ABSTRACT 
 
 
 
 
Title of Disertation / Thesis: AN INVESTIGATION OF EXCLAMATIVES 
IN ENGLISH AND JAPANESE: SYNTAX 
AND SENTENCE PROCESING 
  
 Hajime Ono, Ph.D, 2006 
  
Disertation / Thesis Directed By: Profesor Howard Lasnik, Department of 
Linguistics 
 
 
This disertation is a case study of the syntax of the left periphery, using exclamatives 
in English and Japanese. In the first part, I discuss exclamatives in Japanese in detail 
by focusing on the properties of the exclamative wh-phrases and particles that 
function as licensors for wh-phrases in exclamatives. We argue that licensing 
exclamatives involves at least thre functional heads: Finite, Focus, and Mood. 
Especialy, the necesity of the Mood head diferentiates exclamatives from 
interogatives. On the other hand, we claim that having these thre functional 
projections does not type the clause as exclamative, and show that the presence of a 
wh-phrase of a distinct form is in fact a crucial part of the clause-typing for 
exclamation. This conclusion supports the claim that clause type should not be 
directly encoded into syntax as an independent functional category. The second part 
of disertation deals with English exclamatives. We show that sluicing is available in 
English exclamatives, suggesting that focus is playing a role for the availability of 
  
sluicing, asuming that both interogatives and exclamatives involve focus. Another 
conclusion about English exclamatives is that exclamative wh-clauses are licensed, 
not by selection, but by being c-commanded by a factive operator or a factive 
predicate. This goes against the traditional observation; our conclusion is empiricaly 
justified based on the observation that it is possible to license exclamative wh-clauses 
by a non-local licensor. We argue that this property is similar to what has been 
observed for the aggresively non-D-linked wh-phrases, acounting for the 
distribution and behavior of those non-standard wh-phrases. Finaly, we investigate 
how Japanese exclamatives are procesed by native speakers of Japanese with an on-
line self-paced reading study and two of-line sentence fragment completion studies 
on the procesing of wh-exclamative sentences in Japanese. These studies investigate 
the real-time formation of sentential structures with higher functional categories, and 
show that the parser imediately engages to build syntactic structures with discourse-
oriented higher functional projections before coming across the head, favoring the 
incremental procesing model. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
AN INVESTIGATION OF EXCLAMATIVES IN ENGLISH AND JAPANESE: 
SYNTAX AND SENTENCE PROCESING 
 
 
 
by 
 
 
Hajime Ono 
 
 
 
 
Disertation submited to the Faculty of the Graduate School of the 
University of Maryland, College Park, in partial fulfilment 
of the requirements for the degre of 
Doctor of Philosophy 
2006 
 
 
 
 
 
 
Advisory Commite: 
 
   Profesor Howard Lasnik, Chair 
   Profesor Norbert Hornstein 
   Profesor Juan Uriagereka 
   Asociate Profesor Colin Philips 
   Profesor Robert De Keyser, Dean?s Representative 
 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Hajime Ono 
2006 
 
 
 
 
 
 
 
 
 
 
 
 
 
 i 
 
ACKNOWLEDGEMENTS 
 
There are a lot of people who I am indebted to and to whom I have to expres my 
gratitude, since they have contributed to this disertation substantialy in diferent ways 
al these years, either on the profesional or on the personal front. I would like to 
acknowledge their contributions and to say thanks. 
 
First, I would like to thank my thesis supervisor, Howard Lasnik. He is one of the first 
linguists I noticed that his name is mentioned everywhere in every book when I started 
studying linguistics. At that time, I did not have a slightest idea that eventualy he directs 
my disertation. From each clas I took, I learned very important lesons from Howard. I 
am always impresed by his enthusiasm for teaching. I am grateful to Norbert Hornstein 
for his support as a teacher and as a department chair. I enjoyed his visit to our graduate 
student office, and chating with him was always entertaining. It was also privileged to 
work for his clas as a TA. Juan Uriagereka was in fact a key person for my migration 
from California to Maryland. I was a colloquium coordinator when he visited Irvine, and 
the appointment I took with him left a very strong impresion of the department in 
Maryland. From the Day 1 in Maryland, I start thanking him for providing the 
opportunity to transfer to Maryland. I am delighted to have Colin Philips in my thesis 
commite, since sentence procesing is one of the things I haven?t studies until I got to 
Maryland. This new area of study widened my perspectives, and created many great 
opportunities and experience. Also, I would like to thank other profesors at the 
department, especialy Tonia Bleam; I was a TA for her clas in the semester I defended 
my thesis, and I was supposed to work more, but she was so kind that she tried to reduce 
the amount of my duties as a TA. Without her help, I could not get through the last few 
weks before the defense. 
 
I was extremely a lucky person in Maryland for being surrounded by many great friends. 
Without them, the thre years in aryland would have been very boring. Importantly, I 
learned from them a lot as much as I learned things from clases. I would like to mention 
my thre musketers in Maryland: Masaya Yoshida, Takuya Goro, and Tomohiro Fuji. 
They made my life in aryland very fruitful and memorable. 
 
Of course, I should say that other friends in Maryland have greatly contributed to this 
disertation. I have wonderful memories with each of them, but I am extremely grateful 
to Phil Monahan, Stacey Conroy and Jon Sprouse for helping me to edit this disertation. 
 
Before coming to Maryland, I was a student at University of California, Irvine. I would 
like to show my sincere appreciation to Naoki Fukui and Jim Huang, who created a great 
department in Irvine. I was very sad to witnes that many wonderful people had to leave 
the department. 
 
Another important person I should mention is Grant Goodal, who is my first mentor in 
my life as a graduate student. He supervised my master thesis in University of Texas at El 
Paso. I don?t remember how many times I visited his office to ask a bunch of various 
 ii 
questions. Without his academic and social support, I would not be here. In every major 
event in my graduate student life, I needed his advice. 
 
Also I am grateful for my teachers at International Christian University, especialy 
Motoko Yoshioka and Tomoyuki Yoshida. Not only they taught me a wonderful world of 
linguistics, but they encouraged me when I first mentioned that I wanted to go to a 
graduate school in the US. 
 
In the final stage of writing this disertation in Hiroshima, I owe a lot to people in 
Hiroshima University, especialy Hiromu Sakai, Jun-ichi Tanaka, and Megumi 
Yoshimura, who continuously created a great environment for me. Although it took a 
litle longer to finish my disertation than they might have imagined, they made my life in 
Hiroshima very special. 
 
Finaly, I would like to thank my family members. My parents were amazingly patient; 
they have never complained of me not coming back from the US much earlier (and they 
did not ask me what I have been doing exactly!). I always fel their trust and 
encouragement. 
 
 iv 
CHAPTER 1: INTRODUCTION...................................1 
1. Introduction............................................1 
2. Things not included......................................1 
3. Organization............................................3 
CHAPTER 2: LICENSING JAPANESE XCLAMATIVES.............4 
1. Introduction............................................4 
2. Thre Main Parts.........................................6 
3. Finite: no.............................................10 
4. Focus: noda...........................................13 
4.1 Focus particle......................................13 
4.2 No-da and Adnominal Form...........................17 
4.3 Some other observation with noda......................24 
5. Mood: (da)roo..........................................24 
5.1 Ordering and Morphology.............................25 
5.2 Complementation...................................27 
5.3 Restriction with other Modals..........................29 
5.4 Point of View......................................33 
6. Lower Modal Elements and Negation........................40 
6.1 Evidential: rasi, yooda...............................41 
6.2 Epistemic: bekida, hazuda.............................42 
6.3 Negation..........................................45 
7. Summary.............................................47 
CHAPTER 3: LONG DISTANCE DEPENDENCIES IN JAPANESE 
EXCLAMATIVES..............................................48 
1. Introduction...........................................48 
2. Licensing Exclamatives: Multi-clause........................48 
2.1 Complement Selection...............................48 
2.2 Long Distance Dependency...........................56 
2.3 Island Efects......................................58 
2.3.1 Factive Island.................................58 
2.3.2 Relative Clause Island...........................60 
2.3.3 Wh-Island....................................60 
3. Interaction with Interogatives and Feature Checking............62 
4. Exclamative wh-phrase: nante.............................68 
4.1 Variations of the exclamative wh-phrase..................69 
5. Semantics.............................................71 
5.1 Two degres must be ?related?.........................71 
5.2 Two wh-phrases in one DP............................74 
5.3 The denotation of multiple exclamatives..................76 
6. Summary.............................................78 
CHAPTER 4: EXCLAMATIVES AND SLUICING...................80 
1. Introduction...........................................80 
2. Some Basic Properties of Sluicing..........................81 
2.1 CP status..........................................81 
2.2 P-stranding generalization.............................83 
 v 
2.3 Swiping..........................................85 
2.4 Pseudo-sluicing.....................................87 
3. Island Efects in Exclamatives.............................88 
4. Sluicing..............................................90 
4.1 Factivity..........................................91 
4.2 Material Left Behind.................................92 
4.3 Data.............................................95 
4.3.1 Left Branch Condition...........................95 
4.3.2 Other Islands..................................96 
5. Summary.............................................99 
CHAPTER 5: LICENSING CONDITIONS OF EXCLAMATIVES AND 
INTERVENTION EFECTS....................................100 
1. Introduction: Two puzzles................................100 
1.1 Selection of Exclamatives............................100 
1.2 Licensing Exclamatives at a Distance...................102 
2. den Dikken & Giannakidou (2002).........................103 
2.1 Phrase Structure and Landing Sites of Wh-phrases.........103 
2.2 Wh-the-Hel......................................104 
2.3 Basic Analysis....................................105 
2.4 Wh-the-Hel in Multiple Wh-Constructions...............107 
2.5 Intervention Efects.................................108 
2.5.1 No In-Situ Wh-the Hel Phrase....................108 
2.5.2 Lack of Scope Interaction.......................109 
3. Exclamatives..........................................110 
3.1 Complementary Distribution..........................110 
3.2 Properties of Exclamative Wh-Phrases..................111 
3.3 The Polarity-Based Approach.........................113 
3.4 An Alternative....................................114 
4. Conclusion...........................................114 
CHAPTER 6: PROCESING OF JAPANESE XCLAMATIVES......116 
1. Introduction..........................................116 
2. Syntax of Japanese Exclamatives..........................119 
2.1 Exclamative Wh-Phrase.............................119 
2.2 Exclamative Particle................................121 
3. Experiment 1: Sentence Completion Task....................122 
3.1 Procedures.......................................123 
3.2 Results..........................................124 
3.2.1 Task Acuracy................................124 
3.2.2 Licensing Particles.............................124 
3.2.3 Where the fronted wh-phrases are interpreted........126 
3.3 Discussion.......................................127 
3.3.1 In Situ Wh-phrases............................127 
3.3.2 Fronted h-Phrases...........................128 
4. Experiment 2: Self-Paced Reading Task.....................132 
4.1 Participants.......................................132 
4.2 Materials and Design...............................132 
 vi 
4.3 Procedure........................................134 
4.4 Results..........................................134 
4.4.1 Task Acuracy and Data Analyses.................134 
4.4.2 Reading Time................................134 
4.5 Discussion.......................................135 
5. General Discussion.....................................138 
6. Appendix A: items in the sentence completion task A...........139 
7. Appendix B: items in the self-paced reading task..............141 
CHAPTER 7: CONCLUSION....................................146 
REFERENCES................................................147 
 
 1 
 
CHAPTER 1: INTRODUCTION 
 
1. Introduction 
This disertation should be taken as a case study of the ?Fine Structure of the Left 
Periphery Structure? first extensively discussed by Rizi (1997) and further elaborated by 
Cinque (1999). There is litle debate over the influence of Rizi?s (1997), especialy in 
research concerned with numerous European languages. Yet, it sems fair to say that few 
studies have been done on head-final languages such as Japanese (but se a series of 
studies by Watanabe). One widely shared cross-linguistic observation is that the kind of 
information that is ?visible? in the left periphery in European languages, presumably by 
the phrases that are in a specifier position of a certain functional category, is ?visible? in 
the right periphery as an overt realization of the head of a functional category in 
languages such as Japanese (se Fukui & Sakai 2003, for instance). The current work 
investigates to what extent we can capture the empirical observations in Japanese using 
Rizi?s (1997) framework. I hope that this disertation satisfies those who are interested 
in Rizi?s (1997) framework in general. 
As the central theme connecting al of the chapters in this disertation, we choose 
exclamative sentences. One virtue of the studying exclamative sentences in Japanese is 
that the licensing of exclamatives involves various functional projections in the CP 
domain, the head of which has to be overt. By conducting a detailed investigation of such 
a construction, we can learn various intriguing properties of those functional projections. 
One dificulty we encountered in the course of investigation is that there are few studies 
on exclamatives in the literature, regardles of the language in question. Thus, we are 
compeled to look at the construction in detail, and we have to be careful to establish 
generalizations before ofering an acount. Although we tried as much as we could, it 
sems that much descriptive work is stil necesary. To the extent that we succed in 
describing the empirical data on exclamatives, I hope that this disertation satisfies those 
who are ?exclamative-aficionados?. 
 
 
2. Things not included 
Let me start by listing certain things that are not discussed in this disertation. At one 
time I hoped to describe this disertation as ?a thesis that talks about everything that has 
been said about exclamatives.? But, it was an unrealistic goal. The reason why I do not 
hide the following facts in footnotes is that I hope that a principled explanation wil come 
out in future; they deserve to be the seds for a squib, at least. 
One very interesting topic, that I nonetheles do not have much to say about, is 
the internal structure of exclamative wh-phrases in English. There is some discussion of 
this in Japanese in chapter 2, on the other hand. The word order within the exclamative 
wh-phrases exhibits quite peculiar properties. It looks like the fronted phrase consists of 
two parts as in (1) and (2): a wh-phrase and a ?full? DP. 
 
Fronted wh-phrases 
(1) [What a big car] he bought! 
 2 
(2) [How big a car] he bought! 
I put scare quotes around ?full? because the DPs that can appear inside of an exclamative 
wh-phrase sem to be quite limited. For example, the DP has to be either singular with a 
determiner a, or plural. Note that a definite determiner the, or a quantifier such as some is 
never alowed to appear in the exclamative wh-phrases. 
 
Quite limited DP 
(3) *[What the big car] he bought! 
(4) *[What some big car] he bought! 
One may notice that DPs with such or so obey a similar restriction, which is discussed in 
a great detail in a disertation by Norbert Corver and his subsequent work (Corver 1990, 
1997a, 1997b). The previous discussion on Degre Phrases is quite fascinating, but there 
is nothing substantial that I can add to it (e.g., Bresnan 1973, Jackendoff 1977). A piece 
of work by Hans Bennis (Bennis 1995) and references cited therein might be a good 
place to start a nice cross-linguistic (including non-European languages like Japanese) 
comparison on this isue. Se Kennedy & Merchant 2000 also. 
The second topic I keep silent about is the compositional semantics of 
exclamatives. This is certainly related to the first topic above. I have not sen any 
analysis that works out the semantic contribution of each part of the fronted wh-phrase in 
exclamatives. Where it is necesary, I wil talk a litle bit about the semantics of 
exclamatives (se chapter 2), but otherwise, I folow the analysis found in Zanuttini & 
Portner (2003). As far as I know, there is very litle writen on the semantics of 
exclamatives. The work I am familiar with includes Abels (2004), Oda (2003), Gutierez-
Rexach (1996), and d?Avis (2004), etc., in addition to Zanuttini & Portner (2003).  
Another topic not discussed in this thesis is the restriction on preposition 
stranding in exclamatives. Nelson (1997), citing Emonds (1985) for the original 
observation, has a discussion showing that, compared to wh-interogatives, the fronted 
exclamative wh-phrases do not like to be Ps, as shown below (se also Obenauer 1994). 
Of course, investigating exactly when preposition pied-piping is alowed in wh-
interogatives itself is an interesting topic (se e.g., Hornstein & Weinberg 1981, Baltin 
& Postal 1996, Abels 2003), and so is investigating what acounts for the contrast 
betwen interogatives and exclamatives. Unfortunately, I do not have anything to add. 
The basic paradigm is shown here: 
 
No pied-piping 
(5) a. * To what a crook he lent his house! 
 b.  To which crook did he lend his house? 
(6) a. * With what strange men she danced! 
 b.  ith which men did she dance? 
(7) a. * In what a charming house they live! 
 b.  In which house do they live? 
Another thing I regret I have to leave for the future research is an investigation on 
the cross-linguistic variations of exclamatives, e.g., betwen Japanese and Italian. 
Zanuttini & Portner (2003) extensively used Paduan data, and there are a few studies in 
 3 
German (e.g., Reis 2002, d?Avis 2004), Spanish and Catalan (e.g., Vilalba 2001), and 
French (e.g., Obenauer 1994, Nelson 1997), but the amount of work in the literature is 
too litle to even be compared with that in interogatives. Se Ono (2003, 2004) for an 
atempt to derive a parametric variation in exclamatives betwen English and Japanese. 
 
 
3. Organization 
Let me briefly summarize each chapter. Chapter 2 investigates the precise licensing 
mechanism of Japanese exclamatives. As mentioned above, it wil be shown that thre 
functional projections in the CP domain play crucial roles. Specificaly, we argue that the 
functional projection Mood is an important ingredient. Materials introduced in that 
chapter contribute to understanding the approach proposed by Rizi (1997) and Cinque 
(1999), ilustrating how those ?higher functional projections? are realized in Japanese. 
Such work also contributes to understanding how certain sentence final particles in 
Japanese play a role in syntax. Note that notions such as mood have been discussed 
mainly in the semantic and/or pragmatic literature, but the current work strongly argues 
their syntactic significance (se related discussion in Hara 2006). The main conclusion in 
Chapter 2 is that the nature of the restrictions of mood particles is in fact syntactic. 
Chapter 3 mainly discusses long distance dependencies in Japanese exclamatives. 
We provide a detailed description of where exclamatives are found when embedded. By 
considering previous work based on English (Eliott 1974, Grimshaw 1979), we show 
that the distribution of Japanese exclamatives is diferent from that of English 
exclamatives. Some peculiarities are left for the future research, but we hope that the 
current work contributes to a beter understanding of the nature of the embedded clause in 
Japanese. Another isue discussed in Chapter 3 is the denotation of multiple exclamatives. 
In Chapter 4, we discuss sluicing in English exclamatives. This is a case study to 
examine to what extent there is paralelism betwen interogative and exclamative 
clauses. We offer a detailed description of how sluicing is realized in exclamatives, 
hoping that it helps to reveal some important isues in sluicing. 
Chapter 5 is also about English exclamatives. We point out some similarities 
betwen exclamatives and aggresively-non-D-linked wh-interogatives. Building on the 
recent analysis on the later (den Dikken & Giannakidou 2002), we provide an analysis 
that acounts for the distribution of English exclamatives. 
Chapter 6 is a psycholinguistic study on Japanese exclamatives. With thre 
experiments, we investigate how exclamatives are procesed. We show that procesing of 
exclamatives are somewhat similar to the procesing of interogatives, providing an 
insight into how the human sentence parser builds linguistic representations. 
 
 
 4 
 
CHAPTER 2: LICENSING JAPANESE EXCLAMATIVES 
 
1. Introduction 
The major isue discussed in this section is the required materials in Japanese 
exclamatives.
1
 Smal though they are, it wil be shown that they play a very crucial role in 
licensing Japanese exclamatives. The main conclusion that wil be reached is that the 
exclamative wh-phrase in Japanese is licensed by thre functional heads: no (Finite), da 
(Focus), and roo (Mood). The mood morpheme roo plays an especialy important role, 
for it diferentiates exclamatives from interogatives. In other words, the Finite and Focus 
heads can also be involved in licensing interogatives, but the requirement of the Mood 
head is unique to exclamatives. Also, Finite and Focus particles are optional in 
interogatives in many cases, while they must show up overtly in exclamatives. We argue 
that such a requirement is based on the properties of the wh-phrases involved in 
exclamatives. Difering from other wh-phrases found in interogatives, the exclamative 
wh-phrase nante must be licensed by the Mood head roo, and it cannot be used as an 
interogative wh-phrase. Such deterministic properties of the exclamative wh-phrase 
suggest that the ?locus of exclamation? in exclamative clauses is the presence of the 
exclamative wh-phrase, not the presence of the thre functional heads per se. 
The conclusion atained here contributes to the understanding of the fine details of 
CP structures in Japanese, echoing the view advocated by Rizi (1997) and Cinque 
(1999). By using a syntactic projection of Mood, we argue that notions such as ?speaker?s 
point of view? or ?evidentiality? that have been dealt with primarily in semantics or 
pragmatics have a realization in syntax. Since Japanese has a variety of sentence final 
particles that are thought to be conditioned in discourse context, it sems quite relevant to 
use Japanese in order to investigate how the above-mentioned ?semantic/pragmatic? 
notions interact syntacticaly. It should be emphasized that the syntactic characterization 
of mood has not been discussed much in the minimalist framework (Chomsky 1995, 2000, 
2001). It is hoped that this work can join with others initiating a fruitful investigation in 
much more detail.
2
 
As a specific implementation of the CP structure of Japanese, we argue that the 
CP structure contains at least thre functional projections that are strictly ordered in the 
following way (se Ishihara & Hiraiwa 2002 for discussion of FiniteP and FocusP): 
 
                                                
1
 This chapter includes a lot of work done in collaboration with Tomohiro Fuji. I cannot 
thank him enough for tackling the numerous problems discussed here. 
2
 Se Hara (2006) and Tenny (to appear) for discusion of evidentiality in Japanese, also 
Uchibori (2000) and Watanabe (1996) for subjunctive mood in Japanese. 
 5 
(1) CP structure for exclamatives 
    MoodP  
    u 
    FocusP Mood 
    r roo 
   FiniteP Focus 
   u da 
   IP  Finite 
  6 no 
  ? nante ? 
 
What is significant is that each functional head is realized as an overt particle in Japanese. 
The strict ordering among those particles is straightforwardly acounted for by 
identifying each particle as a head of the functional category, the hierarchical 
configuration of which have been independently justified crosslinguisticaly (se related 
work in Rizi 1997, Cinque 1999, den Dikken 2003, Lipt?k 2001, among others). A 
potential alternative analysis against the use of functional projections would be to say that 
the particle da is just a copula that heads an independent clause (and the particle no is a 
nominalizer). Under such an acount, the structure of an exclamative sentence like the 
following is not mono-clausal, but bi-clausal. 
 
(2) John-wa nante ookina piza-o tabeta-no-da-roo 
 J-TOP NANTE big piza-ACC ate-NML-COP-MOD 
 ?What a big piza John ate!? 
One problem of such an analysis is that it predicts that it would be possible to re-iterate 
those particles. As the following example indicates, it is not possible to repeat those 
particles. 
 
(3) * John-wa nante ookina piza-o tabeta-no-na-no-da-roo 
  J-TOP NANTE big piza-ACC ate-NML-COP-NML-COP-MOD 
On the other hand, an analysis in which each of those particles heads a functional 
projection can provide a straightforward acount, asuming that each of those functional 
projections is alowed to project only once in a clause. Hence, at least at this point, the 
approach that utilizes functional projections sems superior to the one that does not. 
In the next section, we show that each of the functional categories is required in 
exclamatives, which leads us to a conclusion that the exclamative wh-phrase nante has to 
be licensed by those functional projections. Also we conclude that unlike other so-caled 
wh-phrases in Japanese, the wh-phrase nante is not an indeterminate pronoun, but it must 
be used as an exclamative wh-phrase. We further discuss implications for isues of clause 
type in Japanese exclamatives. In section 3, we investigate the properties of the head of 
Finite Phrase no, following Hiraiwa & Ishihara?s work in Japanese. We argue that the 
particle no in exclamatives is the same particle as the particle no in cleft sentences. To the 
extent that the particle no plays a role in marking the presupposition in the cleft sentences, 
it is natural to conclude that the particle no in exclamatives has to do with the factivity 
 6 
observed with exclamatives. In section 4, we look at the focus particles in Japanese. We 
claim that the particle is not a copula, but it is a functional projection that is responsible 
for determining the scope of focused element, developing an idea by Sano (2002). Also, 
we observe that the distribution of the focused particle da is constrained to the extent that 
the adnominal form is restricted. Section 5 is devoted to the discussion of the mood 
morpheme roo. We motivate the claim that it is the head of the Mood Phrase by looking 
at some co-occurrence restriction with other evidential adverbs. Such an observation 
leads us to claim that the mood morpheme roo indicates the speaker?s judgment toward 
the propositional content of the sentence. We therefore connect properties of 
exclamatives with the isues on evidentiality. Section 6 also discusses some other 
evidential morphemes and epistemic morphemes that interact with the mood morpheme 
roo. We claim that the interaction betwen exclamatives and those lexical modal 
elements further supports the analysis that the mood morpheme roo is the head of a 
functional projection, based on the idea by Cinque (1999) that those higher functional 
categories are strictly ordered. 
 
 
2. Thre Main Parts 
This section aims to show that Japanese exclamatives require thre functional 
projections: Finite no, Focus da/desu, and Mood roo. This requirement of exclamatives 
indicates that Japanese exclamatives are minimaly diferent from interogatives, since 
interogatives in general do not require the finite marker and the mood marker in the 
structure. This does not mean that those functional heads cannot appear in interogatives, 
but the interogative wh-phrases can be properly licensed even without those. 
One complication is that those smal markers are sometimes invisible due to some 
phonological / morphological reduction. Finite, Focus and Mood heads can be present in 
wh-interogatives as wel, though they can be morphologicaly reduced in an informal / 
impolite register. Due to this reason, in the crucial cases, we use examples in more formal 
or polite speech registers, where morphological reduction (e.g. case marker drop) is 
generaly restricted (cf. Yoshida & Yoshida 1996). For instance, in a formal register, a 
complex form no-desu or no-de-ar-u (FIN-P-EXIST-PRESENT) is used, instead of no-da. 
The asumption is that when nothing but a complex form (no-desu / no-dearu) appears in 
the structure, it is not a reduced form (since it is in a formal register), and no mood 
morpheme is present in the structure. The contrast in (4) ilustrates that the Mood marker 
is required in exclamatives. Example (4a) is the case where a complex form is used, and 
by asumption there is no Mood marker present in the structure, and the example is not 
aceptable.  
 
Mood morpheme required (yoo is an alomorph of roo) 
(4) a. * John-wa nante kasiko-i -no-desu 
   J-TOP NANTE inteligent-PRES -FIN-FOC.POLITE 
 b.  John-wa nante kasiko-i -no-des -yoo  
   J-TOP NANTE inteligent-PRES -FIN-FOC.POLITE -MOD 
   ?How very inteligent John is!? 
On the other hand, when only no-da appears in the structure in an informal register, there 
is a possibility that there is a Mood morpheme that is invisible due to morphological 
 7 
reduction. If so, the aceptability of (5) makes sense since even though there is no visible 
Mood morpheme present if there is a covert mood morpheme in the structure. 
 
Morphological reduction 
(5) John-wa nante kasiko-i -no-da (-roo) 
 J-TOP NANTE inteligent-PRES -FIN-FOC  -MOD  
 ?How very inteligent John is!? 
We acount for this restriction of obligatory mood marker in exclamatives by arguing 
that an exclamative wh-phrase nante in Japanese requires the Mood head as one of its 
licensers. It suggests that the exclamative wh-phrase bears [mood] feature that must be 
checked by the particular Mood head roo. It is plausible that it also bears a [focus] feature 
that interogative wh-phrases do. 
Now, consider the contrast in (6). An exclamative wh-phrase cannot occur in (6a) 
since Mood roo is not present in the sentence, which is uttered in a register that resists 
morphological reduction, acording to our asumption. The fact that an interogative wh-
phrase can appear in the same environment leads us to conclude that no Mood marker is 
involved in licensing the interogative wh-phrases. 
 
Without the mood marker 
(6) a. * John-wa nante kasiko-i-no-desu(-ka) 
   J-TOP NANTE inteligent-PRES-FIN-FOC.POLITE(-Q) 
   ?How very inteligent John is!? 
 b.  John-wa donokurai kasiko-i-no-desu-ka 
   J-TOP how inteligent-PRES-FIN-FOC.POLITE-Q 
   ?How inteligent is John?? 
The above noted conclusion does not say that the presence of Mood marker roo 
determines a wh-clause as exclamative. Thus, this particular head is not something like a 
head for an ?Exclamative Phrase?. In fact, the Mood head roo can appear in 
interogatives quite naturaly. 
 
Mood ?roo? may occur in interogatives 
(7) a. John-wa dono hito-to kekkonsita -no-da-roo-ka 
  J-TOP which person-with got.maried -FIN-FOC-MOD-Q 
  ?Which person did John mary (, I wonder)?? 
 b. John-wa Mary-to kekkonsita -no-da-roo-ka 
  J-TOP -with got.maried -FIN-FOC-MOD-Q 
  ?Did John mary Mary (, I wonder)?? 
Asuming that the morpheme roo in (7) is the same element we have sen in 
exclamatives, we conclude that the presence of this morpheme does not determine the 
clause type. 
One thing that is unclear how to handle at this point is the presence of the 
Question particle ka. Although it is possible that the particle can appear at the end of 
exclamative clauses, whether or not it is a required element needs further investigation. 
Suppose tentatively that the particle ka is in fact playing a role in order to license 
exclamatives. Unlike the mood morpheme roo, we are led to say that the Question 
 8 
particle can undergo morphological reduction even in a formal speech. Since the 
asumption that morphological reduction is extremely limited in a formal speech is 
important to justify the obligatory presence of the Mood head, it is rather dificult to say 
the Question particle ka is required in exclamatives, unles there is an independent 
motivation to diferentiate the mood head and the question particle with respect to the 
availability of morphological reduction. 
The isue of whether the particle ka is one of the licensors for exclamatives might 
have a link with to what extent the exclamative wh-phrase nante is an ?indeterminate? 
pronoun. It has been noted since Kuroda?s disertation (Kuroda 1965; also se Hagstrom 
1998 and Shimoyama 2001) that wh-phrases in Japanese function as a quantifier, the 
quantificational force of which is determined by the particle atached to it.
3
 For instance, 
the particle ka, for which we have been caling the Question marker, can appear next to 
the wh-phrases or the wh-phrase plus some noun. NPs with the Question marker ka 
function as an existential quantifier. Take dare ?one
person
?, nani ?one
thing
?, and dono ?one-
N? as representative cases of indeterminate pronouns. 
 
Question marker with the wh-phrase 
(8) Dare-ka-ga ki-masi-ta 
 WH-Q-NOM come-POLITE-PAST 
 ?Someone came.? 
(9) John-wa nani-ka-o tabe-masi-ta 
 J-TOP WH-Q-ACC eat-POLITE-PAST 
 ?John ate something.? 
(10) John-wa dono-hon-ka-o yomi-masi-ta 
 J-TOP WH-book-Q-AC read-POLITE-PAST 
 ?John read some book.? 
There is another particle in Japanese that encodes quantificational force. The particle mo, 
when atached to an indeterminate pronoun, creates a universal quantifier / a Negative 
Polarity Item (se some recent discussion in Takahashi 2002, Kishimoto 2001, 
Yatsushiro 1996, and references cited above). 
 
Particle MO with the wh-phrase 
(11) Dare-mo ki-mas-en-desi-ta 
 WH-MO-NOM come-POLITE-NEG-COP-PAST 
 ?No one came.? 
(12) John-wa nani-mo tabe-mas-en-desi-ta 
 J-TOP WH-MO eat-POLITE-NEG-COP-PAST 
 ?John ate nothing.? 
(13) John-wa dono-hon-mo yomi-mas-en-desi-ta 
 J-TOP WH-book-MO read-POLITE-NEG-COP-PAST 
 ?John read no book.? 
                                                
3
 We are also indebted to Hiromu Sakai and Masaya Yoshida for their insightful 
discussion regarding to this point. 
 9 
Finaly, when the Question marker ka appears as a verbal suffix, and when an 
indeterminate pronoun appears in its c-command domain, they function as an 
interogative wh-phrase. 
 
Question marker with the wh-phrase 
(14) Dare-ga ki-masi-ta-ka 
 WH-NOM come-POLITE-PAST-Q 
 ?ho came?? 
(15) John-wa nani-o tabe-masi-ta-ka 
 J-TOP WH-ACC eat-POLITE-PAST-Q 
 ?What did John eat?? 
(16) John-wa dono-hon-o yomi-masi-ta-ka 
 J-TOP WH-book-AC read-POLITE-PAST-Q 
 ?Which book did John read?? 
Now, the wh-phrase nante departs from the paradigm ilustrated above. As we have sen, 
the wh-phrase nante appears in front of an adjective or an adverb, then it is reasonable to 
asume that it is a realization of Degre Phrase (Corver 1990). We can imagine that when 
the particle mo is atached to a DP containing the wh-phrase nante, the DP would mean 
what ?however adjective NP? would mean, functioning as a universal quantifier / 
negative polarity item. It turns out that the example is not gramatical. 
 
(17) * John-wa nante mizikai ronbun-mo yome-nakat-ta 
  J-TOP NANTE short paper-MO read-NEG-PAST 
  ?John did not read a paper, however short.? 
If we want to expres the meaning we intended, some other material (somewhat specific 
to adjectival modification) has to be used. 
 
(18) John-wa donnani mizikai ronbun-mo yome-nakat-ta 
 J-TOP however short paper-MO read-NEG-PAST 
 ?John did not read a paper, however short.? 
Furthermore, the particle ka cannot be used with wh-phrase nante to make an existential 
quantifier. Again, the resulting form is ungramatical. 
 
(19) * John-wa nante mizikai/nagai ronbun-ka-o yomi-masi-ta 
  J-TOP NANTE short/long paper-KA-AC read-POLITE-PAST 
  ?John read a somewhat short/long paper.? 
As we saw with the case of the particle mo, some other material that is explicitly 
existential has to be used. 
 
(20) John-wa ikuraka/aru-tedo mizikai/nagai ronbun-o yomi-masi-ta 
 J-TOP somewhat/some-extent short/long paper-ACC read-POLITE-PAST 
 ?John read a somewhat short/long paper.? 
 10 
Finaly, as we have sen above, the wh-phrase nante cannot be used as an interogative 
wh-phrase. 
 
(21) a. * John-wa nante nagai ronbun-o yomi-masi-ta-ka 
   J-TOP NANTE long paper-ACC read-POLITE-PAST-Q 
   ?How long a paper did John read?? 
 b.  John-wa donokurai nagai ronbun-o yomi-masi-ta-ka 
   J-TOP what.extent long paper-ACC read-POLITE-PAST-Q 
The observations above lead us to conclude that the wh-phrase nante is not indeterminate. 
In fact, the wh-phrase nante is specific to exclamatives. We can speculate that the 
requirement of a specific mood marker roo (presumably encoded as a mood feature) is 
the primary factor for the lack of indeterminate property of the exclamative wh-phrase 
nante. On the other hand, it is reasonable to claim that the focus feature asociated with 
nante does not determine this property, since other indeterminate pronouns sem to have 
the focus feature. A piece of evidence comes from the observation that existential and 
universal quantifiers cannot be a (genuine) topic (atested in many languages; ref. E.Kis 
1995).
4
 
 
(22) *As for someone/everyone, John likes him. 
(23) (*) {Dare-ka / dare-mo}-wa kita 
   WH-KA / WH-MO-TOP came 
  ?As for someone/everyone, he came.? 
The point we are trying to establish is that given the observation of the wh-phrase nante 
being ?determinate?, there is no strong expectation that the particle ka is one of its 
licensors. Of course, it is logicaly possible that the particle ka is playing a role for 
licensing elements other than indeterminate pronoun, but other data suggest that the 
presence of the particle ka is optional at best. 
Below, we provide description of Finite particle no, Focus particle da and a few 
mood markers including roo in some detail. What should be achieved there is to make 
sure that the asumption that those items are in fact part of the functional projections in 
the CP structure. 
 
 
3. Finite: no 
The particle no has a close connection with the focus particle da, which is discussed in 
section 4. When the focus particle appears as a verbal suffix, the particle no has to occur 
as wel. Examples are quite unaceptable without the particle no. 
 
                                                
4
 Japanese examples below are aceptable if the quantifiers are interpreted as a so-caled 
?contrastive? topic, which exhibits diferent properties from a genuine / information topic. 
 11 
?No? required 
(24) a.  John-ga kaeta-no-da 
   J-NOM left-FIN-FOC 
 b. * John-ga kaeta-da 
   J-NOM left-FOC 
   ?JOHN left.? 
We argue that the particle no is the head of the functional projection Finite Phrase, 
following Hiraiwa & Ishihara (2002). They first observe that the presuppositional clause 
in the cleft construction is always marked with the particle no. 
 
Cleft 
(25) [John-ga kata]-no]-wa [hon-o san-satu]-da 
  J-NOM bought-FIN-TOP  book-ACC thre-CLASIFIER-FOC 
 ?It is thre books that John bought.? 
Traditionaly, this morpheme no has been caled a nominalizer, and the idea comes from 
the observation that the sentential subject is also marked with no or koto (Tokieda 1950). 
 
Nominalizers 
(26) [John-ga sono hon-o kata]-no/koto]-wa igai-da 
  J-NOM that book-ACC bought-NOMINALIZER-TOP unexpected-COP(FOC) 
 ?It is unexpected that John bought that book.? 
On the other hand, the particle no in the cleft construction does not alow an alternation 
betwen no and koto. Furthermore, the particle no cannot be replaced with some other 
nominal expresion. This suggests that the particle no in cleft constructions is a highly 
gramaticalized form. 
 
No other options 
(27) *[[John-ga kata]-koto]-wa [hon-o san-satu]-da 
   J-NOM bought-fact-TOC  book-ACC thre-CLASIFIER-FOC 
 ?It is thre books that John bought.? 
(28) *[[John-ga kata]-mono]-wa [hon-o san-satu]-da 
   J-NOM bought-thing-TOP  book-ACC thre-CLASIFIER-FOC 
One of Hiraiwa & Ishihara?s (2002) main claims is that the no-da construction in (29) is 
the underlying structure for the cleft construction. 
 
 12 
(29) Hiraiwa & Ishihara 
    TopP 
    u 
   (Topic)-wa r   
    FocP  Top  
    u   
     r   
    FinP  Foc  da  
    u   
     r   
    TP  Fin   no  
    6    
        
 
Acording to them, the cleft structure is derived by the following two steps of movement. 
First, a focused constituent (XP in the tre below) within a TP moves to [Spec,FocP]. 
Then, the whole FiniteP undergoes remnant movement to [Spec,TopP]. 
 
(30) Hiraiwa & Ishihara 
  TopP      TopP  
  u     qp  
 (Topic)-wa  r   FinP - wa ru 
  FocP Top   ty FocP Top 
  u       
  XP r   TP Fin XP ty  
  FinP Foc  da  5 no t Foc da  
  u        
   r       
  TP Fin   no       
  6       
   t
XP
        
 
 
To the extent that the link betwen the cleft construction and the no-da construction is 
strong, it is plausible that the function of the particle no in the cleft construction is the 
same as the particle no in exclamatives. Specificaly, the particle no marks a part of the 
structure as a ?presuppositional clause? in the cleft sentences. It is not inconceivable to 
make a connection with this and the factivity of exclamatives, which has been widely 
known since Grimshaw (1979). 
 
 
 13 
4. Focus: noda 
The so-caled noda construction, where the particles no-da appears at the end of the 
sentence, has been discussed by various linguists, aiming to figure out the properties of 
this construction from at least thre viewpoints, each of which suggests that the usage of 
no-da takes up diferent functions. We do not mean to suggest that each one of them is 
unrelated or a diferent lexical item, but providing a uniform analysis for al of them is 
not the purpose of this section. It also indicates that the exact characterization sems to be 
extremely dificult. In this section, we concentrate on the properties of no-da that are 
related to focus. In section 4.1, we introduce an analysis by Hiraiwa & Ishihara (2002) in 
which they argue that noda construction is the base for cleft and sluicing constructions in 
Japanese using the Focus projection. We also discuss an analysis of the dake ?only? focus 
construction by Sano (2002), where he hints at some kind of dependency betwen focus 
and the presence of the particles no-da. To the extent that they are succesful, it sems 
plausible to argue that noda constitutes the Focus head in CP structure. In section 4.2, we 
observe that a clause marked with no-da cannot appear with the complementizer koto or 
no. Furthermore, such a clause cannot be used to modify a nominal head. We introduce 
two possible acounts for the restriction on the distribution of a clause with no-da. One 
acount is based on the claim by Murasugi (1990) that the size of relative clauses in 
Japanese is smaler than CP. The other acount is based on the licensing of adnominal 
forms. Following Hiraiwa (2002), this acount claims that the clauses with no-da cannot 
appear as a relative clause since the verbal complex created with no-da cannot license the 
adnominal form. We se that the approach based on the licensing of the adnominal form 
is superior to the one based on the size of a relative clause, but the crucial data is 
provided in section 5. 
 
 
4.1 Focus particle 
One major observation, which we wil follow, is that the construction involves focus. For 
instance, it has been claimed that in example (31), either the whole sentence or a part of it 
is ?emphasized?. When a part of the structure is the locus of focalization, it caries some 
pitch-acent (se Hiraiwa & Ishihara 2002, Sano 2002). Related to that point, for example, 
Hiraiwa & Ishihara (2002) claim that the noda construction (or a noda form) is a basis for 
the cleft and sluicing construction in Japanese, as mentioned in the previous section.
5
 
 
(31) John-ga hon-o Mary-ni ageta-no-da 
 J-NOM book-ACC -DAT gave-FIN-FOC 
 ?It is that John gave a book to Mary.? 
Before we start the discussion of no-da in the perspective of focus, some discussion of a 
copula da in Japanese sems to be in order, since it appears that the copula-like element 
is involved in exclamatives. A slight complication is that the form da is a reduced form of 
                                                
5
 Se Watanabe (2003) for a brief discussion of the focus nature of no-da particles, where 
he suggests that the copula da is the head of the Focus Phrase. Also Schafar (2002) 
suggests a historical link betwen noda constructions and focus construction (Kakari 
Musubi) in Old Japanese. 
 14 
a more complex de-ar-u.
6
 The complex form is the more formal and polite form. As the 
gloss shows, the de-ar-u form consists of a postposition de, a verb stem ar, folowed by a 
tense inflection (present) u.
7
 
 
Japanese copula 
(32) a. John-wa gakuse-da 
  J-TOP student-COP 
  ?John is a student.? 
 b. John-wa gakuse-de-ar-u 
  J-TOP student-P-exist-PRES 
  ?John is a student.? 
Both forms have the past tense form. 
 
Past tense form 
(33) a. John-wa gakuse-dat-ta 
  J-TOP student-COP-PST 
  ?John was a student.? 
 b. John-wa gakuse-de-at-ta 
  J-TOP student-P-exist-PST 
  ?John was a student.? 
Acording to this morphological alternation, the noda construction can be analyzed as the 
reduced form of no-de-ar-u ?nominalizer-postposition-exist-pres?. In general, al the 
instances of a copula da can be replaced with the complex form de-ar-u. We wil come 
back to this alternation again where it becomes relevant. 
 
Basic example: noda  
(34) a. John-ga hon-o Mary-ni ageta 
  J-NOM book-ACC -DAT gave 
  ?John gave a book to Mary.? 
 b. John-ga hon-o ary-ni ageta-noda 
  J-NOM book-ACC M-DAT gave-FOC 
 c. John-ga hon-o ary-ni ageta-no-de-ar-u 
  J-NOM book-ACC M-DAT gave-FOC 
(35) a. John-wa gakuse-da 
  J-TOP student-COP 
  ?John is a student.? 
 b. John-wa gakuse-na-noda  (?na? = pre-nominal form of ?da?) 
  J-TOP student-COP-FOC 
 c. John-wa gakuse-na-no-de-ar-u 
  J-TOP student-COP-FOC 
                                                
6
 Obviously, this is not a new observation. Various Japanese philologists knew this for a 
long time. 
7
 Se some discussion of de-ar-u in Tada (2002). 
 15 
There is one more observation related to the focus property of no-da. 
Investigating one kind of focus marker dake ?only?, Sano (2002) made an interesting 
observation. Acording to him, when the focus marker dake appears in the embedded 
context (he uses noun complementation), it is rather hard for the constituent with dake to 
take wide scope with respect to the matrix predicate. Consider example (36) and two 
scenarios in (37) that are potentialy available. When the example is read in a neutral 
intonation, the example is incompatible with the scenario in (37b). 
 
?Dake? in the embedded clause: hard to get the wide scope 
(36) Titi-wa [Aiko-ga Taroo-dake-ni au] koto-o yurusita 
 father-TOP  A-NOM T-only-DAT met FN-AC alowed 
 ?Aiko?s father alowed that she wil met only Taro.? 
(37) a. ?narow? scenario: Aiko is not wiling to met people in general, except Taro, 
  who she likes a lot. Her father wants her to met other people, but he finaly  
  alowed her to met only Taro. 
 b. ?wide? scenario: Aiko?s father is very strict. He does not want her to met 
  men in general, except Taro, who he realy likes. The only person who her 
  father alowed her to met is Taro. 
What is relevant to our current concern is the folowing. Sano (2002) says that a wide 
scope reading of dake becomes possible in example (36) when it caries a pitch acent, 
and in such a case, it ?invites ? a modal element that suggests the speakers emphatic 
asertion, such as noda.?
8
 In fact, he notes that the narow scope reading disappears in 
(38). 
 
With pitch-accent 
(38) Titi-wa [Aiko-ga Taroo-DAKE-ni au] koto-o yurusita (-noda) 
 father-TOP  A-NOM T-only-DAT met FN-AC alowed  -FIN-FOC 
 ?Aiko?s father alowed that she wil met only Taro.? 
One interpretation of Sano?s observation is that the particle da is an overt 
realization of the functional head Focus, which has a dependency with a phrase marked 
with dake with a pitch acent. Although there has to be some way to explain why the 
presence of no-da is optional, this optionality sems to be restricted. Sano notes that this 
optionality is only available at the root clause, and when example (38) without no-da is 
further embedded by another noun complementation, the wide scope interpretation is not 
available even if dake caries a pitch acent. In (39), a verb yurusita ?alowed? is 
embedded under a noun zizitu ?fact?. More data wil be introduced in later sections 
showing that the particles no-da cannot appear betwen a predicate and a noun in the 
noun complement structure. But, we wil se a trick to place no-da in a noun complement 
structure soon. 
 
                                                
8
 Sano cals noda as a modal element, but he does not define any specific semantic 
contribution for it. 
 16 
Wide scope not available 
(39) [Titi-ga [Aiko-ga kongo Taro-dake-ni au] koto-o yurusita] 
  father-NOM  A-NOM from.now T-only-DAT met FN-AC alowed 
 zizitu-wa nai 
 fact-TOP NEG 
 ?There is not such a thing that her father alowed that Aiko mets only Taro 
 from now on. 
?No-da? in noun complementation 
(40) * [Titi-ga [Aiko-ga kongo Taro-dake-ni au] koto-o 
   father-NOM  A-NOM from.now T-only-DAT met FN-AC 
  yurusita-no-na] zizitu-wa nai 
  alowed-FIN-FOC fact-TOP NEG 
  ?There is not such a thing that her father alowed that Aiko mets only Taro 
  from now on. 
As far as I can tel, the judgment is subtle, but when dake caries a pitch acent, the 
constituent with dake cannot take wide scope with respect to the predicate yurusita 
?alowed.? In fact, I think the example is not good, either. Such an observation suggests 
that when dake caries a pitch acent, it cannot take narow scope, and if an appropriate 
licensing head (Focus, in our terms) is not found, the example is not aceptable. 
Asuming that there is no Focus head in a noun complement structure, a constituent with 
dake cannot be licensed in a position that was available in (38), namely in a clause with 
yurusita. 
Sano notes an ingenious way to make a wide scope reading of dake available. He 
introduces (41) and observes that dake takes wide scope with respect to the predicate 
yurusita ?alowed.? As a mater of fact, it is quite fascinating to encounter nante in this 
example. However, it sems that this nante does not have much to do with the 
exclamative wh-phrase nante. The particular nante in example (41) can be replaced with 
nado-toiu ?like that,? which shows the proposition is a representative case, hence, nante 
here is also a focus marker of sorts. The exact nature of this phrase is unclear, but the 
judgment on the scope relation is correct and interesting. Example (41) is another case 
that demonstrates the same scope interpretation that I came up with. 
 
Wide scope available with a ?focus marker? 
(41) [Titi-ga [Aiko-ga kongo Taro-dake-ni au] koto-o 
  father-NOM  A-NOM from.now T-only-DAT met FN-AC 
 yurusita-nante] zizitu-wa nai 
 alowed-NANTE fact-TOP NEG 
 ?There is not such a thing that her father alowed that Aiko mets only Taro 
 from now on. 
(42) [Titi-ga [Aiko-ga kongo  Taro-dake-ni au] koto-o 
  father-NOM  A-NOM from.now T-only-DAT met FN-AC 
 yurusita-no-da-toiu] zizitu-wa nai 
 alowed-FIN-FOC-C fact-TOP NEG 
As far as those paradigms are concerned, we can conclude that the optionality we found 
at the root clause is not available in the embedded context, at least in noun complements. 
 17 
At the root clause, a covert Focus head, which can be sen as no-da if it is overt, can 
establish a dependency with a constituent containing a pitch acented dake; on the other 
hand, in the noun complement structure, if there is no visible Focus head, then there is no 
Focus head. That explains the lack of the wide scope interpretation of dake in (39). 
There is one more thing that needs to be said about (39). One might wonder why 
the constituent containing the pitch acented dake can not take the widest scope at the 
root clause. Since it is the root clause, the covert Focus head should be available, and 
there should be an interpretation in which the dake phrase takes scope over the negated 
predicate at the root clause. Consider (43), which should have the interpretation we are 
interested in, especialy with an overt no-da. In fact, the example is quite awkward. (44a) 
spels out the intended interpretation, which implies something like (44b). 
 
(43) ? [Titi-ga [Aiko-ga kongo Taro-dake-ni au] koto-o  
   father-NOM  A-NOM from.now T-only-DAT met FN-AC 
  yurusita] zizitu-wa nai-no-da 
  alowed fact-TOP NEG-FIN-OC 
(44) a. It is only Taro such that it is not true that Aiko?s father alowed that she 
  mets him from now on. 
 b. For other men, it is true that Aiko?s father alowed that she mets them 
  from now on. 
It looks like the intended interpretation is pragmaticaly quite strange. There may be a 
way to fix this example somehow (maybe change the noun in the noun complementation 
or change the predicate at the root clause), but it would require more discussion of the 
dake-phrase. I wil leave this as an unsolved puzzle.
9
 
 
 
4.2 No-da and Adnominal Form 
Now, it was hinted at in the previous paragraphs that there are a few pieces of evidence 
that the particles no-da exhibit some properties that are unexpected if they consist of a 
genuine noun and copula. The basic observation is that the genuine copula is compatible 
with complementation with the formal nouns koto or no, but the particles no-da or their 
complex form nodearu cannot. A similar observation is also found with relative clauses: 
the genuine copula can appear imediately before the head noun of the relative clause, 
but the particles no-da or no-dearu cannot. Based on these observations, two possible 
acounts are introduced. As far as the empirical data discussed here, those two acounts 
are both succesful, granting that the underlying asumptions for these acounts. The 
decisive empirical data wil be discussed when we turn to the discussion of the Mood 
head. 
The first observation is the following: the examples in (45) and (46) show that a 
sentence with a copula at the end can be embedded under so-caled formal nouns as long 
as the copula is in a pre-nominal form. 
                                                
9
 Sano (2002) does not discuss the possibility of the dake-phrase taking the widest scope. 
Maybe he considers some syntactic island efects, but the discussion is beyond the scope 
of this section. 
 18 
 
Copula and its pre-nominal form 
(45) Sono kooen-wa kirei-da 
 that park-TOP clean-COP 
 ?That park is clean.? 
(46) a. [sono kooen-ga kirei-na] koto-wa yuume-da 
   that park-NOM clean-COP FN-TOP famous-COP 
 b. [sono kooen-ga kirei-na] no-wa yuume-da 
   that park-NOM clean-COP FN-TOP famous-COP 
  ?The fact that the park is clean is famous.? 
Whether the copula is realized as a complex form or not does not mater.
10
 
 
(47) a. [sono kooen-ga kirei-dearu] koto-wa yuume-da 
   that park-NOM clean-COP.CX FN-TOP famous-COP 
 b. [sono kooen-ga kirei-dearu] no-wa yuume-da 
   that park-NOM clean-COP.CX FN-TOP famous-COP 
  ?The fact that the park is clean is famous.? 
On the other hand, the following examples show that a sentence with the particles no-da 
at the end resists being embedded with a formal noun. 
 
Under ?koto? or ?no? 
(48) a. * [Mary-ga hon-o kata-nona] koto-wa akiraka-da 
    -NOM book-ACC bought-FOC FN-TOP obvious-COP 
 b. * [Mary-ga hon-o kata-nona] no-wa akiraka-da 
    -NOM book-ACC bought-FOC FN-TOP obvious-COP 
   ?The fact that Mary bought the book is obvious.? 
A similar observation is available with the complex form no-dearu. This complex form is 
not distinct betwen the conclusive and pre-nominal forms. 
 
Under ?koto? or ?no? 
(49) John-ga kaeta-no-dearu 
 J-NOM left-FN-COP.CX 
 ?It is that John left.? 
(50) a. * [John-ga kaeta-no-dearu]-koto-wa akiraka-da 
    J-NOM left-FN-COP.CX-FN-TOP obvious-COP 
 b. * [John-ga kaeta-no-dearu]-no-wa akiraka-da 
    J-NOM left-N-COP.CX-FN-TOP obvious-COP 
The examples do not improve when the form appears in a relative clause. Regular 
copulas and their complex form can appear in relative clauses, but the particles no-da and 
their plausible complex form cannot. 
 
                                                
10
 From now on, I am not concerned with the internal structure of the complex form of 
the copula, and it is simply glossed as ?COP.CX?. 
 19 
A genuine copula 
(51) a. kono hana-wa iro-ga  kirei-da 
  this flower-TOP color-NOM beautiful-COP 
  ?As for this flower, the color is beautiful.? 
 b. [iro-ga  kirei-na] hana  
   color-NOM beautiful-COP flower 
  ?a flower the color of which is beautiful? 
(52) a. kono hana-wa iro-ga  kirei-de-ar-u 
  this flower-TOP color-NOM beautiful-COP 
  ?As for this flower, the color is beautiful.? 
 b. [iro-ga  kirei-de-ar-u] hana 
   color-NOM beautiful-COP flower 
  ?a flower the color of which is beautiful? 
Relativization unavailable 
(53) a. John-ga hon-o kata 
  J-NOM book-ACC bought 
  ?John bought a book.? 
 b. [John-ga kata] hon 
   J-NOM bought book 
  ?a book that John bought? 
(54) * [John-ga kata-no-na] hon 
   J-NOM bought-FIN-FOC book 
  ?a book that John bought? 
(55) * [John-ga kata-no-de-ar-u] hon 
   J-NOM bought-FIN-FOC book 
  ?a book that John bought? 
One possible explanation can be given by caling on the structural size of relative 
clauses and noun complementation with formal nouns in Japanese. First let us asume the 
left periphery CP structure by Rizi (1997), abstracting away from Topic Phrase and 
Mood and Question heads for a moment. 
 
(56) CP structure (Rizi 1997) 
  ForceP 
  u 
 Force  FocusP 
   r 
  Focus FiniteP 
    u 
   Finite IP 
     6 
 
Asuming that FocusP is above FiniteP, we can acount for why the particles no-da 
cannot appear in relative clauses or noun complementation if we can independently 
 20 
establish that those clauses or complementation is at most FiniteP. The idea is that the 
particles no-da cannot appear since the clause with no-da is too big to be inside the 
relative clauses or noun complement structures. With respect to the relative clauses in 
Japanese, although the ?size? of relative clauses is stil an isue of debate, Murasugi 
(1991) argues that the size of the Japanese relative clause is IP.
1
 For our current concerns, 
the size of the relative clauses does not have to be IP, but we can provide a 
straightforward acount if relative clauses and noun complementation structure is at most 
FiniteP. 
One might question this acount for the observations noted above, by saying that 
particles no-da or no-dearu cannot appear in those environments due to the fact that they 
lack a proper verb form, namely an adnominal form. Acording to this acount, relative 
clause formation and noun complementation formation require a verbal element to have a 
proper adnominal form, but since no-da and no-dearu are not adnominal forms, they 
cannot appear in those contexts. 
There are dificulties with this counterproposal. First, it is unclear what the proper 
adnominal form has to be. The advocates of the ?size? acount agre that they do not 
have an adnominal form, but that is what they aim to acount for. Acording to the size 
acount, functional heads that are placed higher than FiniteP do not have an adnominal 
form. Then, as far as they can motivate the size of relative clauses and noun complement 
formation independently, their acount holds. Also, if the ?adnominal form? is meant to 
be literaly no-da and no-dearu, this leads to a strange situation. This ?form? acount has 
to say that when the copula da is simply realized as na as an adnominal form, it can 
appear in relative clauses and noun complement formation as shown in (51b), but it 
claims that na appearing after no is not an adnominal form (and the example (54) is not 
aceptable). Also, acording to this acount, dearu (the complex form of the copula) is a 
legitimate adnominal form in (52b), while dearu appearing after no is not a legitimate 
adnominal form. Since the endings of those elements are exactly the same, just saying 
that no-da or no-dearu do not have an adnominal forms does not get us anywhere. 
What we sem to need then is an explicit mechanism of adnominal form licensing. 
Investigating the so-caled nominative / genitive conversion (NGC) in detail, Hiraiwa 
(2000a, 2002) has just that kind of mechanism we need for licensing adnominal forms. 
We introduce some basic facts about NGC, and discuss what we can learn about the 
Focus head from his analysis. 
Descriptively, NGC is a phenomenon in which a genitive case-maker can 
sometimes be used where a nominative case-maker is expected. For instance, a 
nominative marked DP can show up with a genitive case marker inside relative clauses 
and noun complement structures. 
 
                                                
1
 There are some other work that argues for a similar point; se Fukui & Takano 1998, 
Takeda 1999, for instance. Se Kaplan & Whitman (1995), Ishii (1991) for arguments 
that relative clauses in Japanese are in fact CPs. Also I should note that Rizi (1997) 
shows that the size of relative clauses in Italian is in fact ForceP, based on the 
observations that Topics and left dislocated phrases can appear inside the relative clauses. 
 21 
Available in relative clauses and nominal complement 
(se Inoue 1976, Ura 1993, Watanabe 1996, Hiraiwa 2000a, 2002) 
(57) a. [John-{ga/no} yakusita] hon 
   J-NOM/GEN translated book 
  ?a book that John translated? 
 b. John-wa [kinoo Mary-{ga/no} kita]-koto/no-o sir-anakat-ta 
  J-TOP  yesterday -NOM/GEN came-FN-AC know-NEG-PST 
  ?John did not know that Mary came yesterday.? 
 c. [syoorai daizisin-{ga/no} okiru] kanoosei 
   future big.earthquake-NOM/GEN happen possibility 
  ?the possibility that there is a big earthquake? 
On the other hand, such an alternation is not available in some other embedding 
structures. For example, it is not available in complementation that involves to-
subordination. 
 
Unavailable in ?to?-complement and ?toiu?-nominal complementation 
(58) a. John-wa [Mary-{ga/*no} hasita]-to ita 
  J-TOP  -NOM/GEN ran-C  said 
  ?John said that ary ran.? 
 b. [syoorai daizisin-{ga/*no}  okiru]-to iu] zyoohoo 
    future big.earthquake-NOM/GEN happen-C information 
  ?the information that there is a big earthquake? 
 c. [syoorai daizisin-{ga/*no}  okiru]-to no] zyoohoo 
    future big.earthquake-NOM/GEN happen-C information 
  ?the information that there is a big earthquake? 
The availability of NGC in relative clauses and noun complement structures was the 
long-standing generalization in the literature, but by providing the following additional 
data (taken from Hiraiwa 2002), Hiraiwa (2000a, 2002) shows that the true generalization 
is that in (59) (?ADN? = adnominal form). 
 
(59) NGC in Japanese is only licensed by the predicate adnominal form. 
Examples in (60) and (61) show that NGC is available in environments other than relative 
clauses or noun complement structures. The clauses in the examples are something like 
an adverbial clause. When verbs are used, it is hard to se exactly what the verb form is, 
since the adnominal form of those verbs is the same as the conclusive form. 
 
NGC possible 
(60) John-wa [ame-{ga/no} yamu made] office-ni ita 
 J-TOP  rain-NOM/GEN stop.PRS.ADN until office-at be-PST 
 ?John was at his office until the rain stopped.? 
(61) John-wa [Mary-{ga/no} yonda yori] takusan-no 
 J-TOP  -NOM/GEN read.PST.ADN than many-GEN 
 hon-o  yon-da 
 books-ACC read-PST 
 ?John read more books than Mary did.? (Watanabe 1996:396) 
 22 
However, the distinction betwen the adnominal form and the conclusive form is clear 
when the copula is used. Hiraiwa (2000a, 2002) shows that in the same environment, the 
adnominal form of the copula na shows up rather than the conclusive form da. 
 
Adnominal form of copula 
(62) John-wa izyoo-na made-ni sinkeisitu-da 
 J-TOP extraordinary-COP.ADN extent-to nervous-COP.PRS 
 ?John was extraordinarily nervous.?                 cf. (60) 
(63) John-no koto-ga simpai-na yorimo 
 J-GEN thing-NOM worried-COP.ADN than 
 Mary-ga simpai-da 
 -NOM worried-COP.PRS 
 ?I am worried about Mary rather than about John.?      cf. (61) 
Based on the generalization in (59), he proposes the following: 
 
(64) A syntactic C-T-v-V head amalgamate, which is formed via AGRE, corresponds 
to the predicate adnominal form. This head amalgamate has a ?-feature that can 
check genitive Case as wel as nominative Case. 
Acording to his analysis, what is important for creating the head amalgam noted above 
is for the complementizer to be afixal. Being afixal, it requires a verb to raise to (with 
functional projections on its way as wel) to form a morphological unit, which is 
ilustrated in the following diagram. 
 
(65) Amalgamate formation 
    CP 
    ru  
   TP  C [+Af.] 
   ru  
  vP  T[?] 
     
 DP [?] v? 
   ru 
  VP  v  
   
   V 
 
One prediction of his analysis is that when the head amalgama is not available 
NGC is not available. One such case is found when the complementizer is not afixal. We 
have already sen one case where the complementizer is overtly realized as to. The 
examples are repeated here, to ilustrate that although some of those cases are noun 
complement formation, the complementizer is overt. Then, acording to his analysis, the 
head amalgamation is not possible, the adnominal form is not licensed, and NGC is hence 
not available. 
 
 23 
(66) a. John-wa [Mary-{ga/*no} hasita]-to ita 
  J-TOP  -NOM/GEN ran-C  said 
  ?John said that ary ran.? 
 b. [syoorai daizisin-{ga/*no}  okiru]-to iu] zyoohoo 
    future big.earthquake-NOM/GEN happen-C information 
  ?the information that there is a big earthquake? 
 c. [syoorai daizisin-{ga/*no}  okiru]-to no] zyoohoo 
    future big.earthquake-NOM/GEN happen-C information 
  ?the information that there is a big earthquake? 
Now, let us discuss what we can learn about the no-da particle from his analysis 
of the adnominal licensing mechanism. Recal that our basic observation is that the 
adnominal form of the copula na and its complex form dearu can appear in relative 
clauses and noun complement structures, while the aleged adnominal form no-na (from 
no-da) and its complex form no-dearu cannot appear in these environments. We can 
interpret the observation about no-da to mean that the adnominal forms of those particles 
are not licensed. If we adopt Hiraiwa?s analysis of adnominal formation, the facts 
indicate that the head amalgam cannot be formed. Furthermore, one case in which the 
head amalgamation fails is one where the complementizer is overtly realized and is not 
afixal. What this tels us is that the particles no-da / no-dearu are quite diferent from the 
copula, and are in fact a realization of the complementizer. It does not say exactly what 
the functional category is, but that is one conclusion we can draw. 
In this section, we introduced two possible acounts for the observations 
concerning the particles no-da. One acount cals for the specific CP structure proposed 
by Rizi (1997) and argues that the particles no-da cannot occur in relative clauses and 
noun complement structures because the size of those clauses is too smal for no-da, a 
functional projection of Focus head, to appear. To the extent that the asumption that the 
size of the relative clauses and noun complement structure is at most holds, this acount 
can succesfully acount for the facts observed in this section. We have sen that there is 
an alternative acount, which pays special atention to the adnominal form that sems to 
be required in relative clauses and noun complement structures. Initialy, the lack of 
adnominal form itself sems to be what needs to be explained, and given the specific 
licensing mechanism for the adnominal form proposed by Hiraiwa (2000a, 2002), this 
analysis has an interesting implication for the characterization of the particles no-da. 
Given that one important ingredient for the licensing of adnominal forms is for the 
complementizer to be afixal, this approach can acount for the fact that no-da cannot 
occur in relative clauses and noun complement structures by the failure of the head 
amalgamation. This in turn indicates that the particles no-da are not a kind of special 
copula, but an overt realization of a kind of complementizer. It does not determine which 
functional category in CP structure the particles no-da correspond to in a system such as 
Rizi (1997). Certainly it is compatible with the analysis that the particles no-da is a 
Focus head, combined with the discussion from the previous section. As far as the 
empirical data discussed so far is concerned, both analyses sem to be able to provide a 
straightforward answer to the paradigm. However, in section 5, we wil se that 
adnominal licensing acount is superior to the ?size? acount. Before we proced, we wil 
take a smal digresion that introduces some observations concerning no-da. We are not 
trying to provide an acount for these cases now, but certainly it would be great if future 
 24 
research makes available a way to unify al the properties of no-da discussed in this 
chapter. 
 
 
4.3 Some other observation with noda 
Kuno (1973), later followed up by McGloin & Terakura (1978), shows that the particle 
appears in a discourse when the speaker tries to provide a reason for his previous 
utterance. So, in those cases, the utterance can be paraphrased by using -kara-desu ?-
because-copula?. One complication noted by those authors is that the particles no-da also 
shows up at the end of the sentence that the speaker indicates is a consequence of the 
state of afair represented by the previous statement. Obviously, the paraphrase by 
?because? does not work in these cases. 
Second, Kuroda (1973) claims that the particles no-da have something to do with 
the distinction betwen the reportive and non-reportive style. For instance, certain 
adjectives of sensation are sensitive to the presence of the first person subject (se Kuno 
1973 McCawley 1978, 1979 and Tenny to appear as wel). On the other hand, this 
restriction goes away if the sentence is followed by the particles no-da. 
 
Person selection 
(67) a.  Watasi-wa kanasi/sabisi 
   I-TOP sad/lonely 
   ?I am sad/lonely.? 
 b. * Tanaka-san-wa kanasi/sabisi 
   T-Mr.-TOP sad/lonely 
   ?r. Tanaka is sad/lonely.? 
(68) a.  Watasi-wa kanasi-noda 
   I-top sad/lonely-NODA 
   ?I am sad/lonely.? 
 b.  Tanaka-san-wa kanasi-noda 
   T-Mr.-top sad/lonely-NODA 
   ?r. Tanaka is sad/lonely.? 
 
 
5. Mood: (da)roo 
This section reviews properties of a morpheme (da)-roo that is caled a presumptive 
mood marker (Johnson 2003). Other authors use a slightly diferent term, such as an 
evidential marker (Hara 2006), but that does not concern us here. The morpheme 
indicates the judgment of the speaker toward the proposition to which the morpheme 
ataches. Various observations are introduced, and we argue that those are acounted for 
by taking the mood morpheme as a head of the functional projection Mood Phrase (or 
Evidential Phrase) in the CP structure. Following the conclusion that Focus Phrase is 
above Finite Phrase (Rizi 1997), we show that Mood Phrase is above Focus Phrase. 
In section 5.1, we investigate morphological properties of the mood particle roo. 
It wil be shown that there is a specific ordering restriction with respect to the particle, 
and when it is used in exclamatives, it has some efect on the distribution of the 
 25 
preceding morpheme no. In section 5.2, we observe that the mood morpheme roo cannot 
appear in relative clauses. Looking at the two alternative acounts introduced in section 4, 
we conclude that the analysis that is based on the licensing mechanism of the adnominal 
form of the verb is superior to the analysis that relies on the size of the relative clause. 
Section 5.3 introduces an observation that the mood morpheme cannot co-occur with 
another verbal suffix rasi ?sem?. We argue that the co-occurrence restriction is 
acounted for by asuming that the mood morpheme roo contains a pronominal element 
that refers to the speaker of the utterance. Finaly in section 5.4, we discuss the notion 
?point of view?. Building on the previous work such as Tenny (to appear), we propose a 
mechanism that captures the observation with respect to the mood morpheme roo. 
 
 
5.1 Ordering and Morphology 
There is an ordering restriction betwen the mood morpheme and the focus particle. The 
mood morpheme has to occur after the focus particle noda when they occur in the same 
clause. 
 
Ordering restriction 
(69) a.  John-ga Mary-o tataita no-da roo 
   J-NOM -ACC hit FIN-FOC MOD 
 b. * John-wa Mary-o tataita roo no-da 
   J-NOM -ACC hit MOD FIN-FOC 
   ?It sems that it is John who hit ary.? 
The existence of this morpheme roo is most clearly sen in a sentence ending with 
a copula as shown below. The morpheme roo appears imediately after the copula. The 
sentence with roo indicates that the speaker has some judgment about the propositional 
content. For instance, an adverb such as osoraku ?probably? can also appear optionaly. 
 
Mood ?roo?: presumptive mood 
(70) a. John-wa gakuse-da 
  J-TOP student-COP 
  ?Probably, John is a student.? 
 b. John-wa (osoraku) gakuse-da-roo 
  J-TOP  probably student-COP-MOD 
  ?Probably, John might be a student.? 
The judgment by the speaker of the utterance with the morpheme roo does not have to be 
limited to something like ?uncertainty?. The morpheme can co-occur with other adverbs 
that indicate the speaker?s confidence. 
 
(71) John-wa kito kuru-da-roo 
 J-TOP certainly come-COP-MOD 
 ?(I am certain) John wil come.? 
In terms of morphology, the distribution becomes a litle complicated when the 
morpheme appears imediately after a verb or an adjective. In the cases shown below, 
the morpheme has to be daroo. 
 26 
 
After a verb or an adjective 
(72) John-wa osoraku hon-o kau-*(da)roo 
 J-TOP probably book-ACC buy-MOD 
 ?Probably, John would buy a book.? 
(73) John-wa osoraku kasiko-i-*(da)roo 
 J-TOP probably smart-PRS-MOD 
 ?Probably, John would be smart.? 
If da is a realization of the copula in (72) and (73), this is a litle mysterious, since in 
general a copula shows up after a nominal expresion, and obviously verbs and adjectives 
are not nominal expresions. To reconcile this situation, we suggest that the form daroo is 
in fact a reduced form of no-da-roo. Asuming that no is some kind of nominal 
expresion (presumably a Finite head), the obligatory presence of da in daroo can be 
explained. Also, when no is added in the above examples, we do not detect any semantic 
diference, and the vowel o may be dropped as wel. This indicates that it is easy for the 
sequence of particles to undergo some morphological / phonological reduction. 
 
After a verb or an adjective 
(74) John-wa osoraku hon-o kau-n(o)da-roo 
 J-TOP probably book-ACC buy-NODA-MOD 
 ?Probably, John would buy a book.? 
(75) John-wa osoraku kasiko-i-n(o)da-roo 
 J-TOP probably smart-PRS-NODA-MOD 
 ?Probably, John would be smart.? 
Regarding the link betwen this reduction phenomena and exclamatives, the absence of 
no from no-da-roo in exclamatives is never alowed, as shown in (76). We took this 
observation to mean that the particles no-da participate in a checking relation with the 
exclamative wh-phrase. Therefore, morphological reduction is not available.
12
 
 
Required ?no? in exclamatives 
(76) John-wa nante atui hon-o yomu-*(no)-da-roo 
 J-TOP NANTE thick book-ACC read-FIN-FOC-MOD 
 ?What a thick book John reads!? 
 
 
                                                
12
 It might be too simplistic to say that the morpheme does not participate in ?any? 
checking relation when morphological reduction is possible. The exact condition when 
morphological reduction is possible needs further investigation. This raises a question, of 
course, what is the diference betwen the reduction of noda and that of roo. At the 
matrix clause, even in exclamatives, the mood marker roo can be invisible, yet is able to 
participate in licensing exclamatives. 
 27 
5.2 Complementation 
One observation that is common with the finite and focus particle no-da is that sentences 
with roo cannot be embedded under koto or no. 
 
Unable to be embedded under ?koto? or ?no? 
(77) a. * [John-ga gakuse-da-roo]-koto-wa yuume-da 
    J-NOM student-COP-MOD-NML-TOP famous-COP 
 b. * [John-ga gakuse-da-roo]-no-wa yuume-da 
   J-NOM student-COP-MOD-NML-TOP famous-COP 
  ?It is famous that John would be a student.? 
(78) a.  [John-ga gakuse-na]-koto-wa yuume-da 
    J-NOM student-COP-NML-TOP famous-COP 
 b.  [John-ga gakuse-na]-no-wa yuume-da 
    J-NOM student-COP-NML-TOP famous-COP 
   ?It is famous that John is a student.? 
The paralelism goes further. Sentences with roo cannot be embedded within a relative 
clause. 
 
Unable to appear in relative clauses 
(79) a. * [John-ga osoraku kata-daroo] hon 
    J-NOM probably bought-MOD book 
   ?the book that John probably bought? 
 b.  [John-ga osoraku kata] hon 
    J-NOM probably bought book 
   ?the book that John probably would have bought? 
(80) a. * [osoraku sono hon-o kata-daroo] hito 
    probably that book-ACC bought-MOD man 
   ?the man who probably bought that book? 
 b.  [osoraku sono hon-o kata hito 
    probably that book-ACC bought man 
   ?the man who probably would have bought that book? 
(81) a. * John-wa [Mary-ga ringo-o kita-daroo]-no-o tabeta 
   J-TOP  -NOM apple-ACC cut-MOD-NML-ACC ate 
   ?John ate the apple which Mary would have cut.? 
 b.  John-wa [Mary-ga ringo-o kita]-no-o tabeta 
   J-TOP  -NOM apple-ACC cut-NML-AC ate 
   ?John ate the apple which Mary cut.? 
Recal the discussion in 4.2, where we introduced two possible acounts for the 
observation that the particles no-da cannot occur in noun complement structures and 
relative clauses. Under one acount that we caled the ?size acount?, it is asumed that 
noun complement structures and relative clauses are at most Finite Phrases in the sense of 
Rizi (1997). The hypothesis makes a prediction that any functional projections (realized 
as a morpheme) that are located higher than FiniteP cannot occur in noun complement 
structures and relative clauses. Since there is independent evidence that the particles no-
 28 
da are the head of FiniteP and FocusP, the size acount can succesfully explain the 
unavailability of the particles no-da in noun complement structures and relative clauses. 
On the other hand, we introduced an alternative acount that we cal the ?adnominal 
form? acount. This acount claims that the particles no-da cannot appear in those 
structures since the adnominal form cannot be licensed in the particles no-da. One 
requirement for licensing the adnominal form is that the complementizer has to be null 
and afixal. Since the particles no-da contain the overt complementizer no, the adnominal 
form cannot be licensed with the particles no-da. Crucialy, under this acount, the size of 
the clause is irelevant for determining whether an element can appear inside the relative 
clauses or noun complement structure. As far as the adnominal form is licensed, the 
particular item can appear in those environments. 
There is one piece of data about the distribution of da-roo that leads us to acept 
the adnominal form acount, rather than the size acount. Recal in the previous section 
that we showed that the copula da has a complex form de-ar-u. The same alternation is in 
fact available for no-da, and the complex form would be no-de-ar-u. The only diference 
betwen the forms of the alternation is purely stylistic, as far as we can tel. In other 
words, the complex form is a formal form. Now, the natural extension of the paradigm 
sems to be that there is a complex form of da-roo (or in fact no-da-roo under our 
morphological reduction analysis), which is no-de-ar-oo. As we might have guesed, the 
complex form exhibits (almost) al the property we found with no-da-roo. For instance, 
the complex form no-de-ar-oo cannot be embedded under the formal nouns or relative 
clauses. This is what we expect to observe. 
 
Complex form 
(82) * [John-ga kaeta-no-de-ar-oo]-koto-wa yooini soozoo-dekiru 
   J-NOM left-N-P-EXIST-MOD-FN-TOP easily imagine-can 
  ?It is very easy to imagine that John left.? 
(83) * [John-ga kata-no-de-ar-oo] hon 
   J-NOM bought-N-P-EXIST-MOD book 
  ?the book that John bought? 
However, when no is dropped, examples are aceptable. 
 
Complex form without ?no? 
(84)  [John-ga kaeta-de-ar-oo]-koto-wa yooini soozoo-dekiru 
   J-NOM left-P-EXIST-MOD-FN-TOP easily imagine-can 
  ?It is very easy to imagine that John left.? 
(85)  [John-ga kata-de-ar-oo] hon 
   J-NOM bought-P-EXIST-MOD book 
  ?the book that John probably bought? 
There is one more important point in this paradigm. When no-de-ar-oo is atached to a 
sentence ending with the copula (unlike the examples above where verbs are used), we 
got the expected result. The structure that involves the sequence of those particles is not 
alowed to appear in noun complementation with a formal noun. Now, we can make the 
example beter, by deleting no, but at the same time, the copula na (a pronominal form of 
the copula) must disappear as wel, in order for an example to be aceptable. 
 29 
 
Copula plus ?no? 
(86) * [John-ga gakuse-na-no-de-ar-oo]-koto-wa yooini soozoo-dekiru 
   J-NOM student-COP-N-P-EXIST-MOD easily imagine-can 
  ?It is very easy to imagine that John would be a student.? 
(87)  [John-ga gakuse-de-ar-oo]-koto-wa yooini soozoo-dekiru 
   J-NOM student-COP-N-P-EXIST-MOD easily imagine-can 
  ?It is very easy to imagine that John would be a student.? 
This is a quite unexpected contrast acording to the size acount. Based on the ordering 
restriction betwen the finite and focus particles no-da, and mood particle roo, we 
reached an independent conclusion that the Mood head is placed higher than the Focus 
head. The size acount argues that the reason why the finite / focus particles no-da cannot 
appear in relative clauses and noun complement structures is that those structures cannot 
acommodate a structure larger than FiniteP. Asuming Rizi?s (1997) hierarchy of CP 
structure where FocusP is higher than FiniteP, and employing the conclusion that the 
Mood head is higher than FocusP, the mood particle roo should never be able to appear 
in relative clauses or noun complement structures, asuming that the complex form de-ar 
and the focus particle no-da are the same in relevant respects. 
 
(88) CP structure 
    MoodP 
    ru 
   FocusP Mood 
   ru roo 
  FinP Focus 
 6 de-ar 
  ? 
 
On the other hand, the adnominal acount has a straightforward explanation for 
why the presence or the absence of the particle no maters, and why examples are 
aceptable when there is no particle no. Recal that acording to the adnominal acount, 
the finite / focus particles no-da cannot appear in relative clauses or noun complement 
structures since there is an overt complementizer no that intervenes in the verbal 
amalgamation that licenses the adnominal form. Acording to this acount, the form de-
ar-roo is an adnominal form, and the form is licensed since there is no overt no, which 
blocks the amalgamation. Since the adnominal form is licensed, the form can appear in 
relative clauses and noun complement structures. 
 
 
5.3 Restriction with other Modals 
The next observation with the morpheme roo is a co-occurrence restriction with another 
evidential modal-like element rasi or yoo ?sem, likely.? Examples are unaceptable 
when they occur in the same clause. 
 
 30 
Co-occurrence restriction 
(89) a.  John-wa sono hon-o kata rasi no-desu 
   J-TOP that book-ACC bought sem FIN-FOC 
   ?John semed to have bought that book.? 
 b. * John-wa sono hon-o kata rasi no-des-yoo 
   J-TOP that book-ACC bought sem FIN-FOC-MOD 
   ?John semed to have bought that book.? 
On the other hand, if rasi ?sem? is replaced with an epistemic modal element, examples 
are aceptable. 
 
Epistemic modals 
(90) a. John-wa sono hon-o kau beki-na no-desu 
  J-TOP that book-ACC buy should-COP FIN-FOC 
  ?John should buy that book.? 
 b. John-wa sono hon-o kau beki-na no-des-yoo 
  J-TOP that book-ACC buy should-COP FIN-FOC-MOD 
  ?John should buy that book.? 
First, the fact that there is a contrast in (89), but not in (90) suggests that the distribution 
of roo has something to do with evidential modals; this supports our claim that roo is a 
presumptive mood marker, which is an indicator of speaker?s judgment. Discussing some 
similar materials, Tenny (to appear) offers an insight that evidential materials are 
?anchored? to notions such as the speaker or the hearer. Without going into her exact 
analysis, we can make use of her insight and say that evidential materials contain a 
pronoun-like element. Since the evidential morphemes in the above examples are at the 
matrix clause, and they represent the speaker?s judgment, let us asume that both rasi 
and roo cary a pronoun ?I-speaker?. Under the asumption that those pronouns obey 
Principle B, we can explain why rasi and roo cannot occur in the same clause, that is, 
rasi is localy c-commanded by roo, both of which cary the same index.
13
  
One piece of support comes from the observation that rasi and roo can co-occur 
in interogatives as shown below. 
 
Interogatives 
(91) a. John-wa itu sono hon-o kata rasi no-desu-ka 
  J-TOP when that book-ACC bought sem FIN-FOC-Q 
  ?When does he sem to have bought that book?? 
 b. John-wa itu sono hon-o kata rasi no-des-yoo-ka 
  J-TOP when that book-ACC bought sem FIN-FOC-MOD-Q 
  ?When does he sem to have bought that book?? 
Our analysis can acommodate this fact since it has been known that the ?anchor? of 
evidential element is shifted in interogatives. 
 
                                                
13
 Tenny?s (to appear) implementation is slightly diferent, but she atempts to explain the 
same observation. 
 31 
The Interogative Flip 
(92) a. The bicyclist evidently has escaped injury in the crash. 
 b. Has the bicyclist evidently escaped injury in the crash? (Tenny to appear (43)) 
The adverb ?evidently? represents a judgment of the speaker in the declaratives, while the 
one in interogatives represents a judgment of the hearer. We can interpret this contrast to 
suggest that the mood morpheme roo in (91b) caries a pronoun ?you-hearer? in 
interogatives. Since this eliminates the Principle B violation, the example is 
gramatical.
14
 There is one more piece of evidence for the binding-theoretic approach. 
The morphemes rasi and roo can cary the same pronoun, as far as they are not in the 
same clause. In the following example, the clause that contains rasi is embedded under 
an atitude predicate omou ?think.? Also, the morpheme roo appears in the matrix clause. 
Acording to our analysis, the matrix roo caries a pronoun ?you-hearer? since it is an 
interogative clause, and the embedded rasi also caries the same pronoun since the 
matrix subject is ?you-hearer?. 
 
Diferent Clause 
(93) Anata-wa [John-ga sono hon-o kata rasi-to] omotteiru 
 you-TOP  J-NOM that book-ACC bought sem-C think 
 nodes-yoo-ka 
 FOC-MOD-Q 
 ?Do you think that John sems to have bought that book?? 
Although the matrix evidential morpheme yoo (a phonological variant of roo) c-
commands the embedded evidential morpheme rasi, the example is gramatical, since 
they are not in the same clause, avoiding the Principle B violation.
15
 
One consequence of this analysis is that there are in fact two Mood heads that 
interact each other. Acording to Cinque-style cartography of CP structure, it suggests 
that there is one Mood head that can occur lower than Focus head, and the other one that 
can occur higher than Focus head. We say that they interact, since Principle B of the 
Binding Theory regulates a certain combination of them, due to the pronouns they 
contain. Nonetheles, they function diferently at the same time. Acording to our claim, 
the Mood roo is playing an important role in licensing the exclamative wh-phrase. On the 
other hand, the lower Mood head does not participate in this licensing relation. In other 
words, exclamatives are not compatible with evidential modals other than ro. Examples 
(94) and (95) show that evidential modals rasi or yooda cannot license the exclamative 
wh-phrases. 
 
                                                
14
 Whether the other modal element rasi ever alows this kind of alternation betwen the 
speaker and the hearer in interogatives should be worked out more in detail. It is quite 
relevant to ask why rasi does not alter its pronoun in the way roo does, if in fact it does 
not. 
15
 It is not clear if the proposed analysis can say something about the interaction betwen 
the ?pronoun? in the mood morpheme and the ordinary pronouns. This should be 
investigated in the future research. 
 32 
Evidential modals: ?rasi? and ?yooda? 
(94) * John-wa nante ookina kuruma-o kata-rasi-no-desu 
  J-TOP NANTE big car-ACC bought-sem-FIN-FOC 
  ?What a big car John sems to have bought!? 
(95) * Mary-wa nante hayaku hasita-yoona-no-desu 
  -TOP NANTE fast ran-likely-FIN-FOC 
  ?How fast Mary is likely to have run!? 
In fact, given that the higher Mood head is required in exclamatives, the lower one cannot 
even show up in exclamatives.
16
 Although the exact semantic contribution of those two 
separate Mood heads must be investigated further, we have to leave that for the future 
research. 
One related observation with respect to the incompatibility of rasi and roo is that 
epistemic modals such as beki ?should? and hazu ?must? do not interact with roo, at least 
in the sense noted above. These two morphemes can co-occur in the same clause, and in 
fact these epistemic modals are aceptable in exclamatives. 
 
Epistemic modals: ?beki? and ?hazu? 
(96) John-wa takusan-no hon-o yomu-beki-data-no-da-roo 
 J-TOP many-GEN book-ACC read-must-COP-FIN-FOC-MOD 
 ?John must have read many books.? 
(97) Mary-wa takusan-no sinnyuusyain-o yatou-hazu-data-no-da-roo 
 -TOP many-GEN employee-ACC hire-should-COP-FIN-FOC-MOD 
 ?Mary should have hired many new employees.? 
Epistemic modals: ?beki? and ?hazu? 
(98) John-wa nante takusan-no hon-o yomu-beki-data-no-da-roo 
 J-TOP NANTE many-GEN book-ACC read-must-COP-FIN-FOC-MOD 
 ?How very many books John must have read!? 
(99) Mary-wa nante takusan-no sinnyuusyain-o yatou-hazu-data-no-da-roo 
 -TOP NANTE many-GEN employee-ACC hire-should-COP-FIN-FOC-MOD 
 ?How very many new employees Mary should have hired!? 
A possible acount for this is to claim that epistemic modals do not have a pronoun that is 
present in evidential morphemes. In that way, Principle B efects are not applicable in the 
structure, and the examples are aceptable. Another way to explain the contrast betwen 
the evidential and epistemic morphemes is to say that those two have the same kind of 
pronoun, but to stipulate that the pronoun asociated with evidential morphemes and the 
one with epistemic morphemes do not interact. It could be that the interaction of the 
                                                
16
 Alternatively, the incompatibility betwen exclamatives and evidential morphemes 
rasi or yooda might be acounted for by asuming that there is one slot in Mood Phrase 
(or Evidential Phrase), which needs to be occupied by roo in exclamatives. Then, in 
exclamatives, it is impossible for other evidential modal to occur in the same CP structure. 
Unfortunately, this acount has to stipulate that something special is going on in 
interogatives. Recal that in interogatives, both rasi and roo can co-occur, and it is 
unlikely that they are represented by the same Mood head. 
 33 
pronouns is observed only when the pronouns discussed here are in similar functional 
heads, such as the higher Mood and the lower Mood. 
 
 
5.4 Point of View 
This section investigates the distribution and properties of no-da-roo. The general 
conclusion is that the distribution of no-da-roo is regulated by a rule that employs notions 
such as ?Point of View? or ?perspective? (Sels 1987, Nishigauchi 1999, Kuno 1972, 
Uriagereka 1995, Speas 1999, among others). Building on observations in the literature 
and other new observations, we propose that the clause with no-da-roo has to be acesed 
by the speaker of the whole utterance through some other person. Specificaly, we argue 
that there is an operator in the clause with no-da-roo that has to be licensed at the matrix 
clause (se also Hara 2006 for relevant discussion). It is argued that such movement is 
responsible for linking the speaker and the clause with no-da-roo. 
Let us introduce some basic observations. There sem to be at least two types of 
interogative predicates in Japanese. When those predicates take an embedded clause, one 
type can optionaly put an additional complementizer to, while the other type cannot. For 
example, kiku ?ask? and tazuneru ?ask? alow the particle to to appear, but siritagaru 
?wonder? and siraberu ?investigate? do not. 
 
(100) a. John-wa [Mary-ga dare-ni ata]-ka(-to) kiita/tazuneta 
  J-TOP  -NOM who-DAT met-Q-C asked/inquired 
  ?John asked who ary met.? 
 b. John-wa [Mary-ga dare-ni ata]-ka(-*to) siritagata/sirabeta 
  J-TOP  -NOM who-DAT met-Q-C wondered/investigated 
  ?John wondered/investigated who Mary met.? 
The division betwen those two types of predicates correlates with whether or not no-da-
roo can appear in the embedded interogative clause. 
 
(101) a.  John-wa [Mary-ga dare-ni ata-no-da-roo]-ka-to kiita/tazuneta 
   J-TOP  -NOM who-DAT met-FIN-FOC-MOD-Q-C asked/inquired 
   ?John asked who ary met.? 
 b. * John-wa [Mary-ga dare-ni ata-no-da-roo]-ka  
   J-TOP  -NOM who-DAT met-FIN-FOC-MOD-Q-C 
   siritagata/sirabeta 
   wondered/investigated 
   ?John wondered/investigated who Mary met.? 
For those that alow no-da-roo in the embedded interogative clause, when those particles 
appear, the complementizer to is now obligatory. 
 
(102) * John-wa [Mary-ga dare-ni ata-no-da-roo]-ka kiita/tazuneta 
  J-TOP  -NOM who-DAT met-FIN-FOC-MOD-Q asked/inquired 
  ?John asked who ary met.? 
 34 
The generalization is that the presence of the particles no-da-roo requires the presence of 
the complementizer to if the complementizer to is optionaly available for a predicate.
17
 
Another peculiar paradigm involving the presence of no-da-roo is the following: 
usualy, verbs such as omou ?think? cannot embed an interogative clause. Omou ?think? 
selects the complementizer to, and the example below is unaceptable.
18
 With this verb, 
the complentizer to is not optional. 
 
(103) a.  John-wa [Mary-ga sense-ni ata]-to omotteiru 
   J-TOP  -NOM teacher-DAT met-C think 
   ?John thinks that ary met the teacher.? 
 b. * John-wa [Mary-ga dare-ni ata]-ka-to omotteiru 
   J-TOP  -NOM who-DAT met-Q-C think 
   ?John thinks who ary met.? 
However, an interogative clause can appear when no-da-roo is in the interogative 
clause. 
 
(104) John-wa [Mary-ga dare-ni ata-no-da-roo]-ka-to omotteiru 
 J-TOP  -NOM who-DAT met-FIN-FOC-MOD-Q-C think 
 ?John think, who ary met.? 
There is one more observation, which is noted in the literature (Kuroda 1973, 
Kuno 1973, McCawley 1978, Tenny to appear). When a certain clas of adjectives that 
represent ?feling? appear in the matrix clause, the particles no-da(-roo) are required if 
the subject is not the first person. 
 
(105) a.  Watasi-wa kanasi 
   I-TOP  sad 
   ?I am sad.? 
 b. * Tanaka-san-wa kanasi 
   T-Mr.-TOP sad 
   ?r. Tanaka is sad.? 
(106) a. * Watasi-wa kanasi-no-da(-roo) 
   I-TOP  sad-FIN-FOC-MOD 
   ?I am sad.? 
 b.  Tanaka-san-wa kanasi-no-da(-roo) 
   T-Mr.-top sad-FIN-FOC-MOD 
   ?r. Tanaka is sad.? 
The traditional idea, as in Kuroda (1973), is that the examples in (105) are in the 
reportive style, where the speaker of the utterance is playing an objective role. He is 
reporting the state of afairs in an objective viewpoint, and he is not alowed to talk about 
the ?feling? of some other person. By using the particles no-da-roo, he now has the 
                                                
17
 The optionality of the complementizer to only arises with the ?interogative? predicates 
noted above. 
18
 As soon as the nominalizer (or in our analysis, the Finite head) no is added to the 
clause, the example is aceptable. This suggests that it is a reduced form of no-da-roo. 
 35 
viewpoint of somebody else (i.e., Tanaka-san). Now, he can talk about the ?felings? of 
that person. This intuitive idea is supported by the following contrast. Above, we have 
sen that the verb omou ?think? can embed an interogative clause if the interogative 
clause contains no-da-roo. It is not possible to make a matrix yes-no question. 
 
(107) a.  John-wa [dare-ga kita-no-da-roo]-ka-to  omotteiru 
   J-TOP  who-NOM came-FIN-FOC-MOD-Q-C think 
   ?John thinks who came.? 
 b. * John-wa [dare-ga kita-no-da-roo]-ka-to  omotteiru-no-desu-ka 
   J-TOP  who-NOM came-FIN-FOC-MOD-Q-C think-FIN-COP-Q 
   ?Does John think who came?? 
The unaceptable example (107b) can be explained in the following way. Given the 
presence of no-da-roo, the speaker of the whole utterance has the perspective of the 
matrix subject John. This means that the speaker has full control of John?s mind, and it is 
quite strange to inquire about the content of John?s thought, since the speaker should 
know everything about what John thinks. 
The notion of perspective has been used to explain the diference betwen the use 
of pronoun and reflexive in the embedded clause under the atitude predicate such as 
omou ?think? (se e.g., Sels 1987). 
 
(108) a. John
2
-wa [kare
2
-ga itiban-da]-to omotteiru 
  J-TOP  he-NOM best-COP-C think 
  ?John thinks he is the best.? 
 b. John
2
-wa [zibun
2
-ga itiban-da]-to omotteiru 
  J-TOP  he-NOM best-COP-C think 
  ?John thinks ?self? is the best.? 
The diference betwen those examples has been argued to be who has the perspective in 
the embedded clause. In the example with the pronoun kare, it is the speaker of the whole 
utterance who has the perspective, and he reports the state of afairs in an objective mode. 
On the other hand, in the example with the reflexive zibun, the content of the embedded 
clause is described from the perspective of the matrix subject John. Of course, even in 
such a case, it is the speaker who utters the whole sentence, and it means that the speaker 
has aces to the embedded clause through the perspective of the matrix subject, virtualy 
as if the speaker of the whole utterance is the matrix subject. Interestingly, when no-da-
roo is added to the embedded clause, the example with a pronoun is unaceptable. 
 
(109) a. * John
2
-wa [kare
2
-ga itiban-na-no-da-roo]-to omotteiru 
   J-TOP  he-NOM best-COP-FIN-FOC-MOD-C think 
   ?John thinks he is the best.? 
 b.  John
2
-wa [zibun
2
-ga itiban-na-no-da-roo]-to omotteiru 
   J-TOP  he-NOM best-COP-FIN-FOC-MOD-C think 
   ?John thinks ?self? is the best.? 
This data ilustrate that no-da-roo cannot appear in a domain in which the speaker of the 
whole utterance has the perspective directly, not through someone else?s perspective. The 
use of the pronoun in the embedded clause indicates that the content of the embedded 
 36 
clause is described by the objective reporter. Also it shows that no-da-roo can appear 
when the speaker of the whole utterance has aces through the perspective of someone 
else. Here, the speaker indirectly has aces to the content of the embedded clause 
through John?s perspective. The intuitive idea sems to be that no-da-roo can occur 
where the speaker of the whole utterance has an indirect aces through someone else?s 
perspective. 
In order to implement the notion of ?point of view? syntacticaly, Tenny (to 
appear), and Speas & Tenny (2003) argue that there is an Evidential Phrase or Speech 
Act Phrase situated high at the root clause that handles some of the related phenomena 
discussed here. Although we agre with their esential spirit, we do not follow their 
technical implementation. First, we argue for the presence of Perspective Phrase (PerP) at 
the root of the representation. The specifier of PerP must be filed with a null operator 
that originates from the clause where the speaker of the whole utterance takes an indirect 
perspective. Second, recal the obligatory presence of the particle to with the particles no-
da-roo. We take this observation to indicate that the particle to heads a Subordinator 
Phrase (SubP), the specifier of which is the base-generated position of the null operator 
that binds the variable in no-da-roo. 
 
(110)         PerP       
      u       
             
         CP  Perspective    
      ru      
            
          IP      
      6      
      John    VP      
      ru     
       SubP   V omotta 
            
    OP_i ru     
       CP   Sub to   
            
      ru     
      IP   C no-da-roo_i      
      6           
                
The base-generated position of the nul operator indicates that the matrix subject John has 
a perspective toward the embedded clause, which is imediately dominated by the 
atitude predicate. We argue that the movement of the null operator to the root of the 
structure alows the speaker of the whole utterance to have the perspective of the matrix 
subject. 
There is evidence that movement of this null operator obeys ?island? constraints. 
When the clause with no-da-roo is further embedded, the speaker has to be able to take 
 37 
the perspective of each subject in betwen.
19
 Movement of the operator pases through 
the intermediate Perspective Phrase. First, let us take a look at an aceptable example, 
with somewhat simplified structure. 
 
(111) John-wa [zibun-no otoosan-ga [Mary-ga dare-ni ata-no-da-roo]-ka-to 
 J-TOP  self-GEN father-NOM  [-NOM who-DAT met-FIN-FOC-MOD-Q-C 
 omotta]-to sinziteiru 
 thought-C believe 
 ?John believes that self?s father thought who Mary met.? 
(112)         PerP           
      u          
      OP            
         CP  Perspective      
      6          
        IP            
                
     John   V?          
         ru         
       PerP    V        
               
      t   u        
       CP  Perspective     
      6         
        IP           
               
     self?s father V?         
         u        
       CP    V       
      r         
      t   u        
       CP    to       
      6         
       ? no-da-roo-ka       
                  
This two-step movement asures that the speaker of the whole utterance is connected to 
the most deeply embedded clause. The lower part of the movement alows John to take 
the perspective of John?s father, and the higher part of the movement alows the speaker 
to take the perspective of John. In fact, this indicates that the speaker does not directly 
take the perspective of John?s father; the speaker of the whole utterance takes the 
perspective of John?s father though the perspective of John. 
                                                
19
 Obviously, this looks a lot like the blocking efect of long distance reflexives discussed 
in the literature (e.g., Katada 1993, Huang & Liu 2001). 
 38 
There is evidence that it is impossible for the speaker to directly take the 
perspective of John?s father. The following unaceptable example ilustrates the point. 
 
(113) * John-wa [kare-no otoosan-ga [Mary-ga dare-ni ata-no-da-roo]-ka-to 
  J-TOP  he-GEN father-NOM  [-NOM who-DAT met-FIN-FOC-MOD-Q-C 
  omotta]-to sinziteiru 
  thought-C believe 
  ?John believes that his father thought who Mary met.? 
This example is minimaly diferent from (111); the only diference is that in (113) the 
pronoun kare replaces the reflexive zibun. The use of the pronoun ilustrates that the 
matrix subject John does not take the perspective of his father. This unaceptable 
example can be acounted for by asuming that there is no Perspective Phrase in that 
clause as ilustrated below. 
 
(114)         PerP           
      u          
      OP            
         CP  Perspective      
      6          
        IP            
                
     John   V?          
         ru         
       CP   V        
      6         
        IP           
               
     his father  V?         
         u        
       CP    V       
      r         
      t   u        
       CP    to       
      6         
       ? no-da-roo-ka       
                  
Asuming that movement going from the lowest clause to the root violates a Subjacency-
like island constraint (since it skips the intermediate clause), we can acount for the 
unaceptable example. 
Our analysis can acount for the contrast observed in (105) and (106). Let us 
asume that the particles no-da-roo contain a variable bound by the speaker via a null 
operator. Asuming that this places a variable / an index ?I-speaker? on the particles, the 
example in (106a) that contains the pronoun watasi ?I? is a case of a Principle B violation, 
 39 
since the predicate sabisi ?sad? contains the pronoun ?I?. This does not arise when the 
subject is not the first person. The representation is shown below. 
 
(115)  a.  PersP      b.    PersP     
    u       u   
    OP_i        OP_i     
     CP  Pers      CP  Pers  
    ru       ru   
      t_i         t_i     
     IP  no-da-roo_i     IP  no-da-roo_i  
       6         6   
    watasi sabisi_i     Tanaka-san sabisi_k  
Also, the cases where there are no particles, no-da-roo is also acounted for. Recal that 
Kuroda (1973) cals the form without those particles a ?reportive? style. We can take this 
to mean that the speaker is not involved in evaluating the proposition. Then, let us 
asume that in those cases, the head C caries an index ?non-speaker?. We can acount 
for the unaceptable example (105b) by invoking Principle B, since the predicate 
contains a pronoun ?non-speaker?. 
 
(116)  a.  CP       b.  CP  
    ru       ru   
                
     IP  non-speaker_k    IP  non-speaker_k 
       6         6   
    watasi sabisi_i    Tanaka-san sabisi_k  
Next, the incompatibility of siritagaru ?wonder?, siraberu ?investigate? and the 
particles no-da-roo can be explained by asuming that there is no place to generate the 
null operator. Recal that the null operator in the embedded clause is generated at the 
specifier of SubP, headed by the particle to. It is plausible since we independently 
establish the strong connection betwen the presence of no-da-roo and to. Since those 
interogative predicates do not alow the particle to to be generated (hence there is no 
SubP), the nul operator cannot be generated. Therefore, it is impossible for the speaker 
of the whole utterance to take the point of view in the embedded clause. 
Finaly, we have to explain why interogative clauses can appear under omou 
?think? only when the interogative clause contains the particles no-da-roo. Recal that 
when there is no-da-roo in the embedded clause, the null operator is generated at [Spec, 
SubP]. By moving it to matrix [Spec, PersP], it caries an index ?I-speaker?. This 
establishes the link betwen the speaker and the head of SubP to. Intuitively this creates a 
?barier? for the interogative clause to be falsely linked to the matrix predicate omou 
?think?. We can implement this idea by asuming that the interogative clause (or the 
interogative particle ka) contains PRO, which wil be bound by the closest subject-like 
element. Since the speaker is the closest potential controlling entity, this obviates the 
unwanted link betwen the interogative clause and the matrix predicate omou ?think.? On 
the other hand, when there is no no-da-roo in the interogative clause, the 
complementizer / subordinator to does not cary the speaker index, and therefore it does 
 40 
not disrupt the unwanted link betwen the interogative clause and the matrix predicate 
omou ?think?. Since a link betwen them is not a gramatical option due to the semantic 
selection, the example is ungramatical. The following diagram ilustrates the cases. 
 
(117)   a.  CP   b.     PersP       
    ru       u     
           OP_i       
     IP   C       CP  Pers   
       6        6     
     VP          IP       
    ru            
    CP  V omou      VP      
            ru     
   CP   C to        CP    V omou   
  ru                
  IP  C ka       t_i   ru      
                CP   C  to_speaker  
                   
            IP   C ka      
                      
 
 
6. Lower Modal Elements and Negation 
Since we have sen that the Mood head is playing a crucial role in licensing exclamative 
wh-phrases, and since the ood head roo interacts with other modal elements, we 
provide some more discussion on these here. Although we wil label these as ?modal,? 
following Johnson (2003), the categorial status of these elements is unlikely to be 
uniform. We saw two types of modal elements: one kind of modal is compatible with roo 
(which is required in exclamatives), and the other is not. This section discusses these two 
modals in a litle more detail, and we propose a particular left periphery clausal structure 
for Japanese. Specificaly, we argue that there are at least two functional heads that are 
manifested by overt modal elements. It would be nice if the two types of modal elements 
noted above belong to diferent clases of modals, say, in the sense of Cinque (1999). 
This section discusses the interaction betwen exclamatives and the presence of a 
few kinds of modal elements. The basic observation we saw in sections 2 and 5 is that 
evidential morphemes in Japanese, such as rasi and yooda ?likely, sem? are 
incompatible with exclamatives. On the other hand, epistemic morphemes such as hazu 
?must? and beki ?should? can appear in exclamatives. 
Asuming Cinque?s (1999) analysis in which diferent modal elements are 
represented as diferent functional projection in the CP domain, we argue that the 
exclamative particles no-da-roo, especialy roo, also form an evidential morpheme (cf. a 
mood marker; Bhat & Yoon 1992), which plays a crucial role in licensing exclamatives 
in Japanese. The incompatibility betwen exclamatives and evidential morphemes rasi 
or yooda can be acounted for by asuming that there is one slot in the Evidential Phrase 
(or there is only one Evidential Phrase) in the sense of Cinque (1999) (se Hara 2006, 
 41 
Tenny to appear, and many references cited therein for some recent discussion of 
evidentiality in Japanese). One possible implementation of the basic idea is that there is a 
null operator that is asociated with the evidential morphemes rasi and yooda ?sems? 
since they are scope-bearing elements. The incompatibility betwen those morphemes 
and exclamatives can be acounted for if those null operators and exclamative wh-
phrases are competing for the same position. We argue that [Spec,MoodP] is the position 
where those elements are potentialy licensed. 
On the other hand, epistemic morphemes are compatible with exclamatives since 
those morphemes do not compete for the same slot with exclamative licensor in the 
functional projection. Furthermore, it wil be shown that evidential morphemes can 
appear in wh-interogatives in Japanese, suggesting that there is no particular evidential 
morpheme involved in licensing wh-interogatives. Then, we can tie the diference 
betwen exclamatives and interogatives to the presence - in fact requirement - of a 
particular evidential morpheme in the structure. 
 
 
6.1 Evidential: rasi, yooda 
Morphologicaly, rasi and yooda ?sems/likely? occur after a verb with a tense 
morpheme, and it sems that they are highly gramaticalized morphemes, for they 
exhibit some peculiar idiosyncratic properties. For instance, rasi ?likely? shows the 
inflection paradigm of adjectives as shown in (118) and (119), but it is preceded by a 
verb with a tense morpheme. 
 
Adjective-like inflection 
(118) John-ga ki-ta {rasi-i / rasi-kata} 
 John-NOM come-PST  likely-PRS / likely-PST 
 ?It is/was likely that John came.? 
(119) Gasorin-ga {taka-i / taka-kata} 
 gas-NOM  expensive-PRS / expensive-PST 
 ?Gas is/was expensive.? 
The verb occurring imediately before rasi in (118) shows a past tense marker; the verb 
with a non-past tense marker is also aceptable, as shown here. 
 
(120) John-ga ku-ru {rasi-i / ?rasi-kata} 
 John-NOM come-PRS  likely-PRS / likely-PST 
 ?It is/was likely that John comes.? 
In (118), the modal element is imediately preceded by a verb, and in the translation, we 
used ?It is/was likely that ??, but in fact, a complementizer / subordinator to cannot 
appear betwen them. 
 
No complementizer 
(121) * John-ga ki-ta to {rasi-i / rasi-kata} 
  John-NOM come-PST C  likely-PRS / likely-PST 
  ?It is/was likely that John came.? 
 42 
Another modal yooda ?sem? does not have the adjective-like inflectional 
paradigm. It sems reasonable to asume that da at the end of the modal is a copula, 
which suggests that the rest is a nominal expresion, in terms of categorial status. Due to 
the copula at the end, it takes exactly the same inflectional paradigm as the copula. As we 
saw with rasi ?likely?, the tense-marked verb must appear imediately before those 
modals (either past or non-past), and the complementizer to cannot appear betwen the 
verb and the modal. 
 
Yoo is followed by a copula 
(122) John-ga ki-ta {yoo-da / yoo-dat-ta} 
 John-NOM come-PST  sem-COP.PRS / sem-COP.PST 
 ?It sems/semed that John came.? 
(123) John-ga ku-ru {yoo-da / ?yoo-dat-ta} 
 John-NOM come-PST  sem-COP.PRS / sem-COP.PST 
 ?It sems/semed that John comes.? 
Complementizer cannot intervene 
(124) * John-ga ki-ta-to yoo-da 
  John-NOM come-PST-C sem-COP.PRS 
Another methods of complementation we should look at that is usualy available for 
nominal complementation is to-no and toiu. A clause can be embedded under a noun by 
using those complementizers as shown in (125). However, yooda does not take those 
methods of complementation, either. 
 
To-no/toiu complementation 
(125) a. John-ga kaeta to-no uwasa 
  J-NOM left C rumor 
 b. John-ga kaeta toiu uwasa 
  J-NOM left C rumor 
  ?a rumor that John left? 
(126) a. * John-ga kaeta to-no yoo-da 
   J-NOM left C sem-COP 
 b. * John-ga kaeta toiu yoo-da 
   J-NOM left C sem-COP 
   ?It sems that John left.? 
 
 
6.2 Epistemic: bekida, hazuda 
Now, we introduce two other modals beki-da ?should? and hazu-da ?must? that do not 
raise problems when they occur with the modal roo. As is visible from the ending of 
these modals, they have a copula at the end, as we have sen with yoo-da. Then, it could 
be that beki and hazu themselves are categorialy nominal expresions. Although 
categorialy they are similar, there are at least two diferences betwen the two modals. 
At this point, I have no acount for the diference betwen the two modals, and we wil 
simply note the facts. 
 43 
First, when beki ataches to a verb, the verb cannot cary a past tense marker. In 
other words, the verb is in the non-past form. There is no requirement like that for hazu. 
 
Verb cannot be in past tense form with beki 
(127) John-wa gakkoo-e iku-beki-da 
 John-TOP school-to go-must-COP.PRS 
 ?John should go to school.? 
(128) *John-wa gakkoo-e it-ta-beki-da 
 John-TOP school-to go-PST-must-COP.PST 
 ?John should have gone to school.? 
No requirement of tense marking with hazu 
(129) John-wa gakkoo-e iku-hazu-da 
 John-TOP school-to go-must-COP.PRS 
 ?John must go to school.? 
(130) John-wa gakkoo-e it-ta-hazu-da 
 John-TOP school-to go-PST-must-COP.PST 
 ?John must have gone to school.? 
Second, even though both of them can occur in relative clauses, it is possible only 
when they cary a past tense marker. Their behaviors diverge if we look at cases other 
then the past tense paradigm. The following pair ilustrates that they can occur in relative 
clauses. 
 
Beki and hazu can appear in relative clauses if in the past tense form 
(131) [John-ga toku-beki-dat-ta] mondai 
  John-NOM solve-should-COP-PST problem 
 'A problem that John should have solved' 
(132) [John-ga toku-hazu-dat-ta] mondai 
  John-NOM solve-must-COP-PST problem 
 'A problem that John must have solved' 
This observation is crucial since we have aserted a generalization that modals that block 
the dependency betwen the exclamative wh-phrase and the licensing particle cannot 
appear in relative clauses. Later, we show examples in which the exclamative wh-phrase 
is licensed even though the sentence contains these modals. 
Now, when these modal elements do not cary the past tense marker, a mysterious 
patern emerges. Note that when beki and hazu are followed by a non-past pre-nominal 
form -na, examples are not aceptable at al. There is a sharp contrast betwen the pair in 
(131) and (132) where the copula is in the past tense form and the pair in (133) and (134) 
where the copula is in the non-past tense form. It is unclear why this is the case. Second, 
the example with hazu is aceptable if the whole pre-nominal modifier is marked with the 
genitive case marker no. On the other hand, with respect to beki, having the genitive case 
 44 
marker does not help at al. In fact, the example is aceptable when there is no marker 
atached to beki.
20
 
 
Mysterious requirement in relative clauses 
(133) [John-ga toku-beki-?/-*na/-*no] mondai 
  John-NOM solve-should-?/-COP.PRS/-GEN problem 
 'A problem that John should have solved' 
(134) [John-ga toku-hazu-*?/-*na/-no] mondai 
  John-NOM solve-must-?/-COP-PRS/-GEN problem 
 'A problem that John must have solved' 
Despite a few diferences betwen these two modal elements, it is important for to 
note that they do not interfere with the licensing of the exclamative wh-phrase, which is 
shown in (135) and (136). Also note that the presence of either modal element does not 
disrupt the wh-dependency in wh-interogative clauses either. 
 
Hazu and beki do not block wh-exclamatives 
(135) Mary-wa nante takusan-no sinnyuusyain-o yatou hazu-dat-ta  
 ary-TOP wh many-GEN employee-ACC hire must-COP-PST  
 no-da-roo 
 FIN-FOC-MOD 
 `What a lot of new employees Mary was supposed to hire!' 
(136) John-wa nante takusan-no hon-o kau beki-dat-ta no-da-roo 
 John-TOP wh many-GEN book-ACC buy should-COP-PST FIN-FOC-MOD 
 `What a lot of books John should have bought!' 
Hazu and beki do not block wh-interogatives 
(137) Mary-wa donokurai takusan-no sinnyuusyain-o yatou hazu-dat-ta 
 ary-TOP wh many-GEN employee-ACC hire must-COP-PST 
 no-desu-ka 
 FIN-FOC.POLITE-Q- 
 `How many new employees was Mary supposed to hire?' 
(138) John-wa donokurai takusan-no hon-o kau beki-dat-ta 
 John-TOP wh many-GEN book-ACC buy should-COP-PST 
 no-desu-ka 
 FIN-FOC.POLITE-Q 
 `How many books should John have bought?' 
The example of the wh-exclamative with hazu is, as far as we can tel, as good as its 
interogative counterpart. A slight complication is that (135) is interpreted as a 
counterfactual, which might contribute to the dificulty of interpreting the sentence. 
Consider the following context in which (135) can be uttered: Mary is in charge of hiring 
new employees in a company. The company wil open a new division soon, so there is a 
                                                
20
 It is unclear why beki can appear as a bare form only in relative clauses. In other words, 
when it appears in the matrix clause, it has to be folowed by a copula (presumably then it 
can cary tense information). 
 45 
big demand for hiring more workers. She said to a friend of hers, John, "I was asked to 
hire 50 people in one wek, but I could only hire 40 people total." John was surprised at 
the fact that she was supposed to hire 50 people, since the number is realy large. Then he 
can utter something like (135). Note that this interpretation corresponds to the so-caled 
"amount" reading in the sense of Heycock (1995). Informaly, it can be represented as in 
(139). 
 
The "amount" reading 
(139) There is a number such that Mary must have hired that number of people; the 
number is extremely large. 
Another observation to be noted with respect to the interpretation of (135) is that it also 
has the interpretation Heycock (1995) cals the "referential" reading, which could be, 
again informaly, represented as in (140). 
 
The "referential" reading 
(140) There is a set of people that Mary must have hired; the cardinality of that set is 
extremely large. 
The context that would be appropriate for the "referential" reading is the following: She 
said to a friend of hers, John, "I was asked to hire 50 people in one wek, but I could only 
hire 10 people total." John was surprised at the fact that the gap betwen the number of 
people she was supposed to hire and the actual number of people she managed to hire is 
realy large. The "referential" reading is also available with (135), since it is natural to 
utter (135) in that context. With respect to the example with beki, the same observation is 
observed. Examples like (136) are aceptable either with the "amount" reading or the 
"referential" reading. 
Summarizing the discussion so far: modal elements sometimes block the 
dependency betwen the exclamative wh-phrase nante and the licensing particles. We 
have sen that two modal elements, rasi and yooda, disrupt the dependency in 
exclamatives, but not the one in interogatives. Two other modal elements, bekida and 
hazuda, on the other hand, do not block either the dependency in exclamatives or the one 
in interogatives. Also we noted that exclamatives with bekida and hazuda demonstrate 
ambiguity with respect to the scope of degre phrases. Specificaly, both the "amount" 
reading and the "referential" reading sem to be available for the examples. 
 
 
6.3 Negation 
Another element we can discuss in the same vein is negation. Regarding the 
generalization pointing to the correlation betwen the lack of blocking efects in 
exclamatives and the ability to appear within relative clauses, negation conforms to the 
generalization, though negation is not a modal element. The bottom line is that it can 
appear in relative clauses, and it does not block the dependency in exclamatives.
21
 
                                                
21
 Oda (2004) discusses negation in exclamatives in a variety of languages, and concludes 
that exclamatives simply do not admit negation. If the analysis ilustrated in this section 
 46 
The following pair of examples shows that the presence of negation does not 
disrupt the licensing of the exclamative wh-phrase, but it should be noted that the 
example is not ambiguous, unlike the exclamative sentences with a certain modal 
element: it only has the "referential" reading. 
 
Negation is OK and the example is unambiguous. 
(141) Mary-wa nante takusan-no hon-o yom-anakat-ta no-da-roo 
 ary-TOP wh many-GEN book-ACC read-NEG-PST FIN-FOC-MOD 
 ?What a lot of books Mary did not read!' 
The relevant context is that she was supposed to read, say, 50 books (for her 
disertation?). It turns out that she managed to finish only 10 books. The cardinality of 
the set of books she did not read, which is 40, is extremely large. The point in examples 
like (141) cary over to their counterpart in interogatives. We do not know what the 
"amount" reading is supposed to mean with negation. 
There is one modal that does not fit the patern ilustrated above. We do not have 
an acount for it, but give the observation here. A modal ni-tigai-nai 'certain' sems to 
block the licensing of exclamatives, but it does not block the licensing of interogatives; 
nonetheles, it can appear inside a relative clause. 
 
It blocks exclamatives, but not interogatives. 
(142) *John-wa nante takai hon-o kata ni-tigai-nai no-da-roo 
 John-TOP wh expensive book-ACC bought certain FIN-FOC-MOD 
 `What an expensive book John is certain to have bought!' 
(143) John-wa donokurai takai hon-o kata ni-tigai-nai  
 John-TOP wh expensive book-ACC bought certain  
 no-desu-ka 
 fin-FOC.POLITE-Q 
 `How expensive a book is John certain to have bought?' 
It can appear in relative clauses. 
(144) [John-ga kat-ta ni-tigai-nai] hon 
 John-NOM buy-PST certain book 
 'the book that John surely bought' 
A potentialy relevant observation is that the modal ni-tigai-nai is a morphologicaly 
complex expresion: ni can be some sort of a preposition, tigai 'diference' sems to be a 
noun, and nai is a negative morpheme that takes the inflectional paradigm of an adjective. 
It is conceivable that they are some sort of frozen expresion, since, for example, a case-
marker cannot atach to the noun part tigai. Example (146) ilustrates that the nominative 
case-maker can atach to the noun, but then the modal meaning sems to be lost. 
 
(145) Sono hon-wa John-ga kat-ta ni-tigai-(*ga)-nai 
 thatbook-TOP John-NOM buy-PST MOD 
 'It is certain that, that book, John bought.' 
                                                                                                                                            
is on the right track, we are led to conclude that exclamatives can admit negation, but the 
degre restriction cannot be reconstructed across negation. 
 47 
(146) Sono hutatu-no hon-wa nakami-ni tigai-ga nai 
 thatwo.clasifier-GEN book-TOP content-DAT diference-NOM NEG 
 'With respect to their content, there is no diference in those two books. 
Note also that when ni-tigai-nai is used as a modal element, it directly follows the tensed 
verb; on the other hand, in (146), it follows a noun. Then, it sems certain that the case 
where ni-tigai-nai blocks the exclamative sentence shows that it is the modal element that 
is responsible for the unaceptability of (142). 
 Summarizing the section so far, we saw that there are at least two kinds of modal 
elements pertaining to the licensing of the exclamative wh-phrase nante. The first kind of 
modal elements blocks the dependency, and this observation correlates with the fact that 
they cannot appear in relative clauses. The other kind of modal elements (including 
negation), do not block the dependency, and they are alowed to appear in relative clauses. 
Another observation is that when the second kind of modal elements are used in 
exclamatives, there are two interpretations available, except for the negation case. The 
two interpretations correspond to the "amount" interpretation and the "referential" 
interpretation, discussed, for example, by Heycock (1995). 
 
 
7. Sumary 
This chapter discussed thre morphemes that are crucial to Japanese exclamatives. It 
ilustrated that each of the morphemes is a required part of Japanese exclamatives and 
properties of those morphemes are discussed. Especialy, we have discussed the 
properties of the mood morpheme roo, which diferentiate Japanese exclamatives from 
interogatives. It was shown that various properties of the morpheme roo can be captured 
by asuming that it heads a functional projection, which encodes the perspective of the 
speaker of the utterance or point of view, building upon the insight of Tenny (to appear) 
and Hara (2006). Using the general framework proposed in Rizi (1997) and Cinque 
(1999), we can syntacticaly implement some of the notions that have been handled 
mostly in semantics or pragmatics. By using syntactic projections in the CP domain, we 
succesfully captured various observations concerning the mood morpheme roo, showing 
that the phenomena are in fact syntactic. 
 
 48 
 
CHAPTER 3: LONG DISTANCE DEPENDENCIES IN JAPANESE 
EXCLAMATIVES 
 
1. Introduction 
Based on the general properties of a few particles that appear in Japanese exclamatives, 
we now turn to the discussion of the dependency betwen the exclamative wh-phrase 
nante and the exclamative particles. We show that the dependency betwen them can be 
long-distance, as long as various standard island conditions are avoided. In the next 
section, we first investigate restrictions of embedded exclamatives, contrasted with the 
standard embedded interogatives. Obviously, the syntactic environment where 
exclamatives are embedded is quite diferent than that for interogatives (Grimshaw 1979, 
for example). After discussing those restrictions, we examine which island constraints 
exclamatives obey. In section 3, we propose a mechanism intended to capture the island 
efects observed here and observations with respect to the interaction betwen 
interogatives and exclamatives. We argue that various observations introduced there are 
intervention efects. Later in the chapter, we discuss the properties of the exclamative 
wh-phrase nante in detail. We introduce some new observations and discus some 
semantic isues pertaining to the exclamative wh-phrase nante. Section 4 reviews 
properties of the exclamative wh-phrase nante. Some variations of the form are discussed. 
We show that nanto-yuu and nanto, both of which are variants of nante, ilustrate 
diferent syntactic behavior. Nanto-yuu is the form that modifies a projection of a noun, 
while nanto directly modifies either adverbs or adjectives. Finaly in section 5, some 
semantic isues are discussed. Specificaly, we discuss how to obtain the meaning of 
multiple exclamatives found in Japanese. We argue that the two degre phrases involved 
in multiple exclamatives must be related in some particular way. Section 6 summarizes 
the chapter. 
 
 
2. Licensing Exclamatives: Multi-clause 
2.1 Complement Selection 
This section discusses the compatibility of exclamative clauses and certain kinds of 
predicates, as not al predicates can embed exclamatives. Eliott (1971, 1974) and 
Grimshaw (1979) investigated the distribution of English embedded exclamatives, but as 
far as Japanese exclamatives are concerned, there sems to be no study that investigates 
their distribution in detail, at least in the framework we adopt here.
2
 A point we have to 
pay atention to is complementizer selection in Japanese. Predicates in Japanese use 
either the koto/no-complementizer or the to complementizer, although many predicates 
use one predominantly (se Kuno 1973, Nakau 1973, Josephs 1976 for relevant 
discussion). 
One clear observation is that exclamatives cannot be embedded under the koto-
complementizer. This is expected given the discussion in the previous chapter. Recal the 
incompatibility betwen the licensor of exclamatives and the koto-complementizer. Then 
                                                
2
 We take a look at the English cases in Chapter 4. 
 49 
complementation with the to-complementizer is the relevant environment to investigate 
which predicates alow embedded exclamatives. 
It turns out that the distribution of embedded exclamatives in Japanese is quite 
limited. Besides the environment of koto-complementation, it is not true that 
exclamatives can be embedded under any to-complementation. We find the embedded 
exclamatives only under two types of predicates. One type of predicate that alows 
embedded exclamatives is omou ?think? and iu ?say?.
23
 The other is the emotional 
predicate, such as odoroku ?be surprised? and akireru ?be amazed.? In this section, we 
focus on the question of where we observe embedded exclamatives. 
First, verbs that require an interogative clause complement (i.e., tazuneru ?ask? 
and kiku ?ask?) cannot take an exclamative clause as a complement. They require an 
interogative clause with the particle ka to appear in the imediate complement position, 
though the complementizer to can optionaly appear. 
 
Interogative predicates 
(1) a. * John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
   J-TOP  -NOM WH many-GEN student-DAT angry 
   noda-roo-ka]-(to) tazuneta 
   FOC-MOD-C asked 
   ?John asked how very many students Mary got angry at.? 
 b. * John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
   J-TOP  -NOM WH many-GEN student-DAT angry 
   noda-roo-ka]-(to) kiita 
   FOC-MOD-C asked 
   ?John asked how very many students Mary got angry at.? 
Presumably, those examples are unaceptable due to the lack of an interogative 
complement in the embedded clause. Other interogative verbs, such as siritagaru 
?wonder? and siraberu ?investigate,? show the same patern.
24
 
Next, verbs that only take a proposition as their complement are unable to embed 
exclamatives (sinziru ?believe,? syutyoo-suru ?claim?). They can use the complementizer 
to to embed a clause. 
 
(2) John-wa [Mary-ga takusan-no hon-o yonda]-to sinzi-teiru 
 J-TOP  -NOM many-GEN book-ACC read-C believe-be 
 ?John believes that ary read many books.? 
Embedded Exclamatives Imposible 
(3) a. * John-wa [Mary-ga nante takusan-no hon-o yonda 
   J-TOP  -NOM WH many-GEN book-ACC read 
   noda-roo]-to sinziteiru 
   FOC-MOD-C believe 
   ?John believes how very many books Mary read.? 
                                                
23
 I admit that it is peculiar to say that those two verbs form a natural clas. Whether there 
are verbs that patern with those verbs should be investigated further. 
24
 Verbs such as siritagaru ?wonder? and siraberu ?investigate,? on the other hand, do not 
alow the complementizer to to appear. 
 50 
 b. * John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
   J-TOP  -NOM WH many-GEN student-DAT got.angry 
   nodaroo]-to syutyoo-siteiru 
   FOC-MOD-C claim-be 
   ?John claims how very many students Mary got angry at.? 
Second, a clas of factive verbs, sitieru ?know? and oboeteiru ?remember,? do not 
alow exclamative complements. This clas of verbs uses the formal noun koto or no 
when they embed a clause in general, but use of the complementizer to sems to be 
possible in some limited cases.
25
 The factivity asociated with this clas of verbs can be 
shown by the awkwardnes of the following discourse continuation. 
 
Factivity 
(4) a. John-wa [Mary-ga kita]-koto-o siteiru 
  J-TOP  -NOM came-FN-AC know 
  ?John knows that ary came.? 
 
 b. - # demo, Mary-wa kite-nai-yo 
    but -TOP came-NEG-AF 
    ?But, Mary hasn?t come.? 
The speaker of the utterance in (4) cannot negate the embedded proposition. This is one 
of the most widely used tests detecting factivity. The awkwardnes in the continuation 
indicates that the speaker presupposes the truth of the propositional content in the 
embedded clause, which is ?that Mary came?. Examples with arguably non-factive 
predicates such as omou ?think? do not show the awkwardnes, on the other hand. 
 
Non-factive 
(5) a. John-wa [Mary-ga kita]-to omotta 
  J-TOP  -NOM came-C thought 
  ?John thought that ary came.? 
 b. - demo, Mary-wa kite-nai-yo 
   but -TOP came-NEG-AF 
   ?But, Mary hasn?t come.? 
                                                
25
 Se Nakau (1973), Kuno (1973) and Josephs (1976) for discussion of which 
complementizer to use and its semantic and syntactic properties. One instance of to-
clause with siteiru ?know? is as follows, which is slightly strange, compared to the 
example with koto: 
(i) ?John-wa [Mary-ga kita]-to sit-tei-masu-ka? 
(i) John-wa [ary-ga kita]-koto-o sit-tei-masu-ka? 
There is one piece of evidence that the status of the to-clause with the verb siteiru 
?know? is not quite the same as the to-clause with omou ?think.? To my ear, the wh-
dependency across the clause with to and siteiru is not so good. 
(i) John-wa [dare-ga kita]-to omot-tei-masu-ka 
(ii) ?John-wa [dare-ga kita]-to sit-tei-masu-ka? 
It is fairly wel-known that adjunct wh-phrases cannot appear in various islands, but the 
environment where even argument wh-phrases cannot appear is extremely limited. 
 51 
Coming back to exclamatives, regardles of which complementizer those factive verbs 
use to embed clauses, exclamatives are not aceptable under those verbs.
26
 
 
Factive Predicates 
(6) * John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
  J-TOP  -NOM WH many-GEN student-DAT angry 
  noda-roo]-koto-o/to siteiru 
  FOC-MOD-FN-AC/C know 
  ?John knows how very many students Mary got angry at.? 
(7) * John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
  J-TOP  -NOM WH many-GEN student-DAT angry 
  noda-roo]-koto-o/to oboeteiru 
  FOC-MOD-FN-AC/C know 
  ?John remembers how very many students Mary got angry at.? 
Another clas of verbs that we are interested in includes kitai-suru ?expect,? and 
nozomu ?hope.? They are similar to factive verbs in the sense that they usualy use a 
complementizer koto or no to embed a clause, but the complementizer to can sometimes 
appear.
27
 Exclamatives are not licensed under these predicates. 
 
Irealis Predicates 
(8) * John-wa [Mary-ga nante takusan-no hon-o yomu 
  J-TOP  -NOM WH many-GEN book-ACC read 
  noda-roo]-koto-o/to kitai-si-teiru 
  FOC-MOD-FN-AC/C expect-do-be 
  ?John expects how very many books Mary reads.? 
(9) * John-wa [Mary-ga nante takusan-no hon-o yomu 
  J-TOP  -NOM WH many-GEN book-ACC read 
  noda-roo]-koto-o/to nozon-deiru 
  FOC-MOD-FN-AC/C hope-be 
  ?John hopes how very many books Mary reads.? 
The fourth type of verbs includes omou ?think? and iu ?say,? which can embed 
exclamatives. Those verbs use the complementizer to in order to embed a clause, and in 
general koto or no is not used. 
 
Embedded Exclamatives Possible 
(10) a. John-wa [Mary-ga nante takusan-no hon-o yonda 
  J-TOP  -NOM WH many-GEN book-ACC read 
  noda-roo-ka]-to omotteiru 
  FOC-MOD-Q-C think-be 
  ?John thinks, how very many books Mary read.? 
                                                
26
 Raising the question, What is the parametric diference betwen English and Japanese? 
27
 Similar to what we have sen with factive verbs, the distribution of the to-clause is 
quite limited. The to-complementizer sems to block the wh-dependency in general. 
(i) *John-wa [dare-ga kuru]-to kitai-si-tei-masu-ka? 
 52 
 b. John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
  J-TOP  -NOM WH many-GEN student-DAT angry 
  noda-roo-ka]-to ita 
  FOC-MOD-Q-C said 
  ?John said, how very many students Mary got angry at.? 
Finaly, emotive verbs such as odoroku ?be surprised? and akireru ?be amazed? 
can take an exclamative clause with the complementizer to. This clas of verbs is 
diferent from ?factive? verbs such as siteiru ?know? in that they alow embedding a 
clause with the complementizer to. Recal that siteiru ?know? does not use the 
complementizer to to embed a clause. 
 
Embedded Exclamatives Possible 
(11) a.  John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
   J-TOP  -NOM WH many-GEN student-DAT angry 
   noda-roo]-to odoroiteiru 
   FOC-MOD-C surprised-be 
   ?John was surprised how very many students Mary got angry at.? 
 b.  John-wa [Mary-ga nante takusan-no gakuse-ni okotta 
   J-TOP  -NOM WH many-GEN student-DAT angry 
   noda-roo-to akireteita 
   FOC-MOD-C surprised-be 
   ?John was amazed how very many students Mary got angry at.? 
The following table summarizes what we have sen so far. Given that the verbal particles 
in exclamatives are not compatible with koto-complementation, the ability for a predicate 
to use the complementizer to is an absolute requirement. 
 
(12) Summary 
 Koto-comp To-comp 
A. tazuneru ?ask? 
 kiku ?ask? 
No Yes 
exc = No 
B. sinziru ?believe? 
 syutyoosuru ?claim? 
Yes Yes 
exc = No 
C. siteiru ?know? 
 oboeteiru ?remember? 
Yes Yes? 
exc = No 
D. kitaisuru ?hope? 
 nozomu ?wish? 
Yes Yes? 
exc = No 
E. omou ?think? 
 iu ?say? 
No Yes 
exc = Yes 
F. odoroku ?be surprised? 
 akireru ?be amazed? 
Yes Yes 
exc = Yes 
 
There are two remarks to make for the distribution of embedded Japanese exclamatives. 
First, it is unexpected that factive predicates such as siteiru ?know? cannot embed 
exclamatives, given that it has been observed that exclamatives can be embedded with 
factive predicates in languages like English. On the other hand, another clas of factives 
 53 
that we cal ?emotive? predicates, in fact, alow exclamatives to occur in the embedded 
clause. As we mentioned above, the diference betwen the two clases of verbs is that 
the emotive predicates are able to use the complementizer to to embed a clause. 
Asuming that the distribution of embedded exclamatives is regulated at least partialy by 
the semantic selection of the predicate, it is expected that exclamatives in Japanese can be 
embedded with factive predicates. However, the inability of exclamatives to be 
embedded under factive predicates in Japanese can be related to an observation that 
exclamatives resist undergoing nominalization. In other words, due to the property of 
factive predicates that the nominal complementizer koto or no has to be used, and since 
exclamatives are not good at undergoing nominalization, embedding exclamatives with 
factive predicates is unaceptable in Japanese. The independent generalization that 
exclamatives resist nominalization arises from the following observation.
28
 
 
Nominalization 
(13) a. the proposition / fact that John met a tal guy 
 b. the question which guy John met 
 c. *the exclamation what a tal guy John met 
In English, it is possible for a proposition and a question to be nominalized by adding an 
appropriate nominal head, but for some unknown reason, such an option is not available 
with an exclamation. Furthermore, there is a contrast in Japanese betwen interogatives 
and exclamatives with respect to the ability for a case-maker to appear after the clause. 
The following set of data indicates that an interogative clause can be marked with a 
case-maker, suggesting that an interogative clause can be nominalized. On the other 
hand, such an option is never available with an exclamative clause. 
 
Interogatives 
(14) a. John-wa [Mary-ga doko-e ita]-ka(-o) tazuneta 
  J-TOP  -NOM where-to went-Q-ACC asked 
  ?John asked where ary went.? 
 b. John-wa [Mary-no ikisaki]-o  tazuneta 
  J-TOP  -GEN destination-ACC asked 
  ?John asked Mary?s destination.? 
Exclamatives 
(15) a. * John-wa [Mary-ga nante mazusi]-no-da-roo(-o) nageita 
   J-TOP  -NOM WH poor-FIN-FOC-MOD-ACC lamented 
   ?John lamented how very poor Mary is.? 
 b.  John-wa [Mary-no mazusisa]-o nageita 
   J-TOP  -GEN poverty-ACC lamented 
   ?John lamented Mary?s poverty.? 
Another contrast betwen the two is that an interogative clause can appear as a topic 
(marked with a topic marker), while an exclamative clause cannot. 
 
                                                
28
 I would like to thank Juan Uriagereka (p.c.) for insightful discussion on this point. 
 54 
Interogatives 
(16)  [Mary-ga doko-e ita]-ka-wa wakar-anai 
   -NOM where-to went-Q-TOP know-NEG 
  ?It is not known where Mary went.? 
Exclamatives 
(17) * [Mary-ga nante mazusi]-no-da-roo-wa odoroki-da 
   -NOM WH poor-FIN-FOC-MOD-TOP surprise-COP 
  ?It is surprising how very poor Mary is.? 
If those properties of exclamative clauses indicate the extent to which the exclamative 
clause can be nominalized, it sems plausible to claim that the fact that they cannot occur 
with factive predicates does not have so much to do with some semantic restriction. In 
English, the factive predicates alow embedding without nominalization; hence they can 
embed an exclamative clause. On the other hand, Japanese factive predicates never alow 
embedding without nominalization. Therefore, an exclamative clause in Japanese cannot 
be embedded with factive predicates. This observation leads us to suggest that there is a 
parameter with respect to the category that the factive predicates select. 
Some discussion sems to be in order with respect to the distribution of embedded 
exclamatives. Apparently, exclamative wh-clauses in Japanese can occur with verbs such 
as omou ?think? and iu ?say.? On the other hand, they cannot occur with verbs such as 
sinziru ?believe? and syutyoosuru ?claim.? In languages like English, none of these 
predicates can embed exclamatives. The diference betwen the two types of predicates 
can be that the ?content? of belief and claim has to be semanticaly a proposition. On the 
other hand, the content of thought and saying has more fredom. It is possible to ?say? a 
question, an order, or an exclamation, but it is unlikely that it is possible to ?believe? a 
question, an order, or an exclamation. One observation that makes a distinction among 
those predicates is the following. When the ?content? of think and say is asked, the wh-
phrase doo ?how? is the most natural wh-phrase to be used, while when the ?content? of 
believe and claim is asked, the wh-phrase nani ?what? sounds beter than doo ?how.?
29
 
 
With ?how? 
(18) a. Kimi-wa doo omotta-no? 
  you-TOP how thought-Q 
  ?What do you think?? 
 b. Kimi-wa doo ita-no? 
  you-TOP how said-Q 
  ?What did you say?? 
                                                
29
 Juan Uriagereka (p.c.) suggests that one way to handle these facts could be that the to-
clause with think and say in Japanese is a parataxis, while the to-clause with believe and 
claim in Japanese is a complement clause. Se Torrego & Uriagereka (2002). 
 55 
With ?what? 
(19) a. ? Kimi-wa nani-o omotta-no? 
   you-TOP what-ACC thought-Q 
   ?What do you think?? 
 b. ? Kimi-wa nani-o ita-no? 
   you-TOP what-ACC said-Q 
   ?What did you say?? 
With ?how? 
(20) a. ? Kimi-wa doo sinzi-teiru-no? 
   you-TOP how believe-be-Q 
   ?What do you believe?? 
 b. ? Kimi-wa doo syutyoo-suru-no? 
   you-TOP how claim-be-Q 
   ?What do you claim?? 
With ?what? 
(21) a. Kimi-wa nani-o sinzi-teiru-no? 
  you-TOP what-ACC believe-be-Q 
  ?What do you believe?? 
 b. Kimi-wa nani-o syutyoo-suru-no? 
  you-TOP what-ACC claim-be-Q 
  ?What do you claim?? 
There is one more diference betwen those predicates. Recal that one way to clasify 
predicates in Japanese is complementizer selection. Sinziru ?believe? and syutyoosuru 
?claim? alows koto-complementizer, while omou ?think? and iu ?say? do not.
30
 
 
Koto-clause only 
(22)  John-wa [Mary-ga sono ryoori-o tukutta]-koto-o sinzi-teiru 
  J-TOP  -NOM that dish-ACC made-FN-AC believe-BE 
  ?John believes that ary made that dish.? 
(23) * John-wa [Mary-ga sono ryoori-o tukutta]-koto-o omot-teiru 
  J-TOP  -NOM that dish-ACC made-FN-AC think-BE 
  ?John thinks that ary made that dish.? 
On the other hand, omou ?think? alows the koto-clause to appear as long as the to-clause 
also appears. Sinziru ?believe? also alows two clauses to appear at the same time.
31
 
 
                                                
30
 The situation is not so simple, since the complementizer selection also depends on the 
tense of the embedded clause. 
31
 Other predicates such as setumesuru ?explain? can take the koto-clause and the to-
clause at the same time. One thing that is a litle puzzling is that with explain, the order of 
the two clauses sems not to mater; either clause can come before the other. But, it 
sems that, with think, examples sound much beter when the koto-clause precedes the to-
clause. 
 56 
Koto-clause and to-clause at the same time 
(24) John-wa [Mary-ga sono ryoori-o tukutta]-koto-o [Susan-ga  
 J-TOP  -NOM that dish-ACC made-FN-AC  S-NOM  
 tukuru-beki-data]-to  omot-teiru 
 make-should-COP.PAST think-BE 
(25) John-wa [Mary-ga sono ryoori-o tukutta]-koto-o [Susan-ga  
 J-TOP  -NOM that dish-ACC made-FN-AC  S-NOM  
 tukuru-beki-data]-to  sinzi-teiru 
 make-should-COP.PAST think-BE 
A related observation is available in English, as wel. Believe alows an NP 
complement, while think does not (unles the NP is a cognate NP, such as I think a 
thought?, which is a litle peculiar, too).
32
 
 
NP complement 
(26) a. I believe this idea. 
 b. ?I think this idea. 
Also, a diference emerges when those predicates are nominalized. 
 
Nominalization 
(27) a. I have a belief that 2 + 2 = 4. 
 b. ?I have a thought that 2 + 2 = 4. 
It is quite fascinating that the apparently quite mysterious observation in Japanese has 
some link to the observation available in English. It is certainly true that, at this point, it 
is not exactly clear how the distribution of embedded exclamatives in Japanese is 
eventualy explained, but it sems clear that the observed patern is not something 
random. 
 
 
2.2 Long Distance Dependency 
This section discusses cases where the exclamative wh-phrase and its licensor are not in 
the same clause. As usual, the dependency is sensitive to various islands. The general 
picture is a bit complex. It sems that the clas of verbs is not the only factor that is 
relevant for the long distance dependency, but the type of complementizer is also playing 
a role in determining whether a long distance dependency is possible. We wil start by 
looking at cases where the long distance dependency is possible, and in the next section, 
we proced to cases where the long distance dependency is impossible. 
The first environment where the long distance dependency of exclamatives is 
alowed is bridge-verb context. As shown in the previous section, this clas of verb 
selects for the complementizer to, and the exclamative wh-phrase in the embedded clause 
can be licensed by the licensor at the matrix clause. 
 
                                                
32
 I would like to thank Howard Lasnik (p.c.) for providing this fact. 
 57 
(28) a. John-wa [Mary-ga nante takusan-no hon-o yonda]-to 
  J-TOP  M-NOM WH many-GEN book-ACC read-C 
  omot-teiru noda-roo 
  think-be FOC-MOD 
  ?How very many books John thought that Mary read!? 
 b. John-wa [Mary-ga nante takusan-no gakuse-ni okotta]-to 
  J-TOP   -NOM WH many-GEN student-DAT angry-C 
  iu noda-roo 
  say FOC-MOD 
  ?How very many students John says that Mary got angry at!? 
The second case of a long distance dependency is found when verbs such as 
sinziru ?believe? or syutyoo-suru ?claim? use the to-complementizer. Recal that 
exclamative wh-clauses cannot occur as a complement of those predicates, but the 
following examples indicate that the dependency across those predicates is possible. 
 
(29) a. John-wa [Mary-ga nante takusan-no gakuse-ni okotta]-to 
  J-TOP  -NOM WH many-GEN student-DAT got.angry-C 
  syutyoo-siteiru noda-roo 
  claim-be FOC-MOD 
  ?How very many students John claims that Mary got angry at!? 
 b. John-wa [Mary-ga nante takusan-no hon-o yonda]-to 
  J-TOP  -NOM WH many-GEN book-ACC read-C 
  sinzi-teiru  noda-roo 
  believe-be FOC-MOD 
  ?How very many papers John believes that Mary read!? 
Finaly, with verbs like kitaisuru ?hope? or nozomu ?wish,? the long distance 
dependency is possible with the koto-complementizer, though they do not in general 
belong to the clas of bridge verbs. 
 
(30) a. John-wa [Mary-ga nante takusan-no hon-o yomu]-koto-o 
  J-TOP  -NOM WH many-GEN book-ACC read-FN-ACC 
  kitai-si-teiru noda-roo 
  hope-do-be FOC-MOD 
  ?How very many books John hopes that Mary reads!? 
 b. John-wa [Mary-ga nante takusan-no hon-o yomu]-koto-o 
  J-TOP  -NOM WH many-GEN book-ACC read-FN-ACC 
  nozon-deiru noda-roo 
  wish-do-be FOC-MOD 
  ?How very many books John wishes that Mary reads!? 
 
 
 58 
2.3 Island Efects 
2.3.1 Factive Island 
One environment where the long distance dependency is blocked can be clasified as 
factive island cases (Ros 1984, Cinque 1990, Hegarty 1992, etc.). It has been shown in 
the literature that argument wh-phrases can appear inside a koto-clause, but adjunct wh-
phrases naze ?why? cannot (Lasnik & Saito 1984:244, Fukui 1988:509).
3
 The basic 
paradigm using the verb okoru ?get angry? is as follows. 
 
Factive island (Lasnik & Saito 1984: 244) 
(31) a.  [Taroo-ga nani-o  te-ni ireta]-koto-o sonnani okot-teiru-no 
    T-NOM what-ACC got-FN-ACC so  get.angry-be-Q 
   ?What are you geting angry at that Taro got?? 
 b. * [Taroo-ga naze sore-o te-ni ireta]-koto-o sonnani okot-teiru-no 
    T-NOM why it-ACC got-FN-ACC  so  got.angry-be-Q 
   ?Why are you geting angry at that Taro got it?? 
The following example ilustrates that the wh-dependency of exclamatives across the 
koto-clause with factive verbs is not alowed. 
 
Factive island 
(32) ? John-wa [Mary-ga nante takusan-no kaisya-o  
  J-TOP  -NOM WH many-GEN company-ACC  
  keee-siteiru]-koto-o oboe-teiru-noda-roo 
  manage-be-FN-AC remember-be-FOC-MOD 
  ?What a lot of companies John remembers that Mary manages!? 
A similar observation is also available with wh-interogatives with how many. 
 
(33) ? John-wa [Mary-ga ikutu-no kaisya-o  
  J-TOP  -NOM how.many-GEN company-ACC  
  keee-siteiru]-koto-o oboe-tei-masu-ka? 
  manage-be-FN-AC remember-be-POLITE-Q 
  ?How many companies does John remember that Mary manages?? 
(34) ? John-wa [Mary-ga ikutu-no pasokon-o  
  J-TOP  -NOM how.many-GEN computer-ACC  
  mot-teiru]-koto-o sit-teiru-masu-ka? 
  own-be-FN-ACC know-be-POLITE-Q 
  ?How many computers does John know that Mary owns?? 
                                                
3
 They did not use the term ?factive?, but the case is clasified as a complex NP. I use 
the term ?factive? since there is a case where dependency across the koto-clause is 
possible with verbs such as kitaisuru ?hope? and nozomu ?wish?. Note further that the 
adjunct wh-phrase naze ?why? can establish a long distance dependency across the koto-
clause with those verbs. Argument wh-phrases can do the same. 
(i) John-wa [dare-ga yatekuru]-koto-o nozonda-no-desu-ka? 
(i) John-wa [Mary-ga naze yatekuru]-koto-o nozonda-no-desu-ka? 
 59 
The koto-complementizer is also found with verbs such as odoroku ?surprise? or akireru 
?amazed.? The dependency of the wh-phrase nante is blocked across those predicates. 
This observation is also caried over to comparable interogative wh-clauses, and the wh-
phrase how many cannot establish a dependency across the same environment. 
 
Exclamatives across koto 
(35) ? John-wa [Mary-ga nante kantan-na mondai-o toita]-koto-ni 
  J-TOP  -NOM WH easy-be problem-ACC solved-FN-DAT 
  odoroita-noda-roo 
  surprise-FOC-MOD 
  ?What an easy problem John got surprised that Mary solved!? 
(36) ? John-wa [Mary-ga nante rippa-na okonai-o sita]-koto-ni 
  J-TOP  -NOM WH respectable-be behavior-ACC did-FN-DAT 
  akire-teiru-noda-roo 
  amaze-be-FOC-MOD 
  ?What a respectable behavior John got amazed that Mary did!? 
Interogatives acros koto 
(37) ? John-wa [Mary-ga ikutu-no pasokon-o mot-teiru]-koto-ni 
  J-TOP  -NOM how.many-GEN computer-ACC own-be-FN-DAT 
  odoroita-no-desu-ka? 
  surprise-FN-POLITE-Q 
  ?How many computers John got surprised that Mary owns?? 
(38) ? John-wa [Mary-ga keeba-ni ikura tukata]-koto-ni 
  J-TOP  -NOM horse.racing-DAT how.much spent-FN-DAT 
  akireta-no-desu-ka? 
  amazed-FN-POLITE-Q 
  ?How much money John got amazed that Mary spent for horse racing?? 
The following example ilustrates that the koto-clause with surprise is a factive clause. 
 
Factive test 
(39) John-wa [Mary-ga sono kuruma-o kata]-koto-ni odoroita 
 J-TOP  -NOM that car-ACC bought-FN-DAT surprised 
 ?John got surprised that Mary bought that car.? 
 - # demo, Mary-wa  sono  kuruma-o kate-nai-desu 
   but  -TOP  that  car-ACC  bought-NEG-COP.POLITE 
   ?? but, Mary did not buy that car.? 
The awkwardnes of the continuation shown above indicates that the speaker himself 
cannot negate the propositional content of the embedded koto-clause, since it would be a 
direct contradiction with what he said in the first sentence. Furthermore, in contrast with 
the koto-clause, the to-clause with verbs like surprise does not have the same factive 
property. The speaker can negate the propositional content of the embedded clause. 
 
 60 
Factive test 
(40) John-wa [Mary-ga sono kuruma-o kata]-to odoroita 
 J-TOP  -NOM that car-ACC bought-C surprised 
 ?John got surprised that Mary bought that car.? 
 -  demo, Mary-wa sono kuruma-o kate-nai-desu 
   but  -TOP  that car-ACC bought-NEG-COP.POLITE 
   ?? but, Mary did not buy that car.? 
 
 
2.3.2 Relative Clause Island 
Relative clauses do not constitute an island for the dependency of exclamative wh-
phrases. The following examples are aceptable. 
 
(41) John-wa [Mary-ga nante taisetuni siteita piano-o kowasite 
 J-TOP  -NOM nante important did piano-ACC broke 
 simata no-da-roo 
 did FIN-FOC-MOD 
 ?John broke the piano [that Mary consider (it) how very precious] !? 
(42) John-wa [Mary-ga nante taisetuna yubiwa-o ireteita hako-o 
 J-TOP  -NOM nante important ring-ACC contain box-ACC 
 nakusite simata no-da-roo 
 lost did FIN-FOC-MOD 
 ?John lost the box that contains the ring [that Mary consider (it) how very  
 precious] !? 
(43) Sono sense-wa [gakuse-ga nante isyookenme kaita] 
 that teacher-TOP  student-NOM nante hard wrote 
 ronbun-o nakusite simata no-da-roo 
 paper-ACC lost did FIN-FOC-MOD 
 ?The teacher had lost the paper [the student wrote how very hard]!? 
Those examples are aceptable, discounting the fact that they are rather complex. In other 
words, they are much beter than those containing an adjunct wh-phrase naze ?why? 
inside the relative clause. As has been widely observed, argument wh-phrases inside the 
relative clause can take a matrix scope in wh-interogatives in Japanese (se Nishigauchi 
1986, Watanabe 1992, among many others). 
 
 
2.3.3 Wh-Island 
Another case in which the dependency betwen the exclamative wh-phrase and the 
licensors is blocked is the case of wh-islands (ref. Chomsky 1973). In the literature (e.g., 
Nishigauchi 1986, 1990, Watanabe 1992a, 1992b, Takahashi 1993), wh-island efects 
have been reported when the dependency crosses clauses marked with the Q-particle, 
though judgments vary. For example, Nishigauchi (1999:271-2) reports that (44) and (45) 
 61 
can only be interpreted as yes-no questions; the wh-phrase dare ?who? or nani ?what? in 
the embedded clause cannot take a matrix scope.
34
 
 
Nishigauchi (1999) 
(44) John-wa [Mary-ga dare-o turetekuru]-ka i-masi-ta-ka 
 J-TOP  -NOM who-ACC bring-Q say-POLITE-PAST-Q 
 ?Did John say who ary wil bring?? 
 ?*Who did John say whether Mary wil bring?? 
(45) Bil-wa [John-ga dare-ni nani-o ageta]-ka oboete-imasu-ka 
 B-TOP  J-NOM who-DAT what-ACC gave-Q remember-POLITE-Q 
 ?Does Bil remember what John gave to whom?? 
 ?*What does Bil remember to whom John gave?? 
 ?*Whom does Bil remember what John gave to?? 
On the other hand, Watanabe (1992a) observed that in example (46), which is almost the 
same as (45), it is marginaly aceptable for the higher wh-phrase to take the matrix scope, 
while it is impossible for the lower one to do so. 
 
Watanabe (1992a) 
(46) Kimi-wa [dare-ga dono hon-o tosyokan-kara karidasita]-ka 
 you-TOP  who-NOM which book-ACC library-from checked.out-Q 
 siritai-no 
 wonder-Q 
 ??For which person x, you want to know which book x checked out from the 
 library?? 
 ?*for which book y, you want to know who checked out y from the library?? 
Furthermore, Takahashi (1993) provides the folowing observation.
35
 
 
Takahashi (1993) 
(47) Taro-wa [Hanako-ga nani-o tabeta]-ka siritagateiru-no? 
 T-TOP  H-NOM what-ACC ate-Q wonder-Q 
 ?Does Taro wonder what Hanako ate?? 
 ?What does Taro wonder whether Hanako ate?? 
One observation by Takahashi (1993), to which we pay special atention, is that when the 
wh-phrase in the embedded clause undergoes scrambling, the ambiguity disappears, and 
the resulting example can only be interpreted as the matrix wh-question. 
 
                                                
34
 Also, it has been observed by various researchers that wh-island efects are 
considerably weaker if the embedded interogative complementizer is not ka, but ka-doo-
ka. 
35
 Takahashi?s (1993) ?lack of wh-island efect? has been extensively discused recently 
by a series of work by Kitagawa (Deguchi and Kitagawa 2002, Kitagawa 2005), Ishihara 
(2002?). Those authors claim that prosodic contour determines the ambiguity observed by 
Takahashi. It is impossible to review the details of their quite innovative approach here, 
unfortunately. 
 62 
Long distance scrambling 
(48) Nani-o Taro-wa [Hanako-ga t tabeta]-ka siritagateiru-no? 
 what-ACC T-TOP  H-NOM  ate-Q wonder-Q 
 ?What does Taro wonder whether Hanako ate?? 
With this background, the following examples show that the dependency of wh-
exclamatives across a clause with the Q-particle is not good. 
 
Wh-island efects in exclamatives 
(49) ? John-wa [Mary-ga nante takusan-no hito-ni paati-de at-ta 
  J-TOP  -NOM WH many-GEN man-DAT party-at met-PST 
  ka] siritagat-teiru-no-da-roo 
  whether wonder-be-FIN-FOC-MOD 
  ?What a lot of people John wonders whether Mary met at the party!? 
(50) ? John-wa [sono sense-ga nante takusan-no gakuse-ni C-o  
  J-TOP  that teacher-NOM WH many-GEN student-DAT C-ACC 
  ageta]-ka siritagateiru-no-da-roo 
  gave-Q wonder-be-FOC-MOD 
  ?What a lot of students John wonders whether Mary gave a C to!? 
On the other hand, when the exclamative wh-phrase undergoes long distance scrambling 
outside of the embedded clause, examples become aceptable. 
 
Long distance scrambling 
(51) Nante takusan-no hito-ni
2
 John-wa [Mary-ga t
2
 paati-de at-ta 
 nante many-GEN man-DAT J-TOP  -NOM  party-at met-PST 
 ka] siritagateiru no-da-roo 
 whether wonder  FIN-FOC-MOD 
 `What a lot of people John wonders whether Mary met at the party!' 
(52) Nante takusan-no gakuse-ni
2
 John-wa [sono sense-ga t
2
 C-o 
 WH many-GEN student-DAT J-TOP  that teacher-NOM  C-ACC 
 ageta-ka siritagateiru-no-da-roo 
 gave-Q wonder-be-FOC-MOD 
  ?What a lot of students John wonders whether Mary gave a C to!? 
 
 
3. Interaction with Interogatives and Feature Checking 
In this section, we propose a licensing mechanism of exclamative wh-phrases, under 
current minimalist asumptions (Chomsky 1995, 1999, 2000). We have already sen that 
the Mood head and Focus head are playing a particularly important role in licensing 
exclamative wh-phrases, and we claim that exclamative wh-phrases bear at least two 
features: [focus] and [mood]. On the other hand, we claim that interogative wh-phrases 
bear a slightly diferent set of features, as summarized below. 
 
 63 
Feature sets 
(53) a. exclamative wh-phrase: [focus] [mood] 
 b. interogative wh-phrase: [focus] [wh] 
Following den Dikken (2003), we asume that those wh-phrases bear a [focus] feature, 
which has to be checked against a focus head. In addition, as we have claimed, 
exclamative wh-phrases bear a mood feature, while interogative wh-phrases do not. We 
do not know any empirical data suggesting that interogative wh-phrases interact with the 
mood head as exclamative wh-phrases do. Interogative wh-phrases, instead, bear a wh-
feature, which must be checked against the Question head. In Japanese, it is realized as 
the particle ka. 
Furthermore, I argue that FocusP is always involved in licensing interogative wh-
phrases, although the head of FocusP is not always overtly realized in wh-interogatives. 
For example, interogative wh-phrases are licensed in the following way: 
 
Wh-interogatives 
(54) a. John-wa nani-o kai-masi-ta-ka 
  J-TOP what-ACC buy-POLITE-PAST-Q 
 b. John-wa nani-o kat-ta-no-desu-ka 
  J-TOP what-ACC buy-PAST-FIN-FOC-Q 
  ?What did John buy?? 
c.  QuestionP     
  ru     
        
  FocusP Question  ka [wh]  
  ru      
        
  FiniteP Focus da [focus] 
  ru       
         
   IP Finite no    
  6       
  John what bought       
   [focus]      
   [wh]      
 
At this point, it suffices to say that the wh-phrase what undergoes covert movement 
(either a phrase or a feature) into [Spec,FocP], where the [focus] feature can be checked, 
and subsequently, it moves to [Spec,QuestionP] in order to check the [wh] feature. 
Licensing wh-exclamatives works in a similar way. The exclamative wh-phrase 
undergoes covert movement into [Spec,FocP], where the [focus] feature is checked, and 
further moves to [Spec,MoodP] to check the [mood] feature. 
 
 64 
Wh-exclamatives 
(55) a. John-wa nante ookina kuruma-o kata-no-da-roo 
  J-TOP NANTE big car-ACC bought-FIN-FOC-MOD 
  ?What a big car John bought!? 
b.  MoodP    
  u     
   r     
  FocusP Mood  roo [mood] 
  u      
   r     
  FiniteP Focus da [focus] 
  u       
   r      
   IP Finite no    
  6       
  John what a big car bought     
   [focus]      
   [mood]      
 
In section 2.3.3, we have sen that the dependency betwen the exclamative wh-
phrase and its licensor is blocked by an intervening wh-clause. Furthermore, when the 
exclamative wh-phrase undergoes long distance scrambling, the example is much beter. 
 
Interogative wh-clause island 
(56) a. ? John-wa [Mary-ga nante takusan-no gakuse-ni dono  
   J-TOP  -NOM NANTE many-GEN student-DAT which  
   hon-o ageta]-ka siritagateiru-no-da-roo 
   book-ACC gave-Q wonder-FIN-FOC-MOD 
 b.  Nante takusan-no gakuse-ni John-wa Mary-ga dono  
   NANTE many-GEN student-DAT J-TOP  -NOM which  
   hon-o ageta]-ka siritagateiru-no-da-roo 
   book-ACC gave-Q wonder-FIN-FOC-MOD 
   ?What a lot of students John wonders which book Mary gave!? 
A similar contrast is observed when the dependency betwen the interogative wh-phrase 
and its licensor crosses a wh-exclamative clause. 
 
Exclamative wh-clause island 
(57) a. ? John-wa [Mary-ga dono gakuse-ni nante hidoi seseki-o 
   J-TOP   -NOM which student-DAT NANTE bad grade-ACC 
   tuketa-no-da-roo]-to   omotteiru-no 
   gave-FIN-FOC-MOD-C  think-Q 
 b.  Dono gakuse-ni  John-wa [Mary-ga nante hidoi seseki-o 
   which student-DAT  J-TOP   -NOM NANTE bad grade-ACC 
   tuketa-no-da-roo]-to omotteiru-no 
 65 
   gave-FIN-FOC-MOD-C think-Q 
   ?Which student does John think, what a bad grade Mary gave to?? 
Finaly, it should be noted that the (a) examples in (56) and (57) are quite degraded, but if 
the order betwen the two wh-phrases is reversed, the examples are worse. 
 
(58) a. * John-wa [Mary-ga dono hon-o  nante takusan-no gakuse-ni 
   J-TOP  -NOM which book-ACC wh many-GEN student-DAT 
   ageta]-ka siritagateiru-no-da-roo 
   gave-Q  wonder-FIN-FOC-MOD 
   ?How very many people John wonders which book Mary gave!? 
 b. * John-wa [Mary-ga nante hidoi seseki-o dono gakuse-ni  
   J-TOP  -NOM NANTE bad grade-ACC which student-DAT  
   tuketa-no-da-roo]-to omotteiru-no 
   gave-FIN-FOC-MOD-C think-Q 
   ?Which student did John think what a bad grade Mary gave?? 
The paralelism betwen (56) and (57) cals for a uniform acount. What is common 
betwen the interogative wh-clause and exclamative wh-clause is the presence of Focus 
projection and either the Question Phrase or the Mood Phrase, acording to our 
hypothesized functional structure. In (56), the in-situ interogative wh-phrase undergoes 
covert movement to [Spec,QuestionP], and this is the position where the interogative 
wh-phrase takes scope. On the other hand, the scope position for the exclamative wh-
phrase is situated at the matrix clause. It is conceivable that in (56a), the presence of the 
interogative wh-phrase in [Spec,QuestionP] creates a quantifier induced barier for the 
exclamative wh-phrase; therefore, the example is not aceptable, as ilustrated in (59). On 
the other hand, when the exclamative wh-phrase undergoes overt movement, via 
scrambling, to a position high enough to cross the Question Phrase, there is no problem 
for the exclamative wh-phrase to establish the legitimate dependency with its licensor at 
the matrix clause. That acounts for the contrast in (56). 
 
 66 
(59) Quantifier induced barier      
    FocP      
    u     
     r     
    - - -  Focus da   
    VP      
    ru     
    QuesP V     
         
   Wh-Int
3
 ru     
    FocP Question ka   
         
   t
3
 ru     
    - - - Focus    
    IP    
   6    
    VP    
        
   Wh-Exc
2
 t
3
 V   
        
 
 
It is possible to employ a similar story for the contrast in (57). The position that the in-
situ exclamative wh-phrase has to take scope in is the embedded Mood Phrase. Then, it 
undergoes covert movement to [Spec,MoodP]. Now, the in-situ interogative wh-phrase 
has to take a matrix scope, since that is the only available position for it to be licensed. In 
(57a), the presence of the exclamative wh-phrase in [Spec,MoodP] creates a quantifier 
induced barier for the interogative wh-phrase. Therefore, the interogative wh-phrase 
cannot be licensed properly, and the example is not good. When it undergoes scrambling 
to the position above Mood Phrase, there is no intervening element betwen the 
scrambled position and the matrix licensor. 
 
 67 
(60) Quantifier induced barier      
    FocP      
    u     
     r     
    - - -  Focus da   
    VP      
    ru     
    MoodP V     
         
   Wh-Exc
3
 ru     
    FocP Mood roo    
         
   t
3
 ru     
    - - - Focus    
    IP    
   6    
    VP    
        
   Wh-Int
2
 t
3
 V   
        
 
 
Recal that the distinction betwen the overt / covert dichotomy is the crucial ingredient 
for the original observation by Beck (1996). Bariers created by a scope-bearing element 
are only applicable for covert movement, and the restriction is lifted for overt movement. 
One could asume that the overt movement (scrambling, in this case) is driven by a 
diferent feature than the feature that is relevant for creating a barier. Once we take the 
contrast in (56) and (57) seriously, it is quite conceivable that we are dealing with the 
intervention efect (se Takahashi 1990, Pesetsky 2000, among many others). 
There is one paradigm in which the examples are completely unaceptable 
regardles of the order of the interogative and exclamative wh-phrases. 
 
(61) a. * Mary-wa dono hito-ni nante ookii keeki-o 
   -TOP which person-DAT NANTE big cake-AC 
   tabe-sase-ta-no-da-roo-ka 
   eat-CAUSE-PAST-FIN-FOC-MOD-Q 
   ?What a big cake which person Mary make eat!?? 
 b. * Mary-wa nante ookii keeki-o dono hito-ni  
   -TOP NANTE big cake-AC which person-DAT 
   tabe-sase-ta-no-da-roo-ka 
   eat-CAUSE-PAST-FIN-FOC-MOD-Q 
Acording to our acount so far, there sems to be no problem with respect to how 
features are checked in both wh-phrases. Let us asume that one Focus head can check 
[focus] feature on both wh-phrases; since we have been asuming covert phrasal 
 68 
movement, let us say that there is more than one specifier. This is necesary, and it is 
independent from this particular paradigm since we need to be equipped with a 
mechanism that alows multiple wh-interogatives and multiple wh-exclamatives. Now, 
in (61a), the interogative wh-phrase undergoes covert movement to [Spec,QuestionP] 
(through [Spec,FocusP]). This satisfies al the requirements of the interogative wh-
phrase, both [focus] and [wh] features being checked. It is then also possible to raise the 
exclamative wh-phrase to [Spec,MoodP] (again through [Spec,FocusP]). These steps of 
movement correctly check al the features ([focus] and [mood] features) on the 
exclamative wh-phrase. Therefore, the derivation should converge; but the examples are 
not aceptable. 
We argue that the problems with examples in (61) are not so much to do with the 
feature checking, but the requirement of clause-typing. Let us make the following 
minimal asumptions about clause-typing: 
 
(62) a. A clause is typed as ?interogative? if an interogative wh-phrase takes 
  scope in that clause. 
 b. A clause is typed as ?exclamative? if an exclamative wh-phrase takes 
  scope in that clause. 
This gives us a straightforward answer to the unaceptability of examples in (61). In 
those examples, both interogative wh-phrases and exclamative wh-phrases take scope in 
the same clause. Following Chomsky?s (1995:293) claim that a clause has to be 
unambiguously typed, we can acount for the examples in (61) by saying that those 
examples are ambiguous betwen interogative and exclamative clauses. 
 
 
4. Exclamative wh-phrase: nante 
Just as English exclamatives come with how and what forms, examples in (63)-(67) 
ilustrate five sub-types of exclamatory sentences in Japanese. In each sub-type a 
diferent kind (structuraly or categoricaly) of phrase is being exclaimed. 
 
Various examples of exclamatives 
(63) Kanozyo-wa [nante utukusi zyosei]-na no da roo 
 she-TOP  wh beautiful lady-COP EXC 
 ?What a beautiful lady she is!? Adjective within NP 
(64) Kono hana-wa [nante utukusi] no da roo 
 this flower-TOP  wh beautiful EXC 
 ?How beautiful this flower is!? Predicate Adjective 
(65) Kono hana-wa [nante kirei]-na no da roo 
 this flower-TOP  wh beautiful-COP EXC 
 ?How beautiful this flower is!? Predicate Adj. Noun 
(66) Kanozyo-wa [nante kakkoii kuruma]-ni notteiru no da roo 
 she-TOP  wh nice car-DAT drives EXC 
 ?What a nice car she drives!? Adjective within NP 
 69 
(67) Kanozyo-wa [nante hayaku] hasiru no da roo 
 she-TOP  wh fast runs EXC 
 ?How fast she runs!? Adverb 
In this section, diferences observed among the five types ilustrated in (63)-(67) are 
mostly set aside; we concentrate on the general licensing conditions of exclamatory 
sentences in Japanese. 
 
 
4.1 Variations of the exclamative wh-phrase 
Exclamatory sentences in Japanese use the wh-phrase nante. Properties of the 
exclamative wh-phrase are investigated in this section. After ilustrating several 
properties of the exclamative wh-phrase in Japanese, we propose an analysis of the 
structure of DP, which straightforwardly acounts for the mentioned properties of 
Japanese exclamatory sentences. 
Regarding the wh-phrase in exclamatives, there is an apparent optionality 
regarding the choice of wh-phrase in exclamatives. 
 
Variation in the exclamative wh-phrase 
(68) John-wa [nanto-yu / nante / nanto ookina kuruma]-o kata no da roo 
 J-TOP  wh big car-ACC bought EXC 
 ?What a big car John bought!? 
Here, thre diferent forms of wh-phrases in exclamatives are listed. Al thre of them 
work properly in exclamatives; however, when examined carefully, certain diferences 
emerge, which lead us to conclude that there is a particular position in which each wh-
phrase must occur. Consider the following sets of examples: 
 
(69) a.  Kanozyo-wa [
DP
 nanto-yuu utukusi zyosei]-na no da roo 
   she-TOP  wh beautiful lady-COP  EXC 
   ?What a beautiful lady she is!? 
 b. * Kono hana-wa [
AP
 nanto-yuu utukusi] no da roo 
   this flower-TOP  wh  beautiful EXC 
   ?How beautiful this flower is!? 
 c. * Kanozyo-wa [
AdvP
 nanto-yuu hayaku] hasiru no da roo 
   she-TOP  wh fast runs EXC 
   ?How fast she runs!? 
(70) John-wa [
DP
 nanto-yuu otoko]-na no da roo 
 J-TOP  wh man-COP EXC 
 ?What a man John is!? 
Note that nante can occur in al cases as we have sen previously, but the data above 
ilustrate that the distribution of nanto-yuu is certainly restricted. While nanto-yuu can 
occur with an adjective within a DP in (69a), nanto-yuu cannot occur with a predicative 
adjective (69b) or an adverb (69c). One obvious characteristic feature in the observed 
contrast is the existence of a DP. That leads us to suggest that nanto-yuu must be atached 
 70 
to a nominal projection (or an extended projection of NP; se Grimshaw 1997. cf. Abney 
1987). In fact, nanto-yuu can appear without any adjective within a DP, as shown in (70). 
The distribution of nanto, on the other hand, is diferent from the distribution of 
nanto-yuu. First, nanto can occur where there is no nominal projection. 
 
(71) a. Kanozyo-wa [nanto utukusi zyosei]-na no da roo 
  she-TOP  wh beautiful lady-COP EXC 
  ?What a beautiful lady she is!? 
 b. Kono hana-wa [nanto utukusi] no da roo 
  this flower-TOP  wh beautiful EXC 
  ?How beautiful this flower is!? 
 c. Kanozyo-wa [nanto hayaku] hasiru no da roo 
  she-TOP  wh fast runs  EXC 
  ?How fast she runs!? 
(72) * John-wa [nanto otoko]-na no da roo 
  J-TOP  wh man-COP EXC 
  ?What a man John is!? 
It sems that the distribution of nanto and nante partialy overlaps. However, the example 
in (72) shows that nanto requires the presence of an adjective or adverb. When there is no 
adjective or adverb, examples are unaceptable, which suggests that nanto must be 
atached to some non-nominal projection, namely an adjective phrase or an adverb. The 
example in (72) is unaceptable since there is no projection to which nanto can atach. 
The above data is summarized in the following table, which suggests that the distribution 
of nanto-yuu and nanto is diferent. Furthermore, since nante can appear in the widest 
distribution, it sems to be that nante is somewhat ambiguous betwen the two forms. 
 
(73) Summary 
nanto-yuu nante nanto 
*ADVERB/ADJECTIVE ADVERB/ADJECTIVE ADVERB/ADJECTIVE 
NOUN NOUN *NOUN 
 
Based on the distribution of those wh-phrases, we argue that two wh-phrases are 
placed in a diferent position and that they are diferent categories. Specificaly, we 
propose the following structure in order to acount for the properties (se Abney 1987, 
Corver 1997a, Svenonius 1994). 
 
 71 
(74) Two wh-phrases in diferent positions 
      a.   DP       b.   DP 
 2   2 
nanto yuu D?  D NP 
  3 
 D NP DEGP NP 
 2 2 | 
 DEGP NP DEG AP N 
  | nanto ookina kuruma 
 DEG AP N big car 
 ookina kuruma 
 big car 
 
Given that the presence of nanto depends on the presence of an adjective or adverb, a 
natural characterization is that nanto is a head of a degre phrase, as in (74b), that selects 
an adjective or an adverb (Corver 1997a). 
 
 
5. Semantics 
5.1 Two degres must be ?related? 
Having reviewed the properties of exclamatives in the previous section, let us return to 
the contrast observed betwen English and Japanese, namely, the availability of multiple 
exclamatives. In the introduction, we saw that Japanese alows multiple exclamatives 
while English does not. In this section, we discuss properties of multiple exclamatives in 
Japanese. Examples are repeated from the introduction. 
 
Multiple wh-exclamatives in Japanese 
(75) a. [nante osanai kodomo]-ga [nante muzukasi mondai]-o  toita 
   wh young child-NOM   wh dificult problem-ACC solved 
  no da roo 
  EXC 
  ?What a young child solved what a dificult question!? 
 b. John-wa [nante ookina hambaagaa]-o [nante hayaku] tabeta 
  J-TOP   wh  big hamburger-ACC  wh  fast ate  
  no da roo 
  EXC 
  ?How fast John ate what a big hamburger!? 
An observation is that the sentence in (75b) sems to obey a certain pragmatic condition. 
Given the factive nature of exclamatives and general knowledge of the world, in order for 
(75b) to be felicitous, the speaker must have a belief that it takes time to eat a big 
hamburger. Furthermore, the informal meaning of (75b) is the following: ?Given the 
degre to which the hamburger is d-big, the degre to which the time John spent eating 
the hamburger is d?-short is extreme, and given the degre to which the time is d?-short, 
the degre to which the hamburger is d-big is extreme.? What should be noted here is that 
in multiple exclamatives, there at least there two degres exclaimed. The informal 
 72 
meaning of the multiple exclamative shows that multiple exclamatives satisfy the 
condition that two degres must be ?related.? Specificaly, the exclamation of Degre A 
is relative to Degre B, and the exclamation of Degre B is relative to Degre A. The 
following example suggests that when such a condition cannot be obtained, multiple 
exclamatives are unaceptable, even in Japanese. 
 
Degres are NOT related 
(76) * [nante kitanai ie]-ga [nante tookuni] mieru no da roo 
   wh dirty house-NOM  wh far can.se EXC 
  ?What a dirty house can be sen how far!? 
Related, but the ?direction? is not right 
(77) * [nante takai biru]-ga [nante tokaini] aru no da roo 
  wh tal building-NOM  wh urban exist EXC 
  ?What a tal building stands in what an urban area!? 
Related 
(78) [nante takai biru]-ga [nante inakani] aru no da roo 
  wh tal building-NOM  wh rural exist EXC 
 ?What a tal building stands in what a rural area!? 
Example (76) is very awkward. One exclamation is concerned with the ?dirty-nes? of 
the house, and the other is concerned with the distance from which one can se the house. 
The awkwardnes in (76) sems to stem from the fact that it is not easy to make an 
obvious connection betwen the ?dirty-nes? of the house and the distance from which 
one can se the house. In other words, it is not the case that the dirty house is easy to se 
from the distance or hard to se. It is just that the ?dirty-nes? of the house has nothing to 
do with the distance. In such a case, the condition mentioned above can never be satisfied, 
so the example is quite degraded. The example in (77) is again awkward but is 
unaceptable for a diferent reason. It is about exclamation with regard to the height of 
the building and how urban the town is. Apparently, it is not hard to relate the existence 
of tal buildings and the kind (urban or rural) of town under consideration. However, 
what is unnatural here is not what is related to each other, but how they are related to 
each other. Given (77), it sems that the speaker was quite surprised to se a tal building 
in an urban area, say New York City. Who would be surprised to se a tal building (or 
many, as a mater of fact) in Manhatan? In fact, having numerous tal buildings is one of 
the defining characteristics of an urban area. In general, people expect to se tal 
buildings in such an area; the example in (78) is good and natural since the speaker is 
surprised to se a tal building in a rural area. It is reasonable since people do not expect 
to se tal buildings in a rural area. Thus, the contrast betwen (77) and (78) suggests that 
how two degres are related to each other is another important condition to be met in 
multiple exclamatives.
36
 
Another piece of support for this ?relatednes? comes from the following 
observation. When the example in (78) is uttered, it sems that the height of the building 
                                                
36
 I have not yet come up with a formal acount for this observation. Chris Potts and 
Chris Barker suggested to me that Kennedy?s (1997) polar opposition analysis might be 
helpful. 
 73 
does not have to be an extreme degre. Surely it does not have to be in the top 10 of tal 
buildings in the world to create a sense of surprise, though it has to be sufficiently high. 
Consider a seven-story building; no one wil be surprised to se such a building in 
Manhatan, given that there are numerous buildings higher than that building. But, this 
perspective of ?talnes? changes completely in a smal town somewhere in the middle of 
nowhere. In other words, whether or not the height of the building is a worthy cause of 
surprise is relative to the surrounding environment. 
Another interesting case of multiple exclamatives is the following.
37
 
 
Multiple exclamatives with the same adjective 
(79) a. nante tisana akatyan-ga nante tisana huku-o kiteiru nodaroo 
  wh tiny baby-NOM wh tiny cloth-ACC wear EXC 
  ?What a tiny baby is wearing what a tiny cloth!? 
 b. nante sutekina zyoyuu-ga nante sutekina dorama-ni syutuensiteiru 
  wh neat actres-NOM wh neat drama-DAT play.a.role 
  nodaroo 
  EXC 
  ?What a neat actres is playing a role in what a neat drama!? 
In example (79a), the same adjective tisana ?tiny? is used in two diferent phrases. It 
sems that, in exclamatives, the speaker is simply surprised twice. Maybe a more 
appropriate translation of it would be: What a tiny baby is wearing a cloth and how tiny 
the cloth is! The point is that two occurrences of the same adjective are not ?related? in 
the sense discussed above. In other words, there is no strong connection or anything like 
the dependency that we have observed in cases where two diferent adjectives are used in 
multiple exclamatives. Example (79b) ilustrates the same point. One intuition behind the 
?relatednes? is that the degre to which the speaker is surprised depends on the other 
degre. It is not dificult to imagine what kind of interpretation we get if two degres are 
dependent on each other, as in (79b); suppose there are two actreses, A and B, and 
actres A is neater than actres B (say, the actres A is more popular than the actres B). 
Then what we would expect is that when (79b) is uttered, the speaker knows that the 
degre of neatnes of the drama in which the actres A is playing a role is greater than the 
degre of neatnes of the drama in which actres B is playing a role. However, it sems 
that such an interpretation is not available. Therefore, examples in (79) are another kind 
of multiple exclamatives, difering from the examples in which two diferent adjectives 
are used. 
Another set of examples arises from adjectives of quantity, and again the same 
adjectives are used. 
 
Multiple exclamatives with the same adjective 
(80) a. Nante takusan-no otokonoko-ga nante takusan-no onnanoko-to 
  wh many-GEN boy-NOM  wh many-GEN girl-with 
  tukiateiru nodaroo 
  date  EXC 
                                                
37
 I am grateful to Chizuru Nakao and Tomo Fuji (via Chris Tancredi) for providing me 
the examples with the same adjective. 
 74 
  ?How many boys date how many girls!? 
 b. Nante ooku-no hito-ga nante ooku-no okane-o  
  wh many-GEN man-NOM wh  many-GEN money-ACC  
  muda-ni suru nodaroo 
  waste-DAT do EXC 
  ?How many people waste how much money!? 
The example in (80a) can be uttered in a situation where the speaker finds that there are 
surprisingly many boys who date surprisingly many girls. The example in (80b) also 
shows the same point. Again, whether or not the degre of one adjective is surprisingly 
high is independent of the degre of the other adjective. 
 
 
5.2 Two wh-phrases in one DP 
So far, al multiple exclamatives we have sen include one exclamation in one argument 
or adjunct position. For instance, a direct object and an indirect object are exclaimed, or a 
direct object and an adverb are exclaimed. In this section, we wil observe another type of 
multiple exclamative in Japanese. Specificaly, we wil se an instance of multiple 
exclamatives in which there is more than one exclamation within one nominal projection. 
Recal in the previous sections, we discussed diferent wh-phrases in Japanese 
exclamatives. The conclusion was that diferent exclamative wh-phrases are located in 
diferent positions in the structure. What is proposed is represented in the folowing tre 
diagram. 
 
(81) Two wh-phrases in diferent positions 
      a.   DP       b.   DP 
 2   2 
nanto yuu D?  D NP 
  3 
 D NP DEGP NP 
 2 2 | 
 DEGP NP DEG AP N 
  | nanto ookina kuruma 
 DEG AP N big car 
 ookina kuruma 
 big car 
 
Since, in those structures, wh-phrases are in diferent positions, it would be interesting to 
se whether there is any observable semantic efect depending on the elements the wh-
phrase is asociated with. Here, it sems that when the exclamative wh-phrase is 
combined with an adjective or adverb (i.e., nanto), what is exclaimed is the degre of the 
property of the adjective. On the other hand, when the exclamative wh-phrase is 
combined with a nominal projection (i.e., nanto-yuu), what is exclaimed is the property 
of the NP. Thus, the prediction is that the meaning of (81a) is slightly diferent from that 
of (81b). 
Let us start from (81b). In this example, the exclamative wh-phrase nanto is 
combined with an adjective, and the meaning is that the degre of bignes of the car is 
 75 
extreme, hence surprising. We can easily think of a certain kind of big car, such as Chevy 
Suburban or Ford Excursion, a big car we typicaly se on the stret. On the other hand, 
what (81a) means is that a certain property of the big car is quite unexpected, hence 
surprising. A plausible situation is as follows. People usualy think that big cars are not 
economical at al. They consume a lot of gas; every time one fils up at a gas station, it 
costs a lot of money. But imagine that we have a high-tech car, which is big, but 
surprisingly it does not consume much gasoline. Then, one can use (81a) to show one's 
surprise at seing the unexpected property of a big car. The property itself is not specified 
in the example above, but to utter (81a), there must be at least one unexpected property, 
given that the car is large. 
In addition, the analysis proposed here makes a prediction that we should be able 
to use non-gradable adjectives with (81)a), but not with (81b). It should be so since in 
(81b), a wh-phrase is directly atached to an adjective, and the gradable degre of that 
adjective must be the locus of exclamation. On the other hand, in (81a), the locus of 
exclamation is some property asociated with the NP as a whole. As far as one can think 
of some unexpected property asociated with the NP, it is that property that wil be the 
locus of exclamation. The prediction is borne out; here we used a word saikitekina 
'recursive' which is a non-gradable adjective. 
 
(82) Exclamatives with a non-gradable adjective 
 a. * Sore-wa nanto saikitekina ruuru-na no da roo 
   it-TOP WH recursive rule-COP EXC 
 b.  Sore-wa nanto-yuu saikitekina ruuru-na no da roo 
   it-TOP WH  recursive rule-COP EXC 
   'What a recursive rule it is!' 
When a non-gradable adjective is used with nanto, the example is not aceptable while 
nanto yuu has no problem being combined with such an adjective. This supports our 
analysis that the two wh-phrases are located in diferent structural positions. 
There is another consequence of our analysis. So far, we have looked at multiple 
exclamatives that involve more than one phrase in the VP. The following example shows 
that one DP can have more than one exclamative wh-phrase. 
 
(83) a. John-wa [
DP
 nanto-yuu [
DegP
 nanto tisana] sensyu]-na no da roo 
  J-TOP  WH  WH smal player-COP EXC 
  'What a how smal player John is!' 
 b.   DP 
   ru 
 nanto yuu   D' 
       
     D   NP 
      ru 
     DEGP    NP 
       | 
    DEG    AP   N 
    nanto   tisana sensyu 
 76 
     wh    smal player 
 
It is quite surprising that two exclamative wh-phrases can appear in the same noun phrase. 
In fact this is another strong argument for the claim that the exclamative wh-phrases 
nanto-yuu and nanto are placed in diferent positions in the structure. The following 
example also shows that nanto is asociated with an adjective phrase and nanto-yuu is 
asociated with a noun phrase. 
 
(84) * [
DP
 nanto [
DegP
 nanto-yuu tisana] sensyu] 
   WH  WH smal player 
  'What a how smal player' 
When two wh-phrases are used within the same noun phrase, nanto-yuu must precede 
nanto. That is another evidence for the proposed structure. 
Now, let us investigate the meaning of (83). First, there is an exclamation 
regarding the smalnes of the player. The speaker is surprised by the player?s smalnes. 
Second, the presence of nanto-yuu creates an exclamation regarding a property of the 
smal player. Imagine a smal basketbal player who performs extremely wel in games. 
The speaker is surprised by some fantastic performance of the player despite his 
smalnes as a basketbal player. Recal that there is a condition in multiple exclamatives 
in Japanese that the two degres in multiple exclamatives must be related in a natural way. 
Then, in (83), John is surprised by the extreme degre of smalnes of the player given his 
unbelievable performance, and John is also surprised by the extreme degre of the 
player's unbelievable performance given his extreme smalnes. 
 
 
5.3 The denotation of multiple exclamatives 
Having sen some of the properties of Japanese exclamatives, it is unclear how to 
represent the meaning of multiple exclamatives. Zanuttini & Portner (2003) propose a 
semantic analysis for exclamatives, and in their analysis, the presence of the wh-phrase 
and the abstract factive morpheme are the crucial ingredients. The presence of the wh-
phrase is rather obvious, and the existence of the factive morpheme, acording to them, is 
motivated by the fact that exclamatives, if embedded, are incompatible with non-factive 
predicates. 
Another insight by Zanuttini & Portner is that the presence of a wh-phrase in 
exclamatives makes the semantics of exclamatives somewhat similar to the semantics of 
wh-interogatives. Let us asume a standard claim that a wh-interogative denotes a set of 
propositions.
38
 For instance, the denotation of a wh-interogatives (85a) can be 
represented as in (85b). 
 
(85) a. How tal is John? 
 b. {?, John is 5?10? tal, John is 5?11? tal } 
 c. John is 5?11? tal. 
                                                
38
 Zanuttini & Portner (2003) say that diferences betwen approaches claimed by 
Hamblin (1973) and Kartunen (1977) do not mater for their analysis. I wil remain 
neutral on this isue. 
 77 
The pragmatic property of questions seks an answer from someone who participates in 
the conversation. In general, a person who is asked a question such as (85a) specifies one 
proposition from the set presented to him/her. A proposition (85c) is one appropriate 
answer to the question (85a), since the person correctly points out one member of the 
given set denoted by the wh-interogatives. 
Now, Zanuttini & Portner (2003) claim that the pragmatic property of 
exclamatives is diferent from interogatives; exclamatives can be characterized by 
?widening? of the set of propositions denoted by the wh-phrase that determines the 
domain of quantification. 
 
(86) a. How tal John is! 
 b. {?, John is 5?11? tal, John is 6?0? tal } 
 c. {?, John is 5?11? tal, John is 6?0? tal, John is 6?4? tall} 
As in wh-interogatives, the presence of a wh-phrase is a key to generate a set of 
propositions. By uttering an exclamative (86a), the speaker first presents a set of 
propositions which is ilustrated in (86b). This particular set of propositions represents an 
expectation in the context however relevant in the particular discourse; it is refered to as 
?an expected domain?. One obvious diference betwen exclamatives and interogatives 
is that exclamatives do not sek any answer. Then, the pragmatic property of 
exclamatives cannot be the one that someone (usualy not the speaker himself) has to 
point out some proposition from the given set of propositions. The proposal by Zanuttini 
& Porter (2003) is that the speaker, based on the expected domain, widens the domain of 
quantification by adding a new proposition that can crucialy not be found in the expected 
domain.
39
 In (86c), a new proposition John is 6?4? tall is added to the set of propositions. 
They cal this pragmatic property (or ?force?) of exclamatives ?widening?. 
Returning to multiple exclamatives like (87), suppose a situation where you know 
that John ate a 2 pound hamburger in 5 minutes, and it is quite fast that John can eat such 
a big hamburger in such a short period of time in this context, thus it is worth being 
surprised at. 
 
(87) John-wa [nante ookina hambaagaa]-o [nante hayaku] tabeta no da roo 
 J-TOP  wh big hamburger-ACC  wh fast ate EXC 
 ?How fast John ate what a big hamburger!? 
The first thing we do here is to extend Zanuttini & Portner?s (2003) semantic/pragmatic 
analysis slightly. This slight extension is necesary since they do not discuss properties of 
multiple exclamatives, let alone their meaning. Given the context above, I propose that 
the expected domain for (87) be something like (8a). 
 
(88) a. Expected domain 
  {?2 lb, 15 min?, ?3 lb, 17 min?, ?1 lb, 7 min?, ?1/2 lb, 2 min?} 
                                                
39
 Also, there sems to be a restriction with respect to the ?direction of degre?. When the 
talnes of a person is at isue, the value of what is to be added has to be greater than 
others. In the above case, the value 6?4? is larger than 6?0?, then the widening is 
appropriate. Se Kennedy (2000) for some related isues. 
 78 
 b. Widened domain 
  {?2 lb, 15 min?, ?3 lb, 17 min?, ?1 lb, 7 min?, ?1/2 lb, 2 min? ?2 lb, 5 min?}  
What has been changed from the expected domain for the simple exclamative is that now 
the members in the expected domain are pairs of the weight (the size) of the hamburger 
and the time spent to eat the hamburger. For example, eating a 2 pound hamburger in 15 
minutes is normal, then a pair ?2 lb, 15 min? is a member in this set. Other members in 
the set are similar. I follow Zanuttini & Portner (2003) that force of exclamatives is 
widening, then in order to widen the original domain of quantification, it is necesary to 
add a new member to the set, namely a pair of the size of the hamburger and the time 
spent to eat the hamburger. Since it has been known here that John ate a 2 pound 
hamburger in 5 minutes, a new pair ?2 lb, 5 min? is added. Notice, again, that in order for 
this pair to be succesfully added to the original set, it is necesary that the pair should 
not be in the original set.
40
 
One consequence of this representation for the multiple exclamatives is that 
multiple-exclamatives do not arise as a simple addition of two independent simple 
exclamatives. The multiple exclamative (87) cannot be a simple exclamation just on the 
size of the hamburger John ate, since there is no expected set that consists of values of the 
weight of a hamburger. In a similar way, (87) cannot be a simple exclamation just on the 
time spent eating the hamburger, since there is no expected set that consists of values of 
the time spent. What is claimed here is that (87) exclaims the rarity of the combination 
betwen the two degres. This analysis acounts for one property of multiple 
exclamatives: the two degres involved in multiple exclamatives must be related to each 
other. 
 
 
6. Sumary 
In this chapter, we made clear where exclamatives occur in the embedded context. We 
have shown that exclamatives are found in some limited environments. It was shown that 
the distribution of exclamatives in Japanese is slightly diferent than that in English. It is 
not true that exclamatives in Japanese occur only with factive predicates. We showed that 
exclamatives occur in some non-factive contexts; in addition to that, we found that not al 
factive predicates can select Japanese exclamatives. 
With respect to the island efects found in the dependency of exclamatives in 
Japanese, we concluded that those efects can be clasified as intervention efects. We 
proposed the specific mechanism that deals with those efects. 
We also discussed various properties of the exclamative wh-phrase nante and its 
morpho-syntactic variants. Finaly, some semantic isues of multiple exclamatives were 
discussed. We provided some new observations and it was shown that the meaning of 
                                                
40
 Given the expected domain as above, there are at least two ways to find a member that 
is not normal. One, as I have shown, is to add a pair of a large hamburger and a short 
period of time, and the other is to add a pair of a smal hamburger and a long period of 
time. It is clear that (87) cannot be used in a situation where it took a very long time for 
John to eat a very smal hamburger. I suppose this restriction comes from the fact that an 
adjective ooki ?big? and an adverb hayaku ?fast? have been used. 
 79 
multiple exclamatives in Japanese can be succesfully dealt with by extending the 
analysis by Zanuttini & Portner (2003). 
 
 
 80 
 
CHAPTER 4: EXCLAMATIVES AND SLUICING 
 
1. Introduction 
Merchant?s (2001) very detailed work on sluicing recently (re-)initiated much discussion 
on a range of isues. For example, it has been debated what the nature of ?identity 
conditions? are; is it syntactic, semantic, both, or something else? Also, the investigation 
of sluicing raises many questions about the overal architecture of gramar. Elipsis 
phenomena are observed clearly as PF realization (we do not hear parts of sentences 
which we expect to hear), but what can be (or sometimes ?must be?) mising and when it 
is possible does not sem to be regulated just by rules of PF. How we alow diferent 
components to talk to each other is quite a intricate puzzle. An example of sluicing is 
shown below; sluicing is a phenomenon in which a wh-phrase alone appears where the 
whole wh-clause is expected to occur, as originaly observed by Ross (1969). In (1a), the 
wh-phrase who appears where we expect something like the wh-clause in (1b) to occur. 
Note that the interpretation of who is exactly the same as who in (1b); who is interpreted 
as an object of talking to. 
 
Sluicing 
(1) a. John was talking to someone, but I don?t know who. 
 b. I don?t know who John was talking to. 
In this chapter we discuss how exclamatives behave under sluicing. As far as we know, 
al the work on sluicing has been done with interogatives. One might think that showing 
that sluicing is available with exclamatives is good, but not interesting, since 
exclamatives are ?similar to? interogatives: after al, both involve wh- movement, and it 
is relatively easy to show that both obey the same island constraints. We disagre with 
the claim that exclamative sluicing is not interesting. First, it has been noted that the 
existence of wh-movement does not guarante the availability of sluicing; for example, 
relative clauses do not alow sluicing despite the fact that they involve wh-movement 
(Merchant 2001:58, Lobeck 1995:57). 
 
Relative clauses 
(2) a. * Someone stole the car, but they couldn?t find the person who. 
 b. * The judge gave five years each to the adults who participated in the riot, 
   but she hasn?t yet sentences the minors who. 
 c. * Although the place where is unclear, the time when the meting is to be 
   held is posted on the door. 
The same is true for clefts (Merchant 2001:59). 
 
Clefts 
(3) a. * We thought it was Abby who stole the car, but it was Ben who. 
 b. * Somebody stole the car, but no one knew that it was Ben who. 
Second, we contend that the fact that sluicing is available with exclamatives bears a much 
deeper theoretical significance. In fact, exclamative sluicing is an important argument for 
a thesis put forward by Pesetsky & Torego (2001) that wh-interogative clauses and wh-
 81 
exclamative clauses are syntacticaly the same (se further development of their idea by 
Fuji & Ono (2005)). These clauses are just ?wh-clauses? as far as their syntax is 
concerned, and they are interpreted diferently, depending on the structures they are in. 
The thesis is in a sense surprising, since wh-exclamatives and wh-interogatives look 
quite diferent (e.g., lack of subject-auxiliary inversion in exclamatives, diferent 
predicates select those wh-clauses, etc.). Pesetsky & Torego (2001), also Fuji & Ono 
(2005), argue, nonetheles, that diferences betwen interogatives and exclamatives can 
be acounted for on independent grounds. To the extent that their analyses are on the 
right track, ilustrating that sluicing is available with exclamatives offers further support 
for the claim that the two ?wh-clauses? are syntacticaly the same. If the analyses are 
right, it is in a sense predicted that sluicing is available with exclamatives, given that it is 
available with interogatives, and that ?sluicing? in those wh-clauses has to be the same 
thing. We hope the above discussion wil justify why it is very important for us to show 
as much paralelism as possible betwen interogatives and exclamatives. 
Finaly, let us mention one property that makes sluicing extremely interesting: 
sluicing sems to repair some ?island violations.? The following example ilustrates that 
when elipsis does not occur, the example is bad, but the example is aceptable with 
sluicing. 
 
Island repair 
(4) They want to hire someone who speaks a Balkan language, but I don?t 
 remember which (Balkan language). 
 (cf. *I don?t remember which (Balkan language) they want to hire someone 
 who speaks.) 
Later, we wil show that such ?island amelioration? is also observed with examples of 
exclamative sluicing. Again, why do we have to observe various similarities betwen 
interogative and exclamative wh-clauses (as far as such a property is testable in both)? 
We now know the answer: they are syntacticaly the same. 
 
 
2. Some Basic Properties of Sluicing  
2.1 CP status 
Merchant (2001:40?54) provides several arguments in order to show that the sluice (i.e., 
the leftover of sluicing) is actualy a CP, not a DP. This is important for him since 
Riemsdijk (1978), for example, argues that the wh-phrases in sluicing are in fact 
complement of the predicates. Then, acording to those who advocate such a ?wh-phrase 
as a fragment? approach, the structure of the sluice is the folowing. 
 
(5) .. I don?t [
VP
 know [
DP
 who ] . 
We cannot repeat here al the cases provided in Merchant (2001), but we review a few of 
them that are relevant for sluicing in exclamatives.
41
 To the extent that the data in 
                                                
41
 It might not be appropriate to cal the phenomenon ?exclamative sluicing,? since what 
we would like to do here is to show that it is sluicing. Nonetheles, we keep using the 
term for the ease of exposition. 
 82 
exclamative sluicing are paralel with that in interogative sluicing, we can make sure that 
the ?sluicing in exclamatives? discussed in this chapter is realy the same kind of 
phenomenon observed in interogative wh-clauses. 
The first argument, originaly put forward by Ross (1969), is that the sluice is 
available in the position where DPs are banned. Ross (1969) points out that a DP (an NP 
at that time) cannot be extraposed when the adjectival predicates, e.g. clear, sure, certain, 
etc., are used. 
 
Extraposition bad 
(6) a.  [The correct approach] wasn?t clear. 
 b. * It wasn?t clear [the corect approach]. 
On the other hand, CPs can appear in either position. 
 
(7) a. [
CP
 Which of these approaches is correct] is not clear. 
 b. It is not clear [
CP
 which of these approaches is correct]. 
Now, it is perfect for the wh-phrases in sluicing to appear in such an extraposed position. 
This shows that the distribution of the wh-phrase in sluicing matches with that of CPs, 
not that of DPs. 
 
(8) One of these approaches is correct, but it?s not clear [which of them]. 
We can run the same test for exclamative sluicing, using factive adjectival predicates, 
such as odd, exciting, impresive, unbelievable. Those predicates do not take wh-
interogatives either in the subject position, or in the extraposed position. 
 
(9) a. * [Whether he solved those question] is impresive. 
 b. * It is impresive [whether he solved those questions]. 
Also, it is very clear that those predicates do not tolerate an extraposition of a DP. 
 
(10) a.  [His solution] was impresive. 
 b. * It was impresive [his solution]. 
On the other hand, an exclamative clause can appear either in subject position or in the 
extraposed position. 
 
(11) a. [
CP
 How (very) dificult a question he solved] is quite impresive. 
 b. It is quite impresive [
CP
 how (very) dificult a question he solved]. 
Finaly, the exclamative wh-phrase can appear by itself in the extraposed position. This 
suggests that what looks like just the exclamative wh-phrase is in fact a CP in which an 
IP is elided. 
 
Extraposed position 
(12) He told me that he solved an extremely dificult question, and it?s quite 
 impresive [how (very) dificult a question]. 
 83 
The second argument, also pointed out originaly by Ross (1969), is concerned with 
number agrement. In general, when a wh-clause appears in the subject position, it 
triggers singular agrement on the verb. This remains the same even if the wh-phrase 
itself is plural, suggesting that the number marking on the wh-phrase is independent from 
the agrement on the matrix verb in (13). 
 
Singular agrement 
(13) [
CP
 Which problems are solvable] {is/*are} not obvious. 
This agrement patern remains the same in sluicing cases. 
 
(14) Some of these problems are solvable, but [which problems] {is/*are} not obvious. 
We can apply this test in exclamative sluicing. First of al, an exclamative wh-clause with 
a plural exclamative wh-phrase requires singular agrement on the matrix verb. 
 
(15) [
CP
 How (very) many singers he knows] {is/*are} quite impresive. 
Then, as we saw in the interogative sluicing cases, this remains the same even in the 
exclamative sluicing cases. 
 
(16) I heard that he knows very famous singers, and [how (very) famous singers] 
 {is/*are} quite impresive. 
To sum up, there are at least two paralel observations with respect to sluicing in 
interogative wh-clauses and exclamative wh-clauses. Those two points suggest that the 
wh-phrases we can observe in the above examples are remnants of an elipsis proces. 
 
 
2.2 P-stranding generalization 
A further paralelism betwen sluicing in interogatives and exclamatives relates to the P-
stranding generalization noted in Merchant (2001:92). 
 
P-standing generalization 
(17) A language L wil alow preposition stranding under sluicing if L alows 
 preposition stranding under regular wh-movement. 
This generalization is that in languages like English, where preposition (P)-stranding is 
available in regular wh-movement, P-stranding is also available in sluicing. In such a 
language, the remnant of sluicing can be just a wh-phrase DP, even when the wh-phrase 
DP is a complement of a preposition in the original extraction site. On the other hand, in 
languages like Spanish and German, where P-stranding is not available in regular wh-
movement (a preposition wil be pied-piped along with the DP), P-stranding is not 
available in sluicing, either. Hence when the original extraction site for the wh-phrase is a 
 84 
complement of a preposition, the remnant of sluicing wil be a wh-phrase with a 
preposition; leaving a preposition in the elided part of the structure is not legitimate.
42
 
We can test whether this generalization holds in exclamative sluicing by 
constructing corresponding examples in those languages (most of the data on 
interogative sluicing is taken from Merchant (2001)). First, let us start with English 
examples. Example (18a) shows that P-stranding is possible in regular wh-interogatives, 
and example (18b) shows that who, which is the object of the preposition with in the 
antecedent clause, can appear without the preposition in sluicing. 
 
P-standing OK 
(18) a. Who was he talking with? 
 b. Peter was talking with someone, but I don?t know who. 
This patern in English interogatives holds true in exclamative sluicing as wel. 
Asuming that the wh-phrase what a tall guy originates in the complement position of the 
preposition with, (19) suggests that the preposition is left behind in exclamative sluicing. 
 
(19) Peter was talking with a very tal guy, and I can?t believe what a tal guy. 
Spanish is one of many languages that do not alow P-stranding. The following data 
indicates that the preposition has to be pied-piped when the complement wh-phrase 
undergoes movement. In addition, the sluicing example is not good if it lacks the 
preposition. 
 
Obligatory pied-piping 
(20) a. * Qui?n habl? con? 
   who speak.3SG with 
   ?Who does he talk with?? 
 b.  Ana habl? con alguien, pero no s?  
   A speak.3SG with someone but not know.1SG  
                                                
42
 But se Almeida & Yoshida (to appear) for a counterexample to the generalization 
from Brazilian Portuguese. They show that in Brazilian Portuguese, which does not alow 
P-stranding in wh-interogatives, P-stranding is posible in sluicing. 
(i) a. Com quem que a Maria dan?ou? 
  with who that the aria danced ?With whom did Mary dance?? 
 b. *Quem que a Maria dan?ou com? 
  who that the aria danced with ?Who did Mary dance with?? 
(i) a. A Maria dan?ou com algu?m, mas eu n?o lembro com quem 
  the aria danced with someone but I not remember with who 
  <a Maria dan?ou> 
  the aria danced 
  ?Mary danced with someone, but I don?t remember with who.? 
 b. A aria dan?ou com algu?m, mas eu n?o lembro quem 
  the Maria danced with someone but I not remember who 
  <a aria dan?ou com> 
  the Maria danced with 
  ?ary dance with someone, but I don?t remember who.? 
 85 
  ?(con) qui?n. 
     with who 
  ?Ana talks to someone, but I don?t know who.? 
Exclamative sluicing sems to show that the P-stranding generalization holds true. The 
contrast betwen (21b) and (21c) indicates that the preposition has to be pied-piped in 
wh-exclamatives in Spanish. 
 
Obligatory pied-piping 
(21) a.  Ana habl?  con un chico muy alto. 
   A speak.3SG with a boy very tal 
   ?Ana talks to a very tal boy.? 
 b.  Con qu? chico tan alto habl?  Ana! 
   with what boy  so tal speak.3SG A 
   ?What a tal boy Ana talks to!? 
 c. * Qu? chico tan alto habl?  Ana con! 
   what boy so tal speak.3SG A with 
The following sluicing example with exclamatives is not aceptable under the 
interpretation of elided IP. 
 
(22) ? Ana habl? con un chico muy alto, y no puedo 
  A speak.3SG with a boy very tal and no can.1SG 
 crerme qu? chico tan alto <habl? Ana con>. 
 believe what boy so tal  speak.3SG A with 
 ?Ana talks to a very tal boy, and I cannot believe what a tal boy 
 (Ana talks to).? 
The example controls for the possibility that the wh-phrase is just a complement of habl? 
as this verb requires a P complement. This can make sure that the preposition is left in 
the elided site. It sems safe to conclude that P-stranding generalization holds true with 
respect to exclamative sluicing. 
 
 
2.3 Swiping  
One interesting property that is related to pied-piping of prepositions in sluicing is that, in 
certain cases, the preposition can show up after the wh-phrase. Merchant (2002) 
extensively discusses various characteristics of this kind of sluicing, which he cals 
?swiping?. A typical example is shown in (23) (se also Ross 1969, Rosen 1976, 
Culicover 1999, van Craenenbroeck 2004, Sprouse 2005, and references cited therein).
43
 
                                                
43
 One thing we do not discuss here: Rosen (1976) notes that, in certain cases, swiping is 
beter if the preposition does not appear in the antecedent clause (se relevant discussion 
in Merchant (2002), van Craenenbroeck (2004) and Sprouse (2005). Merchant (2002) 
contends that a beter generalization sems to be that swiping is available with the 
preposition if the P is not selected. (i) is good since the P is not an argument of share, 
but the P with someone is a required constituent in (i). 
(i) Howard shares the apartment with someone, but I have no idea who with. 
 86 
 
Swiping 
(23) Peter went to the movies, but I don?t know who with. 
Note that the preposition with that sems to govern the wh-phrase appears to the right of 
the wh-phrase. Such a ?head-final? property is extremely uncommon in English, and 
acording to Merchant, it does not exist in non-sluicing environments. For example, it 
does not occur in wh-interogatives. 
 
(24) a. * Who with did you go to the movies? 
 b.  ith who did you go to the movies? 
He also notes that it is not true that al elipsis can license swiping: VP-elipsis does not. 
 
(25) *Peter went to the movies, but I don?t know who with he did. 
(26)  Peter went to the movies, but I don?t know who with. 
The generalization then sems to be that only sluicing can license swiping, whatever 
principle ultimately explains it. Let us mention one more restriction on swiping. It sems 
that only ?minimal? wh-operators occur in swiping (Merchant 2002:297), where 
?minimal? means ?head.? Then swiping is not available with complex wh-phrases such as, 
which car, how much time, etc., as shown below. 
 
Complex wh-phrases bad 
(27) a. * This opera was writen by an Italian composer in the 19th century, but 
   we?re not sure which composer by. 
 b. * She?s driving, but God knows what town to. 
 c. * He?s renting an apartment with a rich guy, and wait til you hear 
   how rich (of a guy) with. 
To the extent that the generalization of swiping holds, it is expected that swiping is not 
available with exclamative wh-phrases, asuming that a phrase like what a tall boy or 
how tall a boy is not a head. The following example ilustrates that sluicing is aceptable 
with pied-piping of the preposition, but swiping is not available.
4
 
 
                                                                                                                                            
(i) We were with someone. *I forgot who with. 
On the other hand, Nakao, Ono & Yoshida (2006) show that complement Ps license 
swiping when they undergo Heavy Shift or the complement P is a pseudo-gapping 
remnant. 
(ii) a. *John talked [to someone] yesterday, but I don?t remember who to. 
 b. ?John talked yesterday [to someone], but I don?t remember who to. 
(iv) a. *John talked to Mary, and Bil talked to someone. I don?t remember who to. 
 b. ?John talked to ary, and Bil did to someone. I don?t remember who to. 
4
 There is some discussion that preposition pied-piping in exclamatives is not as fully 
available as that in interogatives (Emonds 1985 and Nelson 1997). Although there are 
some variations among the native speakers, people find pied-piping of adjunct-like Ps 
beter than that of argument-like Ps. 
 87 
(28) a.  A: Susan was crying yesterday. 
   B: I know.. for what a stupid reason! 
 b. * B: I know.. what a stupid reason for! 
45
 
 
 
2.4 Pseudo-sluicing  
Merchant (2001:ch.4) is concerned with a potential complaint that sluicing is in general 
derived from a kind of cleft construction. Under such an analysis, the sluicing example in 
(29a) has an underlying form (which he cals ?pseudosluicing?) shown in (29b). 
 
Pseudo-sluicing 
(29) a. Gues who. 
 b. Gues who it was. 
It is not imediately clear whether such a cleft-like analysis has advantages over 
Merchant?s, but it might provide the impresion that the various island amelioration 
efects (to be discussed below) can be acounted for very easily.
46
 In any event, Merchant 
lists 10 arguments to show that pseudo-sluicing is not the general source of sluicing. 
Unfortunately, only two of them can be used for exclamatives, and we show that those 
ilustrate that the cleft structure is not the general source of exclamative sluicing. 
The first argument is based on the fact that adjuncts and implicit arguments are 
not so great as a pivot of a ?bare? cleft. Nonetheles, sluicing examples are aceptable. 
 
Adjuncts bad as a pivot 
(30) a. He fixed the car, but I don?t know how (*it was). 
 b. He fixed the car, but I don?t know why (*it was). 
Implicit argument 
(31) She reads regularly, but I don?t know what (*it was). 
We can apply this to exclamative sluicing. The results are the same. Importantly, 
exclamative sluicing is aceptable. 
 
                                                
45
 With regard to the matrix exclamative sluicing, one might question whether elipsis is 
in fact involved (compared to interogative sluicing), given that it is perfectly natural to 
utter just What a great car! without any linguistic antecedent. This might suggest that 
what is involved is a kind of deep anaphora (se Hankamer & Sag 1976). I think that 
there is nothing special with exclamatives on this point, since Norbert Hornstein (p.c.) 
informed me that there are limited cases even in interogatives that a sort of fragment can 
be uttered without any linguistic antecedent. Consider the folowing context: Fuji and 
Ono are in a TV studio as part of the audience for a political debate in which thre 
candidates for mayor are discussing various isues. After listening to what each candidate 
said, Ono asked Fuji, ?So, who?? Then the impresion that fragments are more readily 
available for exclamatives might not so obvious. 
46
 But se Lasnik (2005) for a clear ilustration that it is not true, since the cleft 
construction obeys island constraints as noted in Chomsky (1977). 
 88 
(32) a. He fixed the car, and I was amazed (at) how very fast (*it was). 
 b. He fixed the car, and I was amazed (at) how very fast it was that he fixed  
  the car. 
(33) a. He solved such a dificult problem, and I was surprised (at) how very 
  wisely (*it was). 
 b. He solved such a dificult problem, and I was surprised (at) how very 
  wisely it was that he solved such a dificult problem. 
Second, sluicing can remedy the violations on Left Branch Condition (LBC) (Ross 1967), 
but the cleft construction with LBC violation is not aceptable.
47
 
 
(34) a. *It is rich that he maried [a woman]. 
 b. *He maried a rich woman?wait til you hear how rich it is! 
We can construct similar examples in exclamative sluicing. 
 
(35) John maried an extremely rich woman, and it is unbelievable how very rich  
 (*it is). 
Those show that it is plausible that even in exclamative sluicing the pseudosluicing (or a 
?bare? cleft) is not a general source. 
 
 
3. Island Efects in Exclamatives 
In this section, we review arguments that exclamatives involve wh-movement. 
Particularly, the fact that exclamatives exhibit various island constraints suggests that 
movement of exclamative wh-phrases involved in exclamatives is characterized by the 
same gramatical operation as in interogatives.
48
 
The dependency of exclamative wh-phrases is unbounded when adequate bridge-
verbs are used. 
 
Unbounded dependency 
(36) a. [What an expensive ring]
1
 John thinks [Mary bought t
1
 ]! 
 b. [How fast]
1
 John thinks [Mary ate the pie t
1
 ]! 
Of course, this is a property that is easily observed in wh-interogatives. 
 
Unbounded dependency 
(37) a. [Which ring]
1
 does John think [Mary bought t
1
 ]? 
 b. [How fast]
1
 does John think [ary ate the pie t
1
 ]? 
Thus, wh-phrases in exclamatives can move long distance. Next, I wil discuss the data 
concerning various island efects with exclamatives. We focus on the following islands 
(e.g., Ross 1967, Chomsky 1973, Huang 1982). 
                                                
47
 Much more detailed examination is to be conducted in section 4.3.1. 
48
 Ross (1967:385) notes a few examples of exclamatives that exhibit island efects. Here 
we do a more comprehensive job. 
 89 
 
(38) a. Relative Clause 
 b. Noun-Complement 
 c. Adjunct Island  
 d. Subject Island  
 e. Sentential Subject Island  
 f. Coordinate Structure Constraint 
 g. That-t 
 h. Left Branch Condition 
In each case, we use what and how-type non-adverbial exclamative wh-phrases; in other 
words, the wh-phrase plugged in is in the form/patern of either what (a) adjective N or 
how adjective (a) N, and not the form of how adverb.
49
 We should emphasize that, as we 
expect, extractions of an exclamative wh-phrase out of those islands are totaly 
unaceptable. 
Let us start with data showing the extraction of an exclamative wh-phrase out of a 
relative clause.
50
 Examples in (39) are not aceptable. 
 
Relative Clause 
(39) a. * [What an expensive ring]
1
 John met [a girl [who bought t
1
 ]! 
 b. * [How expensive a ring]
1
 John met [a girl [who bought t
1
 ]! 
Extraction out of Noun-complement structures is also unaceptable. 
 
Noun-Complement 
(40) a. * [What an expensive ring]
1
 John heard [the rumor [that Mary bought t
1
 ]! 
 b. * [How expensive a ring]
1
 John heard [the rumor [that ary bought t
1
 ]! 
We can also show that Adjunct Islands are also respected in exclamative wh-phrases as in 
interogative wh-phrases. 
 
Adjunct Island 
(41) a. * [What an expensive ring]
1
 John left [after Mary bought t
1
 ]! 
 b. * [How expensive a ring]
1
 John left [after ary bought t
1
 ]! 
Extraction out of subjects is prohibited; it does not mater whether it is sentential or not. 
 
Subject Island 
(42) a. * [What a famous politician]
1
 [a biography of t
1
 ] is going to be published! 
 b. * [How famous a politician]
1
 [a biography of t
1
 ] is going to be published! 
 
                                                
49
 The data on Left Branch Condition is an exception. Se the actual data below. 
50
 For ease of exposition, I simply asume what has undergone movement is either what a 
adjective N or how adjective a N, without providing any evidence. Whether or not there is 
a separate movement within this, say, DP, must be cleared up at some point, as wel as 
whether those words form a constituent, but it might not be so relevant for the purpose of 
ilustrating the fact that the movement observed here obeys islands. 
 90 
Sentential Subject Island 
(43) a. * [What a powerful country]
1
 [that [the UN security council would criticize t
1
 
   at the next meting] has been widely reported! 
 b. * [How powerful a country]
1
 [that [the UN security council would criticize t
1
 
   at the next meting] has been widely reported! 
Exclamative wh-phrases in coordinate structures cannot be extracted out of a coordinated 
phrase. 
 
Coordinate Structure Constraint 
(44) a. * [What a prety girl]
1
 John met [Susan and t
1
 ]! 
 b. * [How prety a girl]
1
 John met Susan and t
1
 ]! 
That-t efects are also observed. 
 
That-t Efect 
(45) a. * [What a cute girl]
1
 John thought [that [ t
1
 is coming to the department]! 
 b. * [How cute a girl]
1
 John thought [that [ t
1
 is coming to the department]! 
Finaly, we wil look at the data to se whether or not Left Branch Condition is respected 
in exclamatives.
51
 
 
Left Branch Condition 
(46) a. * [What a cute girl?s]
1
 John saw [t
1
 pictures]! 
 b. * [How cute a girl?s]
1
 John saw [t
1
 pictures]! 
To sum up, we have sen that unbounded dependencies are observed in exclamatives, and 
movement of the exclamative wh-phrases obeys various islands as has been observed in 
interogative wh-phrases (and other operators like a null operator). 
 
 
4. Sluicing  
In this section, data on sluicing in English exclamatives are introduced. One property of 
sluicing, noted by Ross (1969), which grabs the heart of many syntacticians, is that 
sluicing apparently rectifies various island violations (se Merchant 2001). The goal here 
is simple; first, we show that sluicing is available in exclamatives, and second, not 
surprisingly, island amelioration is observed even in sluicing with exclamatives. Before 
investigating sluicing with exclamatives, a smal discussion of factivity is necesary, 
since it has a close tie with exclamatives, and is crucial for identifying exclamatives 
under sluicing. One basic example of sluicing is ilustrated below: 
 
Sluicing in interogatives 
(47) a. I heard that John met someone, but I wonder who. 
 b. I heard that John met someone, but I wonder who [
IP
 John met t ]. 
                                                
51
 Other kinds of Left Branch Condition, e.g., extraction of an AP out of a DP, wil be 
discussed in the next section. 
 91 
We asume that sluicing is an elipsis of IP, acknowledging that its implementation is a 
controversial mater (se Chung, Ladusaw, and McCloskey (1995), in addition to the 
work cited above). Sluicing leaves a wh-phrase that corresponds to an indefinite in the 
antecedent clause, and the wh-phrase is understood as it caries the same theta-role of the 
corresponding indefinite. In other words, the wh-phrase who in the above example is 
understood as the object of met. 
 
 
4.1 Factivity 
There is one point we have to note in sluicing in exclamatives. In the examples in (48), 
we can tel that there is a mising clause, and wonder is such a verb that takes an 
interogative clause as a complement. We need to make sure that we are dealing with 
exclamatives, by using an appropriate sort of predicate, namely, factive predicates 
(Kiparsky & Kiparsky 1970). This point is crucial especialy in the case where the 
material left under sluicing is how + adjective. Note that just by looking at that material, 
we cannot tel whether it is sluicing with interogatives or exclamatives, since a string 
how + adjective is available in interogatives as ilustrated in the following examples. 
 
(48) John went to Japan, but I wonder how long he wil stay there. 
We can tel that the above examples involve interogatives, since the matrix verb is 
wonder, which only selects an interogative complement. The reasoning behind the idea 
that factive predicates must be used is the following. It has been known that exclamative 
sentences, when embedded, must be selected by factive predicates (Eliott 1974, 
Grimshaw 1979). Thus, if the predicate that embeds exclamatives is an interogative-
taking one (or any non-factive one), examples are unaceptable. 
 
Exclamatives with factive predicates 
(49) a. It is unbelievable what a big car John bought. 
 b. I was surprised at what a large house he lives in. 
Exclamatives with non-factive predicates 
(50) a. *Mary wondered what a big car John bought. 
 b. *It is not clear what a large house he lives in. 
Conversely, an interogative clause, when embedded, cannot be selected by a factive 
predicate; the predicate that embeds an interogative clause must be one of interogative-
taking predicates. 
 
Interogatives with factive predicates 
(51) a. *It is unbelievable which car John bought. 
 b. *I was surprised at which house he lives in. 
Interogatives with interogative predicates 
(52) a. Mary wondered which car John bought. 
 b. It is not clear which house he lives in. 
Using this as a criterion, we cal tel whether a string how + adjective, as wel as other 
remaining materials under sluicing, is a part of an exclamative clause as long as it appears 
 92 
under a factive predicate. In the following sluicing examples with exclamatives, we use 
factive predicates in order to ensure that we are dealing with exclamatives. 
 
 
4.2 Material Left Behind 
Merchant (2001:62) notes the generalization that only wh-operators can remain in the left 
periphery of a clause under sluicing; no complementizer, no verb, and no clitic-like 
element is alowed to survive. For instance, in German, in which the matrix verb 
undergoes movement to C, the verb must not show up under sluicing. 
 
German 
(53) a. Max hat jemand eingeladen. 
  ax has someone invited 
  ?Max invited someone.? 
 b. Echt? Wen (*hat)? 
  realy who   has 
Building on Lobeck?s (1995) work, Merchant (2001:59) proposes the following condition 
for licensing IP-elipsis: 
 
(54) Only the null [+wh, +Q] C of interogatives wil license the null IP. 
Since sluicing is available with exclamatives, (54) must be modified in order to 
acommodate exclamatives to the following.
52
  
 
(55) Only the null [+wh, +Q/+Exc] C wil license the null IP. 
An isue I wil investigate below concerns what material in exclamatives can remain 
under sluicing. The generalization is summarized below. 
 
(56) a. Exclamative wh-phrase as a whole can remain under sluicing: 
  the whole DP; what a long paper, how long a paper 
  the whole Adverb Phrase; how long, how quickly 
 b. Extraction of just the wh-phrase out of DP or Adverb Phrase is not good; 
  * what
1
  ? [ t
1
 a long paper] 
  * how
1
  ? [ t
1
 long a paper] 
  * how
1
  ? [ t
1
 quickly] 
 c. Non-constituent cannot remain under sluicing: 
  *what a long, *what a, *how long a 
Note that the fact that elements listed in (56c) cannot remain under sluicing is not 
surprising since they are not even constituents. However, elements listed in (56b) must be 
handled with care. A similar observation, at least for how, is found in sluicing with 
                                                
52
 Of course, we surely want to make this condition more elegant so that it ties 
interogatives and exclamatives as a natural clas, excluding others, e.g., a relative clause 
formation and fre relatives. We leave this isue for future research. 
 93 
interogatives (se Ross 1969:277, Merchant 2001:164). First, both in interogatives and 
exclamatives, extraction of how alone without sluicing is flatly unaceptable. 
 
Extraction of ?how? alone in Interogatives
53
 
(57) a. *How
1
 does he want [a [ t
1
 detailed] list]? 
 b. *How
1
 did she buy [a [ t
1
 {expensive/fast/big}] car]? 
 c. *How
1
 is your brother [a [ t
1
 smart] doctor]? 
Extraction of ?how? alone in Exclamatives 
(58) a. *How
1
 he wants [a [ t
1
 detailed] list]! 
 b. *How
1
 she bought [a [ t
1
 {expensive/fast/big}] car]! 
 c. *How
1
 your brother is [a [ t
1
 smart] doctor]! 
Second, even with sluicing, examples are not aceptable when the wh-word how that is 
asociated with an adjective is the only element left. 
 
Atributive Adjectives in Interogatives with Sluicing (adapted from Merchant (2001)) 
(59) a. *He wrote a long paper, but I don?t know how
1
 he wrote [a [ t
1
 long] paper]. 
 b. *He wrote a long paper, but I don?t know how
1
 he wrote [a [ t
1
 long] paper]. 
Atributive Adjectives in Exclamatives with Sluicing 
(60) a. *He wrote an extremely long paper, but it is unbelievable how
1
 he wrote 
   [a [ t
1
 long] paper]. 
 b. *He wrote an extremely long paper, but it is unbelievable how
1
 he wrote 
   [a [ t
1
 long] paper]. 
The same observation holds with how, when it is asociated with predicative adjectives. 
 
Extraction of ?how? alone in Interogatives 
(61) a. *How
1
 is he [ t
1
 proud of his son]? 
 b. *How
1
 was he [ t
1
 interested in basebal]? 
Extraction of ?how? alone in Exclamatives 
(62) a. *How
1
 he is [ t
1
 proud of his son]! 
 b. *How
1
 he was [ t
1
 interested in basebal]! 
Predicative Adjectives in Interogatives with Sluicing  
(63) a. *He must be proud of his son, but I don?t know how
1
 he must be [ t
1
 proud  
   of his son]. 
 b. *He must be proud of his son, but I don?t know how
1
 he must be [ t
1
 proud 
   of his son]. 
Predicative Adjectives in Exclamatives with Sluicing  
(64) a. *He is extremely interested in basebal, but it is unbelievable how
1
 he was 
   [ t
1
 interested in basebal]. 
 b. *He is extremely interested in basebal, but it is unbelievable how
1
 he was 
   [ t
1
 interested in basebal]. 
                                                
53
 The intended reading is, for instance in (b), not asking the way or reason she buys a 
fast car, but asking the degre of ?fastnes? of the car. 
 94 
 
Based on these examples, Merchant (2001:166) suggests, citing Corver (1990), that the 
wh-word how asociated with an adjective is a head of a functional projection in DP (e.g., 
DegP taking AP as its complement; se Abney (1987) as wel), and given that heads are 
not alowed to move into SpecCP, examples above ilustrate a case of ilegitimate head 
movement. This explanation can be caried over to the prohibition of the extraction of 
what alone, asuming that what is also a head of a functional projection in DP.
54
 Note that 
extraction of what is never aceptable in exclamatives regardles of sluicing.
5
 
 
Extraction of ?what? in Exclamatives 
(65) a. *What
1
 he wrote [ t
1
 a long paper]! 
 b. *What
1
 he bought [ t
1
 a big car]! 
Extraction of ?what? in Exclamatives with Sluicing 
(66) a. *He wrote an extremely long paper, but it is unbelievable what
1
 he wrote 
   [ t
1
 a long paper]. 
 b. *He wrote an extremely long paper, but it is unbelievable what
1
 he wrote 
   [ t
1
 a long paper]. 
 
There is one caveat for asimilating how in interogatives with how in exclamatives. It 
has been observed that the how + adjective complex found in exclamatives alows an 
insertion of a modifier, such as very or extremely, in front of the adjective (Eliott 1974). 
Such a modifier cannot show up in the how + adjective complex in interogatives, 
however.
56
 
 
very + Adjective in Exclamatives 
(67) a. How (very) expensive a ring he bought! 
 b. How (very) interesting a paper he wrote! 
very + Adjective in Interogatives 
(68) a. How (*very) expensive a ring did he buy? 
 b. How (*very) interesting a paper did he write? 
                                                
54
 The exact syntactic diference betwen what and how in exclamatives is to be 
investigated. 
5
 Use of predicative adjectives in this context is irelevant given the fact that what cannot 
be used in exclamatives with predicative adjectives even pied-piping the whole 
constituent. 
(i) *What proud of his son he is! 
(i) *What interested in basebal he was! 
Thus, we focus on the cases with atributive adjectives here. 
56
 Norbert Hornstein (p.c.) nonetheles provides an example where a modifier very 
appears in interogative wh-phrase: 
(i)   A: Sam bought a realy expensive ring for Mary. It was realy realy expensive. 
    B: Oh, yeah? How very expensive a ring was it? 
It might be that the example is fine if the part very expensive in (iB) is read in a special 
?quote and unquote? intonation of some sort. 
 95 
 
This data does not argue against the idea that the how in exclamatives is a head of a 
functional projection in DP, but it might suggest that how in exclamatives is a functional 
head, but a diferent (maybe higher) head of a functional projection.
57
 
 
 
4.3 Data  
Let us start the discussion with cases in which no island is involved. As has been 
mentioned, factive predicates are used to embed exclamatives. 
 
(69) John wrote an extremely long paper, .. 
 a. .. and it is unbelievable what a long paper John wrote. 
 b. .. and it is unbelievable what a long paper John wrote. 
(70) John wrote an extremely long paper, .. 
 a. .. and it is unbelievable how long a paper John wrote. 
 b. .. and it is unbelievable how long a paper John wrote. 
 c. .. and it is unbelievable how long John wrote a paper. 
Below, we look at various cases with islands. Note that we have established that 
movement in exclamatives obeys those islands, and we se that island violations are 
rectified under sluicing. The materials left after sluicing do not change; only those listed 
in (56a) can survive. Thus, we use those materials in the ilustration below. 
 
 
4.3.1 Left Branch Condition 
One very interesting observation noted by Merchant (2001), which shows a clear contrast 
with cases with exclamatives, is the following. He tested many island constraints to se 
which constraints are rectified under sluicing. One island constraint that is quite relevant 
in our cases with exclamatives is LBC, since LBC regulates the extractability of 
adjectival projections, and exclamatives also involve movement of adjectival projections. 
Merchant (2001) notes that LBC efects disappear under sluicing, as shown below. 
 
Sluicing with LBC 
(71) a. He wants a detailed list, but I don?t know how detailed. 
 b. She bought an {expensive/fast/big} car, but I don?t know how 
                                                
57
 Alternatively, we can speculate that what and how in exclamatives, both occurring at 
the beginning of a clause, are a special complementizer for exclamatives. Following the 
generalization by Merchant (2001:62) that complementizers cannot remain under sluicing, 
the fact that what and how cannot survive after sluicing must be explained by whatever 
analysis acounts for the generalization: a licensing condition of IP-elipsis. Se 
Merchant (2001: chapter 2, section 2). This idea might work with what since the DP 
following what sems to be a DP in which the materials inside are normaly ordered (e.g., 
a long paper), but it might be harmful with how. It requires an additional stipulation for 
the ordering restriction on the materials following how (e.g., long a paper) via a 
mechanism of pied-piping. 
 96 
  {expensive/fast/big}. 
 c. Your brother is a smart doctor, but it?s not clear how smart. 
However, if the nominals that contain the how + adjective complex is inside another 
island, examples are unaceptable (Merchant 2001:175). He shows examples with 
Adjunct Island and Relative Clause. 
 
LBC + Adjunct Island  
(72) a. *She?l be angry if he buys an expensive car, but I don?t know how expensive. 
 b. *He got stresed because his boss wants a detailed list, but I don?t know how 
   detailed. 
LBC + Relative Clause 
(73) a. *She met a guy who bought an {expensive/fast/big} car, but I don?t know 
   how {expensive/fast/big}. 
 b. *They want to hire someone who writes thorough reports, and wait til you se 
   how thorough. 
 c. *She wants to met a guy who buys old paintings, but she didn?t say how old. 
He also shows that this cannot be due to the ban on long distance extraction of how-
phrase by noting that in general, a long distance extraction does not cause any problem in 
sluicing. 
 
(74) He said he needed a detailed list, but wait til you hear how detailed. 
The contrast betwen cases with and without islands is quite peculiar. To make the mater 
even more peculiar, as wil be shown below, exclamatives do not show this contrast; 
namely, extraction of the how + adjective complex sems to be alowed regardles of the 
presence of an island. We have already sen in (70) that examples with sluicing are fine if 
the nominal that contains an adjectival projection is not inside of an island. Below, we 
wil se that examples remain aceptable even if the nominal with an adjectival 
projection is placed inside of an island. This is the case not only with Relative Clause and 
Adjunct Islands, as we wil also se examples with Noun-Complement Islands, Subject 
Islands, Sentential Subject Condition, and Coordinate Structure Constraint. 
 
 
4.3.2 Other Islands 
Examples in (76) and (77) show that a violation of the Complex NP Constraint (Relative 
Clause) is ameliorated if sluicing applies. 
 
Relative Clause + Sluicing 
(75) John met a profesor who wrote an extremely long paper, 
 a. *.., but it is unbelievable [what a long paper]
1
 John met a profesor 
   who wrote t
1
. 
 b.  .., but it is unbelievable [what a long paper]
1
 John met a profesor 
   who wrote t
1
 . 
 97 
Relative Clause + Sluicing 
(76) John met a profesor who wrote an extremely long paper, 
 a. *.., but it is unbelievable [how long a paper]
1
 John met a profesor who  
   wrote t
1
. 
 b.  .., but it is unbelievable [how long a paper]
1
 John met a profesor who 
   wrote t
1
 . 
 c.  .., but it is unbelievable [how long]
1
 John met a profesor who wrote 
   a t
1
 paper. 
The same is true with Noun-Complement Islands. 
 
Noun-Complement + Sluicing  
(77) Susan heard a news that a researcher found an extremely big spider, 
 a. *.., but it is unbelievable [what a big spider]
1
 Susan heard a news that a 
   researcher found t
1
. 
 b.  .., but it is unbelievable [what a big spider]
1
 Susan heard a news that a 
   researcher found t
1
 . 
Noun-Complement + Sluicing  
(78) Susan heard news that a researcher found an extremely big spider, 
 a. *.., but it is unbelievable [how big a spider]
1
 Susan heard news that a 
   researcher found t
1
. 
 b.  .., but it is unbelievable [how big a spider]
1
 Susan heard news that a 
   researcher found t
1
 . 
 c.  .., but it is unbelievable [how big]
1
 Susan heard news that a researcher found 
   a t
1
 spider. 
Though a litle degraded, Adjunct Island violations are also repaired under sluicing. 
 
Adjunct Island + Sluicing  
(79) Bil came to office because he had to sign an extremely important document, 
 a. *.., but it is unbelievable [what an important document]
1
 Bil came to  
   office because he had to sign t
1
. 
 b. ?.., but it is unbelievable [what an important document]
1
 Bil came to  
   office because he had to sign t
1
 . 
Adjunct Island + Sluicing  
(80) Bil came to the office because he had to sign an extremely important document, 
 a. *.., but it is unbelievable [how important a document]
1
 Bil came to ofice 
   because he had to sign t
1
. 
 b. ?.., but it is unbelievable [how important a document]
1
 Bil came to ofice 
   because he had to sign t
1
 . 
 c. ?.., but it is unbelievable [how important] Bil came to office because he 
   had to sign a t
1
 document. 
Other island cases exhibit more or les a similar patern. The state of afairs is that 
examples with how-exclamatives are slightly beter than that with what-exclamatives. 
 
 98 
Subject Island + Sluicing  
(81) A biography of an extremely famous politician is going to be published, .. 
 a. *.., and it is unbelievable [what a famous politician]
1
 [a biography of t
1
 ] is 
   going to be published. 
 b. ?.., and it is unbelievable [what a famous politician]
1
 [a biography of t
1
 ] is 
   going to be published. 
Subject Island + Sluicing  
(82) A biography of an extremely famous politician is going to be published, .. 
 a. *.., and it is unbelievable [how famous a politician]
1
 [a biography of t
1
 ] is 
   going to be published. 
 b. ?.., and it is unbelievable [how famous a politician]
1
 [a biography of t
1
 ] is 
   going to be published. 
 c. ?.., and it is unbelievable [how famous]
1
 [a biography of a t
1
 politician] is 
   going to be published. 
Sentential Subject Island + Sluicing 
(83) That the UN security council would criticize an extremely powerful country at the 
 next meting has been widely reported, .. 
 a. *.., and it is unbelievable [what a powerful country]
1
 [that [the UN security  
   council would criticize t
1
 at the next meting]] has been widely reported. 
 b. ?.., and it is unbelievable [what a powerful country]
1
 [that [the UN security 
   council would criticize t
1
 at the next meting]] has been widely reported. 
Sentential Subject Island + Sluicing 
(84) That the UN security council would criticize an extremely powerful country at the 
 next meting has been widely reported, .. 
 a. *.., and it is unbelievable [how powerful a country]
1
 [that [the UN security  
   council would criticize t
1
 at the next meting]] has been widely reported. 
 b. ?.., and it is unbelievable [how powerful a country]
1
 [that [the UN security 
   council would criticize t
1
 the next meting] has ben widely reported. 
 c. ?.., and it is unbelievable [how powerful]
1
 [that [the UN security council 
   would criticize a t
1
 country at the next meting]] has been widely reported. 
Coordinate Structure Constraint + Sluicing  
(85) John met Susan and an extremely cute girl, .. 
 a. *.., and it is unbelievable [what a cute girl]
1
 John met [Susan and t
1
 ] 
 b. ?.., and it is unbelievable [what a cute girl]
1
 John met [Susan and t
1
 ]. 
Coordinate Structure Constraint + Sluicing  
(86) John met Susan and an extremely cute girl, .. 
 a. *.., and it is unbelievable [how cute a girl]
1
 John met [Susan and t
1
 ]. 
 b. ?.., and it is unbelievable [how cute a girl]
1
 John met [Susan and t
1
 ]. 
 c. ?.., and it is unbelievable [how cute]
1
 John met Susan and [a t
1
 girl]. 
In this section, we saw a series of examples showing that various island efects disappear 
once sluicing has applied. For example, extraction of what a big car or how big a car out 
of various islands is in general prohibited, showing island efects, but once the IP 
following the extracted material has been elided, the examples are aceptable. Moreover, 
it is quite significant that the same contrast is observed with extraction of how big alone. 
 99 
Merchant (2001) notes, with his sluicing with interogative clauses, that such extraction 
(a clear LBC violation) is possible only when no other islands are involved. Once the 
extraction site is buried under another island, e.g., a complex NP, the examples are not 
aceptable. We, then, saw cases where exclamatives are diferent; regardles of the 
existence of other islands, LBC violations are remedied by sluicing. 
 
 
5. Sumary  
In this chapter, we observed that sluicing is available not only for interogatives, but also 
for exclamatives. This forced us to revise a generalization noted by Merchant (2001) that 
only interogative complementizers alow sluicing. We found out that, for sluicing, 
interogatives and exclamatives form a natural clas, excluding relative clauses and fre 
relatives, among others. With respect to material left under sluicing, we suggest that the 
generalization based on interogatives is also applicable. The fact that what and how 
alone cannot remain under sluicing can be explained by asuming Corver (1990)?s 
suggestion that these are a head of some extended projection of AP, and a head cannot 
move into [Spec,CP]. Then, we saw various cases in which islands are involved. We saw 
that, as with interogatives, various island efects can be rectified by sluicing. The most 
interesting finding here is that although Merchant (2001) has observed that LBC 
violations in interogatives can be rectified only when there are no other islands involved, 
we found that LBC violations in exclamatives can be rectified regardles of the 
involvement of other islands. LBC efects disappear not only when there is no other 
island involved, but also when the launching site is embedded under other islands. 
 
 
 
 
 100 
 
CHAPTER 5: LICENSING CONDITIONS OF EXCLAMATIVES 
AND INTERVENTION EFECTS 
 
1. Introduction: Two puzles 
This chapter is concerned with the distribution of exclamative clauses in English. We 
show that exclamatives are licensed by being c-commanded by a factive operator or 
predicate, contrary to the traditional asumption that exclamatives are selected by a 
factive predicate. It wil be argued that the distribution of exclamative clauses can be 
acounted for by asimilating it to aggresively-non-D-linked wh-interogative clauses 
(the hel-wh-interogatives), although the relevant property for the former is factivity, 
while that for the later is non-veridicality. We show that the analysis proposed by den 
Dikken & Giannakidou (2002) can be extended to explaining the distribution of 
exclamatives in English. The general conclusion obtained in this chapter suggests that 
exclamatives and the hel-wh-interogatives are similar in that both involve very intricate 
semantic and pragmatic information. Figuring out the exact properties of both alows us 
to investigate to what extent semantic and pragmatic information play a role in syntax. 
Recal that in chapter 2, we argued that notions such as mood or speaker?s point of view 
play a very important role in syntax, and we encode those by functional projections in the 
CP layer. We hope that the conclusion in this chapter helps us to examine what kind of 
projections are needed for exclamatives in English. 
 
 
1.1 Selection of Exclamatives 
It has been a standard observation that, when embedded, exclamative clauses have to be 
selected by a factive predicate. One repeatedly discussed paradigm, in (1) and (2), is that 
exclamative clauses cannot appear as the complement of an interogative predicates, such 
as wonder or ask, which require an interogative wh-clause (se e.g., Eliott 1971, 1974, 
Grimshaw 1979, Zanuttini & Portner 2003, Fuji & Ono 2005).
58
 
 
Embedded under Factive Predicates 
(1) a.  It?s amazing what a fool he is. 
 b.  Susan regrets what a cheap house her parents bought. 
(2) a. * John wil ask what a fool he is. 
 b. * Susan wonders what a cheap house her parents bought. 
Much atention has been paid to interogative clauses, the distribution of which is 
complementary to that of exclamative clauses as shown in (3) and (4). Interogative 
clauses, such as ones headed by whether, cannot occur as a complement of factive 
predicates, such as be amazing, regret or be surprised at. 
 
                                                
58
 We abstract away from cases where interogative predicates select categories other 
than a clause, such as I asked the name of the new student. Se Grimshaw (1979) and 
Pesetsky (1982, 1991). 
 101 
Complementary Distribution 
(3) a.  Fred wil ask whether he is a fool. 
 b.  Susan wonders whether her parents live in a large house. 
(4) a. * It?s amazing whether he is a fool. 
 b. * Susan regrets whether he lives in a large house. 
It is also wel known that predicates such as know or find out can embed either 
interogative or exclamative clauses. 
 
Embedding Interogatives or Exclamatives (Grimshaw 1979; (8) 
(5) a. John knows whether he is a fool. 
 b. John knows what a fool he is. 
 c. John found out whether he lives in a large house. 
 d. John found out what a large house he lives in. 
Based on the paradigm ilustrated above, Grimshaw (1979) suggests frames for 
complement selection building upon semantic types as shown in (6), which aims at 
explaining al the paterns observed above. 
 
Semantic Selection Frames 
(6) a. [ __ Q ]  interogative taking predicates 
 b. [ __ E ]  exclamative taking predicates 
 c. [ __ Q / E ] both 
The particular intuition behind this is the close paralelism betwen interogatives 
and exclamatives. That is, a certain type of clause (here, we are only concerned with 
clauses) has to be selected by a unique type of predicate when it is embedded. The 
connection betwen a clause and a predicate sems to be very tight in the case of 
interogatives. Interogative wh-clauses have to be selected by one of the interogative 
predicates and interogative predicates have to select an interogative clause as a 
complement. That is how selection works. 
It sems true that exclamative wh-clauses have to be selected by a factive 
predicate when they are embedded, but it is certainly not the case that factive predicates 
have to select an exclamative clause as a complement. Evidently, that-clauses can appear 
as a complement of factive predicates, as shown in (7). Examples in (8) show that the 
sentential complement for predicates such as ask or wonder is always interogative. On 
the other hand, there is no predicate that must select exclamative clauses. If there were 
such a predicate, it would show the patern shown in (9). If we take this point seriously, it 
suggests that the paralelism betwen interogatives and exclamatives is not as complete 
as has been claimed. 
 
That-clause complement with factive predicates 
(7) a.  It?s amazing that John finaly solved the problem. 
 b.  Susan regrets that her parents bought a very cheap house. 
That-clause never goes with ask and wonder 
(8) a. * Fred wil ask that he is a fool. 
 b. * Susan wonders that he lives in a large house. 
 102 
Imaginary exclamative predicate: REGRET 
(9) a.  Fred REGRETs what a cheap house his parents bought. 
 b. * Fred REGRETs that his parents bought a cheap house. 
 
 
1.2 Licensing Exclamatives at a Distance 
The second puzzle is related to the point made above. Since an interogative predicate 
selects an interogative clause, the wh-clause appears as a sister to the interogative 
predicate, as shown in (10). (10b) is ungramatical since there is no interogative clause 
that is a sister of wonder. 
 
Very local relation 
(10) a.  I wonder whether it is unclear Mary likes Christmas shopping. 
 b. * I wonder it is unclear whether ary likes Christmas shopping. 
On the other hand, there is evidence that the relationship betwen the factive predicate 
and the exclamative wh-clause is not selection. First, the predicate be unclear usualy 
cannot select an exclamative wh-clause, as shown in (11), while an interogative wh-
clause can appear as a complement of this predicate, indicating that be unclear is not a 
factive predicate. Now, the minimal pair in (12) shows that exclamative wh-clauses can 
appear as the sister of the predicate be unclear if a factive predicate such as regret or 
realize also appears in a position from which it can c-command the exclamative wh-
clause. Note that predicates such as think, say or claim, being non-factive predicates, do 
not help license the exclamative clause.
59
 
 
Long-Distance Licensing of Exclamative Wh-Phrases 
(11) a. * It is unclear how very big a house he has. 
 b.  It is unclear whether he has a big house. 
(12) a.  Mary regrets/realizes that it is unclear how very big a house he has. 
 b. * ary thinks/says/claims that it is unclear how very big a house he has. 
This observation is quite interesting when compared with the examples in (10). Again, 
the examples in (10) show that an interogative predicate requires an interogative clause 
in its complement position, and an interogative clause is also required to be the 
complement of an interogative predicate when embedded. On the other hand, the 
minimal pair in (12) shows that, unlike a very strict local relation like ?selection?, the c-
command relation betwen the clause and the predicate licenses exclamative wh-clauses. 
                                                
59
 We have to be careful with the choice of the predicate. The paradigm can be 
established with a non-factive predicate, such as be unclear that can select either a that-
clause or an interogative clause. For that reason, we cannot use predicates such as be 
likely, be certain, etc., which only select that-clause. Relevant discussion is found in 
Adger & Quer (1997, 2001) where they discuss cases that the predicates such as be clear 
and be obvious that do not take a complement clause headed by if, but they do when the 
predicate is in question or negation is added. 
 103 
A key to understanding this contrast betwen the interogative and exclamative 
wh-clause wil be found if the distribution of exclamative wh-clauses is investigated from 
a diferent perspective. One conclusion we draw is that although it is true that 
interogative wh-clauses are licensed by selection by an appropriate predicate, 
exclamative wh-clauses are licensed by c-command: an embedded exclamative clause is 
only licit if it is c-commanded by certain veridical items, such as a factive operator (se 
Melvold 1991, Watanabe 1993, Zanuttini & Portner 2003). In order to show that the 
distribution of exclamative wh-clause is regulated by a proper c-command relation, we 
propose a system which is quite similar to the one argued for by den Dikken & 
Giannakidou (2002), who investigated the distribution of ?aggresively non-D-linked? 
wh-phrases (wh-the hel). They argue that wh-the hel phrases are polarity items, and they 
have to occur in a non-veridical context, which is typicaly established by the presence of 
negation, a Q-operator, or an interogative predicate.
60
 We argue that exclamative wh-
phrases are also some kind of ?polarity? sensitive item, but the environment they have to 
occur in is not non-veridical, but a subset of the veridical contexts, namely factivity. 
Therefore, to the extent that the proposal here is succesful, additional support can be 
given to the approach by den Dikken & Giannakidou (2002). 
 
 
2. den Diken & Giannakidou (2002) 
This section summarizes the analysis of wh-the hel phrases proposed by den Dikken & 
Giannakidou (2002). First, we introduce their phrase structure and discussion with 
respect to the landing sites of wh-phrases in English, then we review the critical 
component of their argument. 
 
2.1 Phrase Structure and Landing Sites of Wh-phrases 
Den Dikken & Giannakidou (2002) asume the multi-layered CP structure introduced in 
Rizi (1997) where an IP is dominated by at least thre functional projections: a Focus 
phrase (FocP), a Topic phrase (TopP) and a CP, as shown in (13). 
 
Multi-Layered CP 
(13) [
CP
 ? [ C [
TopP
 ? [ Topic [
FocP
 ? [ Focus [
IP
 ? ]]]] 
Based on this phrase structure, they argue that the landing site of wh-movement in 
English is not uniform betwen the embedded and matrix clauses. Specificaly, wh-
movement targets [Spec,CP] in embedded clauses, but [Spec,FocP] in matrix clauses. 
First, one piece of support for the hypothesis comes from Hungarian data, where a wh-
phrase moves into [Spec,FocP]. Examples in (14) show that in Hungarian, wh-phrases 
need to appear in a position imediately before the verb. Then, when there is another 
focused phrase within a clause, a wh-phrase cannot co-occur with the focused phrase, as 
shown in (15). Examples in (15) are explained straightforwardly if both phrases target the 
same [Spec,FocP]. 
 
                                                
60
 The precise characterization of (non)veridicality is not an isue we can discuss here. 
Se Giannakidou (1999). 
 104 
Hungarian Focus Movement (dDG (45)) 
(14) a. Mari kit h?vott meg?  
  ari(TOP) who-ACC invited PV 
  ?Who did Mari invite?? 
 b. Kiv?nsci vagyok hogy Mari kit h?vott meg. 
  curious I-am that ari(TOP) who-ACC invited PV 
  ?I wonder who Mari invited.? 
Wh-Phrase and Focus Element Cannot Co-occur (dDG (46)) 
(15) a. * Kit MARI h?vott meg? 
   who-ACC ari(FOC) invited PV 
 b. * MARI kit h?vott meg? 
   ari(FOC) who-ACC invited PV 
With respect to the diferent landing sites of wh-movement betwen the embedded and 
matrix clauses, they use Pesetsky?s (1989) observation on the relative order of wh-
phrases and a leftward-moved topic phrases in English. Pesetsky (1989) observes that a 
wh-phrase follows a topic phrase in the matrix clause, while it precedes a topic phrase in 
the embedded clause. Asuming that there is a unique topic phrase in a clause, it sems 
that there is more than one position to which a wh-phrase moves. In the matrix clause 
(16), a wh-phrase moves to [Spec,FocP], which is lower than TopP, and in the embedded 
clause (17), a wh-phrase moves to [Spec,CP], which is higher than TopP. 
 
Pesetsky 1989: topicalization 
(16) a. ?A book like this, why should I buy? (his (39)) 
 b. *Why, a book like this, should I buy? 
(17) a. ?I wonder why, a book like this, I should buy. (his (41)) 
 b. *I wonder, a book like this, why I should buy. 
Wh-the-hel phrases exhibit the same ordering restriction with respect to the topic phrase. 
 
Same Observation with Wh-the-Hel Phrases 
(18) a. ?A book like this, why the hel should I buy? 
 b. *Why the hel, a book like this, should I buy? 
(19) a. ?I wonder why the hel, a book like this, I should buy. 
 b. *I wonder, a book like this, why the hel I should buy. 
 
 
2.2 Wh-the-Hel 
?Aggresively non-D-linked? wh-phrases are wh-phrases with an extra phrase such as the 
hel or on earth. They are characterized as having semantic or pragmatic properties that 
are not readily available with ?regular? wh-phrases (se Pesetsky 1987, den Dikken & 
Giannakidou 2002, Le 1994, and Obenauer 1994 for more detailed discussion). 
Comparing examples in (20), an interogative clause with a wh-the-hel phrase is used 
with an additional psychological atitude toward the answer (e.g., surprise), or it has a 
negative inference such as nobody was supposed to buy the book. 
 105 
 
(20) a. Who bought the book? 
 b. ho the hel bought the book? 
Den Dikken & Giannakidou (2002) introduce quite interesting observations with respect 
to the distribution of wh-the-hel phrases. They note that while wh-the-hel phrases cannot 
usualy occur under a factive predicate like know, this restriction is lifted when the 
predicate is negated. Note further that polarity items such as anyone are also sensitive to 
elements like negation. 
 
Complement of veridical predicates: Usually, ?know? cannot take wh-the-hel 
(21) a.  I know who would buy that book.  (dDG (5)) 
 b. * I know who the hel would buy that book. (with factive predicates) 
But OK if c-commanded by negation 
(22) a.  I don?t know who would buy that book. (dDG (6)) 
 b.  I don?t know who the hel would buy that book. 
Polarity item: anyone 
(23) a. * I know anyone would buy that book. 
 b.  I don?t know anyone would buy that book. 
The paralelism betwen the wh-the hel and polarity items goes further. Wh-the-hel 
phrases are aceptable as long as they are c-commanded by elements like only, nobody, 
or if they are inside of if-clauses. 
 
(24) a. Only John knows who the hel wrote this secret report. (dDG (13-15) 
 b. If John knows who the hel wrote this secret report, he should tel us now. 
 c. Nobody knows who the hel wrote this secret report. 
(25) a. Only John knows whether anyone is aware of this secret report. (dDG (13-15) 
 b. If John ses anybody, he should tel us now. 
 c. Nobody knows whether anyone wrote this secret report. 
Based on these observations, den Dikken & Giannakidou (2002) conclude that wh-the-
hel phrases are polarity items like anyone, which have to be licensed under certain 
contexts (i.e., non-veridical contexts). 
 
 
2.3 Basic Analysis 
In the previous section, we have reviewed some quite intriguing distributional properties 
of wh-the-hel phrases. This section reviews den Dikken & Giannakidou?s (2002) 
analysis for how those observations follow from the polarity sensitive status of wh-the-
hel phrases. Recal the paradigm in (26) where wh-the-hel phrases are not alowed to 
appear under predicates such as know, but they are if the predicate is negated. This 
observation follows straightforwardly from their analysis since the negation wil c-
command the wh-the-hel phrase. With respect to the ungramatical example (26a), there 
is no negation that can c-command the wh-the-hel phrase. 
 
 106 
No Embedding under ?know? 
(26) a. * I know who the hel would buy that book. 
 b.  I don?t know who the hel would buy that book. 
 c.  I don?t know anyone would buy that book. 
At this point, it is mysterious how the wh-the-hel phrase at the beginning of the 
sentence could be licensed, since there is no overt element to the left of the wh-the-hel 
phrase. In order to explain this data, the asumption that the landing site of wh-movement 
in the matrix clause is [Spec,FocP] becomes very important. Den Dikken & Giannakidou 
(2002) argue that there is a Q operator in C, and it can c-command the wh-the-hel phrase 
in [Spec,FocP]. One piece of evidence that a Q operator is a legitimate licenser comes 
from the observation that polarity items are licensed in interogatives in general as shown 
in (28). 
 
Wh-the-hel phrases at the root clause is c-commanded by a Q operator.
61
 
(27) a. Who the hel left? 
 b. [
CP
 C
Q
 [
FocP
 [who the hel]
i
 [ Foc [
IP
 t
i 
left ]] 
(28) Did Bil eat anything? 
There is one more way to satisfy the polarity sensitivity of wh-the-hel phrases. 
Recal that when a wh-the-hel phrase occurs in embedded clauses, it moves to [Spec,CP], 
not [Spec,FocP]. Presumably [Spec,CP] cannot be c-commanded by a Q operator, so the 
wh-the-hel phrase has to have some alternative way to satisfy the requirement. The 
suggestion by den Dikken & Giannakidou (2002) is that the interogative predicate, 
wonder in (29), c-commands the wh-the-hel phrase to satisfy the requirement. 
 
Wh-the-hel phrases at the embedded clause is c-commanded by an interogative 
predicate.
62
 
(29) a. I wonder who the hel left? 
 b. I [
VP
 wonder [
CP
 [who the hel]
i
 C
Q
 [
FocP
 Foc [
IP
 t
i 
left ]] 
So far, we have sen at least thre ways to satisfy the polarity sensitivity of wh-the-hel 
phrases: negation, a Q operator, and an interogative predicate. The analysis reviewed 
here in fact can acount for further properties of wh-the-hel phrases, which we discuss in 
the next section. 
 
 
                                                
61
 Examples of wh-questions with negative polarity item are not good, such as: 
(i) *Who bought anything? 
Later, we discuss their acount for the example above. In short, this is a case of an 
intervention efect. The wh-phrase intervenes betwen the licensor (Q-operator) and 
anything. 
62
 This analysis is only concerned with predicates that must select an interogative clause; 
i.e., predicates such as know are not included. Se den Dikken & Giannakidou (2002). 
 107 
2.4 Wh-the-Hel in Multiple Wh-Constructions 
One property of wh-the-hel phrases that is not sen with regular wh-phrases is that when 
a wh-the-hel phrase in the matrix clause is used in a multiple wh-construction, it does not 
admit a pair-list interpretation. Den Dikken & Giannakidou (2002) observe that a pair-list 
interpretation only becomes available when a wh-the-hel phrase is embedded.
63
 
 
Root / Embedded Asymetry 
(30) a. Who the hel is in love with who? [pair-list reading unavailable] 
 b. I {am wondering/would like to know} who the hel is in love with who. 
       [pair-list reading available] 
They argue that their analysis can acount for this asymmetry betwen the matrix and 
embedded clauses, under the asumption that in-situ wh-phrase must be licensed in the 
Focus projection and there is only one [Spec,FocP]. (31) ilustrates one possible 
derivation; if the in-situ wh-phrase with who undergoes covert movement to [Spec,FocP], 
the wh-the-hel phrase cannot move into [Spec,FocP]. Then, alternatively, the wh-the-hel 
phrase undergoes movement to [Spec,CP]. However, this position cannot be c-
commanded by a Q operator, thus this derivation cannot receive a pair-list 
interpretation.
64
 Now, consider the alternative derivation in (32). Suppose the wh-the-hel 
phrase moves to [Spec,FocP] as has been argued to happen in the matrix clause. Then, the 
in-situ wh-phrase cannot move to [Spec,FocP], since this position is already occupied by 
the wh-the-hel phrase. Asuming that in-situ wh-phrases are licensed in the Focus 
projection, the in-situ wh-phrase in (32) cannot be licensed. 
 
Analysis for (30a) #1 
(31) [
CP
 [who the hel]
i
 C
Q
 [
FocP
 [with who]
j
 Foc [
IP
 t
i
 is in love t
j
 ]] 
 The wh-the-hel phrase is not c-commanded by the Q operator. 
Analysis for (30a) #2 
(32) [
CP
 C
Q
 [
FocP
 [who the hel]
i
 Foc [
IP
 t
i
 is in love [with who] ]] 
 The wh-the-hel phrase is licensed, but not in-situ wh-phrase. 
On the other hand, in the embedded clause, there is a way to license both of the wh-the-
hel phrase and the in-situ wh-phrase, as shown in (33). As we have sen in (31), the only 
way to license an in-situ wh-phrase is by moving it to [Spec,FocP]. Since this is in the 
embedded clause, the wh-the-hel phrase moves to [Spec,CP]. Although this position 
cannot be licensed by a Q operator, there is an interogative predicate, wonder, which can 
c-command the wh-the-hel phrase. Hence, this derivation can receive a pair-list 
interpretation. 
 
                                                
63
 Se some related discussion in Le (1994). Also, we disregard the availability of 
single-pair echo reading; se den Dikken & Giannakidou (2002). 
64
 On the other hand, multiple wh-questions that do not contain wh-the hel phrase are 
licensed since the overtly moved wh-phrase can move to [Spec,CP], making available 
[Spec,FocP] for the in situ wh-phrase. This is possible since the regular wh-phrase does 
not have to be c-commanded by the Q operator. 
 108 
Analysis for(29b) 
(33) ? wonder [
CP
 [who the hel]
i
 C
Q
 [
FocP
 [with who]
j
 [
IP
 t
i
 is in love t
j
 ]] 
 
 
2.5 Intervention Efects 
2.5.1 No In-Situ Wh-the Hel Phrase 
It has been observed that wh-the hel phrases are restricted from appearing in situ 
(Pesetsky 1987, Le 1994, and Obenauer 1994). In contrast with the multiple questions 
we have sen in the previous section, there is no asymmetry betwen the matrix and 
embedded clauses. 
 
(34) a. *Who is in love with who the hel? 
 b. *I {am wondering/ would like to know} who is in love with who the hel. 
Den Dikken & Giannakidou (2002) argue that the lack of in-situ wh-the-hel phrases is an 
intervention efect (se, e.g., Rizi 1990, de Swart 1992, Pesetsky 2000). For instance, 
the licensing of polarity items exhibits sensitivity with respect to the material occurring 
betwen the licensor and license. In (35), a referential expresion Mary does not disrupt 
the relation betwen negation and a polarity item a red cent, while a quantifier every 
charity does. Den Dikken & Giannakidou (2002) argue that with respect to the wh-the-
hel phrases, a quantifier who intervenes betwen the Q operator and the wh-the-hel 
phrase. Hence, this is an intervention efect.
65
 
 
Intervention Efect of Polarity Items 
(35) a.  John didn?t give Mary a red cent. 
 b. * John didn?t give every charity a red cent. 
(36) a. *[ Neg ? [every charity ? [a red cent ? ]] 
 b. *[ Q ? [who  ? [who the hel  ]] 
In the embedded clause, as in (34b), the higher wh-phrase who should be in [Spec,CP], 
which is higher than a Q operator. In this case, what is intervening betwen the Q 
operator and the wh-the-hel phrase is the trace of who.
6
 
                                                
65
 Norbert Hornstein (p.c.) notes that it is not common for an indefinite quantifier to act 
as an intervener in intervention efects noted in the literature, providing the following 
example: 
(i) John doesn?t know when anyone is at home. 
6
 There is some concern whether the prohibition of in-situ wh-the-hel phrases can be 
acounted for in this analysis. In particular, example (ic) is predicated to be aceptable. 
First, acording to their analysis, the in-situ wh-the-hel phrase in (ic) is c-commanded by 
a Q operator in the embedded clause. Then there are some indications that whether does 
not function as an intervener. The aceptable example (ib) suggests that whether does not 
block the licensing requirement of anyone, which is a polarity item. If we asume that the 
set of items functioning as an intervener for wh-the-hel phrases is the same for anyone, 
then the wh-the-hel phrase in the same position should be licensed. 
 109 
 
 
2.5.2 Lack of Scope Interaction 
The second property that fals within the intervention efect apparoach is the lack of 
scope interaction. It has been noted in the literature that example (37a) is ambiguous with 
respect to the scope relation betwen the wh-phrase and a quantifier everyone (cf. May 
1985, Lasnik & Saito 1992, among many others). In one interpretation, the wh-phrase 
takes wide scope with respect to everyone, everyone bought the same thing for Max. For 
example, Susan bought a Betle?s CD, Mary bought another Betle?s CD, etc. In the 
other interpretation, everyone takes wide scope with respect to the wh-phrase. One 
expected situation is that Susan bought a Betle?s CD, Mary bought a pasta machine, and 
Lisa bought a pair of socks. Den Dikken & Giannakidou (2002) observe that such a scope 
ambiguity is not available in (37b), in that the wh-the-hel phrase must take wide scope 
with respect to everyone. Their acount of this observation is that if the wh-the-hel 
phrase is reconstructed to a position that is lower than everyone, which sems to be 
required for the wide scope universal quantifier interpretation, the representation exhibits 
intervention efects. As can be sen in (38b), the universal quantifier c-commands the 
wh-the-hel phrase and the Q operator c-commands the universal quantifier at the same 
time. 
 
(37) a. What did everyone buy for Max? [ambiguous] 
 b. hat the hel did everyone buy for Max? [unambiguous] wh>every 
                                                                                                                                            
(i) a. Who wonders whether Mary met who? 
 b. ho wonders whether ary met anyone? 
 c. *Who wonders whether Mary met who the hel? 
The second indication that whether is not a harmful intervener comes from the following 
examples. Asuming that a predicate like know does not host a Q operator, as they argue 
in their paper, the wh-the-hel phrase in (ia) is licensed by a Q operator that appears 
somewhere to the left of wonder. Again, the aceptable status of (ia) indicates that 
whether does not function as an intervener. 
(i) a. I wonder whether Mary knows who the hel left. 
 b. I wonder whether ary knows anyone left. 
If the above consideration holds, we have to sek some alternative acount for the 
prohibition of in-situ wh-the-hel phrases. 
Furthermore, with respect to the status of who as an intervener, it sems that a 
regular wh-phrase does not function as an intervener with respect to licensing of any. As 
shown in (iia), the polarity item anything is licensed even though who and its trace 
intervene betwen a Q operator and anything, as den Dikken & Giannakidou (2002) note 
in their paper. They claim that this fact indicates that wh-the-hel phrases are a new kind 
of polarity item, and the set of licensors for wh-the-hel phrases does not completely 
overlap that of other polarity items such as anything. 
(ii) a. Who has given anything to Bil? (dDG (48)) 
 b. hat did John give to anyone? 
 
 110 
(38) a.  [ Q ? [what the hal ? [everyone ? ]] 
 b. * [ Q  [everyone ? [what the hel ? ]] 
 
 
3. Exclamatives 
3.1 Complementary Distribution 
In the previous sections, we reviewed the analysis proposed by den Dikken & 
Giannakidou (2002) concerning the distribution of wh-the-hel phrases. In this section, 
we argue that a similar acount is potentialy available for the distribution of exclamative 
wh-phrases. We propose that exclamative wh-phrases are a kind of polarity item, the 
distribution of which is regulated by the presence of an appropriate licensor. In the case 
of wh-the-hel phrases, we have sen that the licensors include negation, a Q operator, 
interogative predicates, and some others. We show that the distribution of exclamative 
wh-phrases is limited to a position where a Factive operator or a factive predicate can c-
command them. Furthermore, we provide some data that ilustrate the intervention efect 
as we have sen in the case of wh-the-hel phrases. First, the following data show that in 
the embedded context, the distribution of wh-the-hel phrases and exclamative wh-
phrases is complementary. This makes sense since it has been shown that wh-the-hel 
phrases, being a polarity item that is sensitive to non-veridicality, cannot occur in a 
context where factivity is at work. On the other hand, if exclamative wh-phrases are 
polarity items that are sensitive to the presence of factivity, it is expected that 
exclamative wh-phrases are not alowed to occur where wh-the-hel phrases can occur 
(except for matrix exclamative wh-phrases; we wil come back to this soon). 
 
(39) a. I know who left. 
 b. *I know who the hel left. 
 c. I know what a big house he has. 
(40) a. I don?t know who left. 
 b. I don?t know who the hel left. 
 c. *I don?t know what a big house he has. 
(41) a. I wonder who left. 
 b. I wonder who the hel left. 
 c. *I wonder what a big house he has. 
(42) a. It is amazing who came. 
 b. *It is amazing who the hel came. 
 c. It is amazing what a big house he has. 
(43) a. It is unclear who came. 
 b. It is unclear who the hel came. 
 c. *It is unclear what a big house he has. 
One piece of evidence for the analysis proposed here comes from the availability of long-
distance licensing of exclamative wh-phrases noted in the introduction. The minimal pair 
in (44) shows that although exclamative wh-phrase cannot usualy appear under a non-
factive predicate such as be unclear, the example becomes aceptable if the whole 
 111 
sentence is further embedded under a factive predicate like realize. Asuming that factive 
predicates require the complement to have a factive operator (presumably the predicate 
and the factive CP are sisters), this factive operator c-commands the exclamative wh-
phrase in (44a). On the other hand, there is no factive operator or factive predicate in 
(44b). 
 
Long Distance Licensing 
(44) a.  Mary regrets/realizes that it is unclear how very big a house he has. 
 b. * ary thinks/says/claims that it is unclear how very big a house he has. 
There is one point we want to make sure of before we proced. Recal that there is 
an ordering restriction betwen the fronted wh-phrase and the fronted topic phrase. Also, 
what is significant is the asymmetry betwen the matrix and the embedded clause. As we 
have sen with regular wh-phrases and wh-the-hel phrases, exclamative wh-phrases obey 
exactly the same restriction as shown in (45) and (46). In the matrix clause, the topic 
phrase must precede the exclamative wh-phrase, and in the embedded clause, the topic 
phrase must follow the exclamative wh-phrase. 
 
Relative Order with Topic Phrases 
(45) a. ?To a jerk like him, what an expensive present you gave! 
 b. *What an expensive present, to a jerk like him, you gave! 
(46) a. ?It?s awful what an expensive present, to a jerk like him, you gave. 
 b. *It?s awful to a jerk like him, what an expensive present you gave. 
The observation above leads us to conclude that, like other wh-phrases in English, the 
landing sites of wh-movement is diferent depending on whether it is the matrix or 
embedded clause. In the matrix clauses, the fronted exclamative wh-phrase moves to 
[Spec,FocP], while in the embedded clauses, the fronted exclamative wh-phrase moves to 
[Spec,CP]. 
 
[Spec, FocP] in the Root Clause, [Spec, CP] in the Embedded Clause 
(47) a.   [
CP
 ?  [ C [
TopP
 [ Top [
FocP
 Wh-Exc [ Foc [
IP
 ?  
 b. ? realize [
CP
 Wh-Exc [ C [
TopP
 [ Top [
FocP
 ?  [ Foc [
IP
 ? 
 
3.2 Properties of Exclamative Wh-Phrases 
In this section, we show how the proposed analysis acounts for some basic facts about 
exclamative wh-phrases. First, we argue that matrix exclamative wh-phrases can be 
licensed by a factive operator in C, which c-commands the exclamative wh-phrases, as 
shown in (48). Note that this follows if we asume that the landing site of the fronted 
exclamative wh-phrase in the matrix clause is not [Spec,CP], but [Spec,FocP]. 
 
FACT Operator in C 
(48) a. What a big house he has! 
 b. [
CP
 C
FACT
 [
FocP
 [what a big house]
i
 [ Foc [
IP
 he has t
i 
]] 
 112 
Next, when exclamative wh-phrases show up in the embedded clause, especialy 
when the exclamative wh-phrase appears in a clause imediately dominated by a factive 
predicate as in (49a), the exclamative wh-phrase is licensed by the factive predicate, since 
the predicate c-commands the exclamative wh-phrase. In such a case, obviously, the 
factive operator cannot c-command the exclamative wh-phrase, and an alternative 
licensor would be the factive predicate itself.
67
 
 
Embedded under Factive Predicates 
(49) a.  It is amazing what a tal man Mary met! 
 b. * It is unclear what a tal man ary met! 
(50) It is amazing [
CP
 [what a big house]
i
 C
FACT
 [
FocP
 Foc [
IP
 he has t
i 
]] 
As a diference betwen interogatives and exclamatives, Eliott (1974) observes 
that polarity items such as anything are not licensed, as in (51). Under our acount, it is 
possible to acommodate this fact very easily. In order to license the exclamative wh-
phrase, there is a factive operator in C. Then, since anything has to appear in a non-
veridical context, it can never satisfy its requirement in exclamatives. Also, we can 
conclude from (51) that in the structure there is no appropriate licensor for a negative 
polarity item. More specificaly, there is no Q morpheme or interogative predicates that 
would license a negative polarity item. 
 
No NPI licensing 
(51) a. *What a rich man has given anything to Bil! (Eliott 1974) 
 b. *What a cheap present John gave to anyone! 
Finaly, a smal discussion on the long distance licensing of exclamative wh-
phrase sems to be in order. We have shown that exclamative wh-phrases can show up 
under a non-factive predicate as long as a factive predicate appears somewhere higher in 
the structure. One important point in such an example sems to be the status of the 
intervening predicate, and predicates such as be unclear can be employed to show that 
long distance licensing is posible. Note that examples are not aceptable when 
interogative predicates such as wonder are used, as shown in (53). This observation, 
nonetheles, does not undermine the current proposal; it is quite plausible that what is 
wrong with the example is not the licensing of exclamative wh-phrases, but a requirement 
of interogative predicates. That is, in the example, the requirement that an interogative 
predicate must have an interogative wh-clause as a complement is not satisfied. A 
similar analysis can be provided for example (54). Again, the example is unaceptable, 
                                                
67
 Alternatively, we could say that there is a factive operator OPfact in [Spec,CP]. Under 
such an approach, the exclamative wh-phrase in the matrix clause is c-commanded by the 
factive operator OPfact. Furthermore, in the embedded clause, both a factive operator and 
an exclamative wh-phrase are in [Spec,CP], asuming multiple specifiers in CP. It could 
be that the exclamative wh-phrase is located in the inner specifier of CP, where it can be 
c-commanded by the factive operator; or the factive operator is in the inner specifier, but 
it c-commands the exclamative wh-phrase that is in the outer specifier, under certain 
asumptions of c-command. 
 113 
but what is wrong is a requirement of the predicate think, that it is unable to take a wh-
clause. 
 
Long Distance Licensing 
(52) a.  Mary realizes that it is unclear what a big house he has. 
 b. * ary claims that it is unclear what a big house he has. 
(53) *Mary realizes that John wonders what a big house his parents have. 
(54) *Mary realizes that John thinks what a big house his parents have. 
 
 
3.3 The Polarity-Based Aproach 
One empirical consequence of the analysis proposed by den Dikken & Giannakidou 
(2002) is that they can acount for some quite mysterious properties of wh-the-hel 
phrases as instances of intervention. We would like to perform a similar task in this 
section, that is, some properties of exclamative wh-phrases are acounted for if viewed as 
instances of intervention. 
First, it has been observed that exclamative wh-phrases are never good if they are 
left in situ (Obenauer 1994, Nelson 1997, Ono 2002, Oda 2002). There is no diference 
regardles of whether they are in the matrix or embedded clause. 
 
No Multiple Exclamatives / No In-situ Exclamative Wh-phrase 
(55) a. *What an expensive present you gave to what a stupid boyfriend! 
 b. What an expensive present you gave to such a stupid boyfriend! 
Same Observation in the Embedded Clause 
(56) *John realizes what an expensive present you gave to what a stupid boyfriend! 
If we asume that an exclamative wh-phrase acts as an intervener, we can acount for the 
observation above as an instance of intervention. Although a factive operator in C c-
commands the in situ exclamative wh-phrase, there is another exclamative wh-phrase (or 
a trace of it) intervening betwen the factive operator and the in situ exclamative wh-
phrase, as shown in (57). 
 
No In-Situ Wh-Exclamatives 
(57) a. *What an expensive present you gave to what a stupid boyfriend! 
 b. [
CP
 . [ C
FACT
 [
FocP
 Wh-Exc [ Foc [
IP
 ? Wh-Exc ]]] 
    |--------*-------------| 
The second observation in which intervention efects are at work is the lack of 
scope interaction with exclamative wh-phrases. As sen in wh-the-hel phrases, the 
fronted exclamative wh-phrase always takes higher scope with respect to a quantifier in 
the example of the sort in (58). This observation can be acounted for by asuming that a 
quantifier everyone acts as an intervener if the exclamative wh-phrase is reconstructed 
below the quantifier in order to take the lower scope with respect to the quantifier. 
 
 114 
No Scope Interaction 
(58) a. What an expensive present everyone bought for Max! [unambiguous] 
 b. hat did everyone buy for Max?   [ambiguous?] 
No Scope Interaction 
(59) a.  [ C
FACT
 ? [ Wh-Exc ? [everyone ? ]] 
 b. * [ C
FACT
  [everyone ? [ Wh-Exc ? ]] 
      |--------*---------| 
 
3.4 An Alternative 
This chapter started by pointing out that c-command is the crucial ingredient in licensing 
exclamatives, rather than selection. This raises an interesting question with respect to the 
driving force of the movement of the exclamative wh-phrase (Norbert Hornstein p.c.). 
Asume that selection is a relation betwen the heads. In our current case, an 
interogative taking verb selects a head of CP with a wh-feature. It can be posited that the 
interogative wh-phrase wil move to [Spec,CP] of that CP, to check some feature on the 
head. Since the feature on the head C has to be checked, an interogative wh-phrase has 
to move to the [Spec,CP] that is local to the interogative taking verb. On the other hand, 
if it is true that exclamative clauses are licensed not by selection but by c-command, the 
relation betwen the factive verb and the complement CP (and its head) does not have to 
involve wh-features. Adopting the insight by Kiparsky & Kiparsky (1970), it is possible 
to asume that the factive verb selects a complement CP with a feature that has something 
to do with factivity, which in turn might necesitate the presence of the factive operator in 
its specifier position. Again, our observation so far tels us that the wh-feature does not 
have to be involved since exclamatives are not licensed by selection and the sentential 
complement for the factive predicate can be a that-clause. Given that, we can ask why the 
exclamative wh-phrase has to move to [Spec,CP]. In the case of interogative wh-phrases, 
an interogative taking predicate requires its complement to bear a wh-feature, the wh-
phrase moves to [Spec,CP] since it can check the feature of C by this movement. On the 
other hand, in the case of exclamative wh-phrases, the head of CP that is selected by a 
factive predicate does not bear a feature that has to be checked by movement of the wh-
phrase. Nonetheles, it looks like the exclamative wh-phrase moves to [Spec, CP]. 
One possible suggestion is that the exclamative wh-phrase moves since the wh-
phrase itself has a feature that has to be checked in [Spec,CP]. The derivation would 
crash if such a wh-phrase remains in situ, so the only possible way to derive a 
gramatical structure is to move the wh-phrase. We could speculate that this suggestion 
that the wh-phrase itself has a feature is tied with the intuition that exclamative wh-
phrases are loaded with discourse-related information, or it involves the speaker?s 
judgment / point of view. If such a property could be linked to the suggestion that the wh-
phrase itself has a feature to be checked, it sems possible to say something similar for 
the wh-the hel phrases (se some related discussion in Ochi (2004)). 
 
 
 115 
4. Conclusion 
This chapter aims at explaining two puzzles in exclamatives. The first puzzle is that while 
predicates exist that take only an interogative complement, there are no ?exclamative 
predicates? that take only exclamatives as their complements. The second puzzle is that 
while embedded interogative clauses have to occur as a sister to a interogative predicate, 
embedded exclamative clauses can be licensed by a non-local factive predicate. The 
answer to the two puzzles is the licensing mechanism of exclamative wh-phrases. Unlike 
interogative wh-phrases, exclamative wh-phrases are licensed not by selection by a 
predicate, but by c-command. This acounts for the observation that long distance 
licensing is possible in exclamative wh-phrases. We proposed that there are at least two 
elements, a factive predicate and a factive operator that can license exclamative wh-
phrases. This property of exclamative wh-phrases is paralel to the properties of wh-the-
hel phrases noted by den Dikken & Giannakidou (2002). Acording to their analysis, 
wh-the-hel phrases are polarity items and they have to be c-commanded by a non-
veridical item, such as an interogative predicate, negation, or a Q operator. With respect 
to exclamative wh-phrases, on the other hand, the set of licensors includes factive 
predicates and factive operators. Then, it is plausible to claim that exclamative wh-
phrases are also some kind of polarity item that is sensitive to a subset of veridical items. 
One consequence of the current analysis is that a system employing some feature E (as in 
exclamative) in a semantic selection frame is shown to be inappropriate. The 
investigation conducted in this paper shows that the licensing mechanism of exclamative 
wh-clauses is diferent from that of interogative wh-clauses: the former involves c-
command, and the later selection. Furthermore, the analysis here has the advantage of 
explaining some quite mysterious facts of exclamatives by regarding them as 
intervention efects. 
 
 
 
 
 116 
 
CHAPTER 6: PROCESING OF JAPANESE EXCLAMATIVES 
 
 
1. Introduction 
In this chapter, the proces of dependency formation in sentence procesing is 
investigated, using experimental data from Japanese interogatives and exclamatives. 
Research on dependency formation is quite abundant in the sentence procesing 
literature; much research has been done using, for example, fronted clause-initial 
interogative wh-phrases in English and related languages. Since the distance betwen the 
fronted wh-phrase (filer, what in (1) and the position where its non-wh variant would 
have occurred (gap, the position marked by ?__? in (1) is potentialy infinite, it creates a 
masive ambiguity in the course of sentence procesing. For instance, from a piece of the 
sentence in (2), there is no way for speakers of English to know where the exact position 
of the gap is beforehand; it could be found in the matrix clause as in (1a), in the 
embedded clause as in (1b), or somewhere further down in the structure. A research 
question that has been asked on numerous occasions is how the parser asociates the filer 
with the gap. 
 
(1) a. What does John believe __ ? 
 b. hat does John believe that Mary ate __ ? 
(2) What does John ? 
 
The Filed Gap Efect 
Using the paradigm shown in (3), Stowe (1986) observed a longer reading time at us in 
the embedded clause in (3b), compared with (3a). 
 
(3) a. My brother wants to know if Ruth wil bring us home to Mom at 
  Christmas. 
 b. My brother wants to know who Ruth wil bring us home to __ at 
  Christmas. 
The longer reading time at us in (3b) is interpreted as the parser initialy trying to 
interpret the fronted wh-phrase who to be a direct object of the verb bring, which would 
asign a theta-role to the fronted wh-phrase. Since the presence of an overt lexical item 
(i.e., us) reveals that the parser?s prediction is wrong, the reading time increases at that 
position, possibly due to the parser revising its initial prediction. It was argued that the 
results show that the parser creates a gap as soon as it encounters the filer, establishing a 
dependency (se related work, such as Crain & Fodor, 1985; de Vincenti, 1991). 
 
Typing Mismatch Efect 
A similar conclusion has been reached in the research using languages such as Japanese, 
in which wh-phrases need not be fronted, and can remain in the canonical position. 
Miyamoto & Takahashi (2002) showed that the presence of the in situ wh-phrase in 
Japanese (i.e., wh-phrases remaining in the canonical position) initiates a search for the 
licensing particle ka at the earliest gramaticaly possible position. By comparing the 
 117 
reading times of the embedded verb position as shown in (4), they observed that the 
reading time of the embedded verb region in (4b) is longer than that in (4a). They cal this 
the Typing Mismatch Efect (TME). Crucialy, they showed that the reading time of the 
embedded verb with the declarative complementizer is faster than that of the embedded 
verb with the Q-particle ka when not preceded by a wh-phrase. 
 
(4) a. senmu-wa kakarityoo-ga dono pasokon-o tukateiru-ka ita 
  supervisor-TOP director-NOM which computer-ACC use-Q said 
  ?The supervisor said which computer the manager uses.? 
 b. senmu-wa kakarityoo-ga dono pasokon-o tukateiru-to ita-no 
  supervisor-TOP director-NOM which computer-ACC use-COMP said-Q 
  ?Which computer did the supervisor say that the manager uses?? 
Note that having the licensing particle ka in the clause with the wh-phrase is not a 
requirement of the gramar, as is clear from the fact that (4b) is gramatical. In either 
(4a) or (4b), the licensing particle ka c-commands the wh-phrase, which is the 
gramatical requirement of the wh-phrase. Therefore, it is argued that the reading time 
contrast observed above, the TME, arises due to a bias by the parser to satisfy the 
gramatical requirement of the wh-phrase as soon as possible. 
 
The Filed Gap Efect in Japanese 
Investigating fronted wh-phrases in Japanese, Aoshima, Philips & Weinberg (2004) 
provide another piece of evidence that the parser prefers to establish a dependency as 
soon as possible. It was shown that the dative-marked sentence initial wh-phrase is 
interpreted in the most deeply embedded clause in a sentence that involves more than one 
clause. Using the Filed Gap Efect paradigm, they observed that the reading time of the 
dative NP (kootyoosense-ni ?principal-DAT?) in the embedded clause is longer in (5a), 
compared to (5b). 
 
(5) a. dono sinnyuusei-ni tannin-wa  sisyo-ga 
  which new student-DAT clas teacher-TOP librarian-NOM  
  kootyoosense-ni mangabon-o susumeta-to imasita-ka 
  principal-DAT comic book-ACC recommended-COMP said-Q 
  ?To which student did the clas teacher say that the librarian recommended  
  the comic book to the principal?? 
 b. dono sinnyuuse-ga tannin-ni  sisyo-ga 
  which new student-NOM clas teacher-DAT librarian-NOM  
  kootyoosense-ni mangabon-o susumeta-to imasita-ka 
  principal-DAT comic book-ACC recommended-COMP said-Q 
  ?Which student saidl to the clas teacher that the librarian recommended  
  the comic book to the principal?? 
They argued that the reading time of the embedded dative NP increases because the 
reader tries to interpret the fronted dative-marked wh-phrase in the embedded clause. 
Since the position is already filed with another phrase, the reader has to revise his initial 
expectation. 
This apparent long distance asociation betwen the sentence initial wh-phrase 
and the embedded verb can be acounted for by saying that the embedded verb is the first 
 118 
available item that can asign a theta-role to the wh-phrase or license the wh-feature by 
the clause-final Q-particle. By positing a gap in the most deeply embedded clause, it is 
possible for the embedded verb to asign a theta role to the wh-phrase, establishing the 
dependency with respect to the theta role or wh-feature. Note that interpreting the fronted 
wh-phrase in the embedded clause is not driven by the gramar, i.e., a legitimate 
structure can be constructed with the dative-marked wh-phrase interpreted at the matrix 
clause. Thus, the results ofer evidence that the parser is biased toward establishing a 
dependency as soon as possible. 
 
Procesing Exclamatives 
Below, we present two kinds of experimental data on exclamatives. Exclamatives are 
quite useful for investigating the interaction betwen the parser?s mechanism for 
establishing the dependencies and its mechanism for building the linguistic representation. 
Acording to the syntactic nature of exclamative sentences in Japanese, exclamative 
sentences require the parser to build very intricate functional projections in the CP layer. 
In exclamatives, the head of the functional projection is overtly realized as a verbal suffix, 
no-da-roo, which we treat as a head of functional projection Exclamative Phrase. 
First, we show that Typing Mismatch Efects are observed with exclamatives, 
suggesting that the presence of the exclamative wh-phrase initiates the readers? 
expectation that the appropriate licensing particles occurs at the earliest gramaticaly 
available position. Second, we show that the sentence initial dative-marked exclamative 
and interogative wh-phrases are handled diferently. Our results, obtained from an off-
line study show that unlike its interogative counterpart, we have no evidence that the 
parser predominantly interprets the sentence initial exclamative wh-phrases in the most 
deeply embedded clause. Contrasting the results obtained with the sentence initial 
interogative wh-phrases, we provide one potential acount for the results of exclamative 
wh-phrases. Under this acount, once-established dependency of the exclamative wh-
phrase with the elements in the matrix clause was not changed even though it is in 
principle possible to re-establish the dependencies to satisfy the requirements as soon as 
possible. Se Ono, Yoshida, Aoshima, & Philips (to appear) for discussion of an 
alternative acount. 
In the folowing section, we first discuss various syntactic properties of Japanese 
exclamatives relevant for the current study. Then, we look at one off-line sentence 
completion experiment and one on-line self-paced reading experiment. In the first 
sentence completion experiment, we show that the in situ exclamative wh-phrases in the 
embedded clause receive completion with the local licensing particle as often as their 
interogative counterparts do. This suggests that the strong local licensing requirement of 
the parser can be observed in the sentence generation task. Then, we show that the 
sentence initial dative-marked exclamative wh-phrases are treated diferently from their 
interogative counterparts. These results reveal that while the sentence initial 
interogative wh-phrases receive more completion with licensing particles in the 
embedded clause than in the matrix clause, the sentence initial exclamative wh-phrases 
receive completions with licensing particles in the matrix clause. Finaly, we show that 
the TME is observed with in situ exclamative wh-phrases by using an on-line self-paced 
(moving window) reading experiment. 
 
 119 
 
2. Syntax of Japanese Exclamatives 
The syntax of exclamatives in Japanese is to some extent similar to, and to other extents 
diferent from, the much-studied syntax of interogatives. They are similar to each other 
to the extent that both involve wh-phrases and certain particles as their licensors. In this 
section, we review some syntactic properties of exclamatives and interogatives relevant 
for the purpose of this study. 
 
2.1 Exclamative Wh-Phrase 
It is wel known that interogative wh-phrases are licensed by the so-caled Q-particle ka 
that appears as a verbal suffix. There is an exclamative counterpart; the exclamative 
particle no-da-roo licenses the exclamative wh-phrase nante.
68
 
 
Interogatives 
(6) John-wa [dono ookina piza]-o tabemasita-ka 
 John-TOP  which big piza-ACC ate-Q 
 ?Which big piza did John eat?? 
Exclamatives 
(7) John-wa [nante ookina piza]-o tabeta-no-da-roo 
 John-TOP  WH-EXC big piza-ACC ate-EXC 
 ?What a big piza John ate!? 
Another constraint is that the appropriate licensor has to c-command the base-generated 
position of the wh-phrase (se Harada (1972) for discussion about interogative wh-
phrases). Examples of exclamatives such as (8b) are ungramatical since the exclamative 
particle no-da-roo in the embedded clause never c-commands the subject of the matrix 
clause. On the other hand, (8a) is gramatical since the exclamative particle c-commands 
the matrix subject. 
 
C-command Requirement 
(8) a.  nante takusan-no hito-ga [sono gakuse-ga mondai-o  
   wh-EXC many-GEN man-NOM  that student-NOM problem-ACC 
   toita-to]  omotta-no-da-roo 
   solve-COMP thought-EXC 
   ?What a lot of people thought that the student solved the problem!? 
 b. * nante takusan-no hito-ga [sono gakuse-ga mondai-o  
   wh-EXC many-GEN man-NOM  that student-NOM problem-ACC 
   toita-no-da-roo-to] omotta 
   solve-EXC-COMP thought 
                                                
68
 The particle used in exclamatives tends to be varied and undergoes quite robust 
morphological reduction, depending on the gender of the speaker and the register of the 
discourse. The form no-da-roo is taken to be the standard form. Se Fuji & Ono (2006) 
and Chapter 2 and 3 in this disertation for the details. 
 120 
Extending the above noted observation, Saito (1989) shows that it is possible to front an 
interogative wh-phrase to the position that is outside the c-commanding domain of the 
Q-particle as long as the base-generated position of the wh-phrase is c-commanded by the 
Q-particle. In (9), the base-generated position of the fronted interogative phrase (marked 
with ?t?) is in the embedded clause and is c-commanded by the Q-particle at the end of 
the embedded clause. 
 
(9) dono hon-o John-wa [Mary-ga tosyokan-kara  t karita-ka] 
 which book-ACC J-TOP  -NOM library-from borrowed-Q 
 siritagateiru 
 wonder 
 ?John wonders which book Mary borrowed from the library.? 
Note that the fronted interogative wh-phrase is interpreted in the scope of the embedded 
clause, since the Q-particle appears in the embedded clause; hence the example is not a 
matrix question. 
We can construct similar examples with exclamatives. Consider the following 
examples. In (10), the exclamative wh-phrase appears in the canonical position where it 
is c-commanded by the exclamative particle no-da-roo. On the other hand, in (11), the 
exclamative wh-phrase is fronted to the sentence initial position.
 69
 
 
(10) John-wa [Mary-ga [nante takusan-no gakuse]-ni A-o 
 J-TOP  -NOM  wh-EXC many-GEN student-DAT A-ACC  
 ageta-no-da-roo-to] akireteita 
 gave-EXC-COMP amazed 
 ?John was amazed what a lot of students Mary gave an A to.? 
(11) [nante takusan-no gakuse]-ni John-wa [Mary-ga t A-o  
  wh-EXC many-GEN student-DAT J-TOP  -NOM A-ACC 
 ageta-no-da-roo-to] akireteita 
 gave-EXC-COMP amazed 
Similar to the examples with interogative wh-clauses, the fronted exclamative wh-phrase 
in (11) is interpreted in the embedded clause, since the exclamative particle determines 
the scope of the exclamative wh-phrase. 
A further similarity betwen interogatives and exclamatives is shown by the fact 
that the distance betwen wh-phrases and the particles can be non-local. In other words, 
they do not have to be in the same clause. In (12) and (13), wh-phrases appear in the 
embedded clause and the particles in the matrix clause. This data demonstrates the 
similarities betwen exlcamatives and interogatives. However, in the next section, I wil 
show a few diferences. 
 
                                                
69
 Se Ono et al. (to appear) for discussion of the environment where long distance 
scrambling is alowed in exclamatives. 
 121 
Long Distance Dependency 
(12) Mary-wa [John-ga dono ookina piza-o tabeta]-to omoimasita-ka 
 ary-TOP  John-NOM which big piza-ACC ate-COMP thought-Q 
 ?Which big piza did ary think that John ate?? 
(13) Mary-wa [John-ga nante ookina piza-o tabeta]-to omotta-no-da-roo 
 ary-TOP John-NOM wh-EXC big piza-ACC ate-COMP thought-EXC 
 ?What a big piza ary thought that John ate!? 
So far, we have sen a few similarities betwen interogatives and exclamatives. Wh-
phrases in both constructions require particular licensing particles, and the licensing 
mechanism is also quite similar. 
 
2.2 Exclamative Particle 
A diference emerges in the cases where the particles appear in an embedded clause. 
When a wh-exclamative clause is embedded, the clause has to be followed by the 
complementizer to as shown in (14). On the other hand, as sen in (15), when wh-
interogative clauses are embedded, there are verbs that alow the clause to appear 
without the complementizer to. Furthermore, (16) ilustrates that there are verbs such as 
siritagaru ?wonder? that requires the clause to appear without the complementizer to. 
 
(14) Mary-wa [John-ga nante ookina piza-o tabeta]-no-da-roo-*(to) 
 ary-TOP  John-NOM wh-EXC big piza-ACC ate-EXC-COMP 
 omotta 
 thought 
 ?Mary thought, ?What a big piza John ate!? ? 
(15) Mary-wa [John-ga dono ookina piza-o tabeta]-ka-(to) kiita 
 ary-TOP  John-NOM which big piza-ACC ate-Q-COMP asked 
 ?Mary asked which big piza John ate. 
(16) Mary-wa [John-ga dono ookina piza-o tabeta]-ka-(*to) 
 ary-TOP  John-NOM which big piza-ACC ate-Q-COMP  
 siritagateita 
 wondered 
 ?Mary wondered which big piza John ate. 
This observation can be captured by asuming that the Q-particle ka occurs in the Comp 
position where the complemtizer to occurs in the embedded clause (se the diagrams 
below). On the other hand, the exclamative particle no-da-roo occurs in a position lower 
than the Comp position. 
 
 122 
(17) a. CP structure for exclamatives  b. CP structure for interogative 
      VP          VP  
       u        u  
      CP    V       CP  V 
    r        r  
 ExclamativeP  Comp to       IP  Comp to / Q ka 
  u        6   
   IP    Exclamative nodaroo   ? dono ?   
     6         
    ? nante ?         
The exclamative particle no-da-roo is the head of Exclamative Phrase, which is 
dominated by CP (se (17a). The Q-particle ka appears where the Comp to occurs 
otherwise (se (17b). The point is that having the Q-particle ka does not have to add any 
extra functional projection specificaly for it, while there has to be an extra functional 
projection in the structure in order to project the exclamative particle no-da-roo. 
To sum up, we saw that exclamative wh-phrases have similar requirements as 
interogative wh-phrases have: namely the c-comand requirement by the appropriate 
licensing particle. Yet, the diferences betwen wh-exclamatives and wh-interogatives 
arise when diferent properties of the licensing particles are investigated. The Q-particle 
ka occupies the position where the Declarative complementizer to occupies otherwise. On 
the other hand, the Exclamative particle no-da-roo occurs at a position lower than the 
Declarative complementizer. Below, we designed two experiments ilustrating how the 
syntactic similarities and diferences have efects on the procesing of dependency 
formation by the parser. 
 
 
3. Experiment 1: Sentence Completion Task 
An off-line sentence fragment completion task was designed to investigate what native 
speakers of Japanese predict, given a sentence fragment that strongly indicates that it is 
part of an exclamative sentence. The aim was to investigate whether the in situ 
exclamative wh-phrases are licensed in the embedded clause or in the matrix clause. The 
results obtained from Experiment 3 in Aoshima, et al (2004) showed that the sentence 
fragments with in situ interogative wh-phrases with multi-clausal structures were 
completed with the licensing particle in the embedded clause, although putting the 
licensing particle only in the matrix clause is also gramaticaly alowed. Similarly, 
completing the sentence with the licensing particle in the matrix clause is also a 
gramaticaly alowed option for the in situ exclamative wh-phrases. If the presence of 
the in situ exclamative wh-phrase creates the expectation of the local existence of the 
licensing particles, as the interogative counterpart does, the relevant sentence fragments 
wil be completed with licensing particles at the embedded clause. The strength of the 
local bias would indicate that in situ exclamative wh-phrases are similar to the 
interogative wh-phrases in that both create an expectation that the licensing particles 
occur in the local clause. 
A second aim was to determine where the sentence initial exclamatives are 
interpreted. The results obtained from Experiment 3 in Aoshima, et al (2004) showed that 
 123 
sentence fragments with the dative-marked sentence initial interogative wh-phrases were 
predominantly interpreted in the embedded clause, measured by the number of 
completions with the licensing particles in the embedded clause and by the inspection of 
the argument structure of the embedded predicates. If the sentence initial exclamative 
wh-phrases are treated as interogative wh-phrases are, the sentence fragments with the 
sentence initial exclamative wh-phrases are completed with the licensing particles at the 
embedded clause. Also the inspection of the argument structure of the embedded 
predicate shows that the embedded predicates should be compatible with the dative 
marked NP. 
 
3.1 Procedures 
Forty-two undergradutate native speakers of Japanese participated in the experiment. 
Experimental materials consisted of thirty sets of sentence fragments with six conditions 
each (se Appendix A for the complete list of items). The thirty sets of items were 
distributed among six lists in a Latin Square design. Each participant saw exactly one of 
the lists intermixed with sixty unrelated filers in a random order.
70
 
Target items consisted of thre NPs, each of which was marked by diferent case 
markers (topic, nominative, and dative). Target items were manipulated by two factors; 
the type of the wh-phrase (exclamative vs. interogative), and by the position of the wh-
phrase (in-situ vs. fronted). (18) ilustrates an example set. 
 
(18) Sample set of experimental conditions for Experiment 1 
 a. Exclamative / In situ (E/IS) 
  sono sense-wa gakuse-ga nante takusan-no onnanoko-ni .. 
  that teacher-TOP student-NOM wh-EXC many-GEN girl-DAT 
 b. Exclamative / Frotned (E/F) 
  nante takusann-no onnanoko-ni sono sense-wa gakuse-ga .. 
  wh-EXC many-GEN girl-DAT that teacher-TOP student-NOM 
 c. Interogative / In Situ (W/IS) 
  sono sense-wa gakuse-ga dono onnanoko-ni .. 
  that teacher-TOP student-NOM which girl-DAT 
 d. Interogative / Fronted (W/S) 
  dono onnanoko-ni sono sense-wa gakuse-ga .. 
  which girl-DAT that teacher-TOP student-NOM 
In Exclamative conditions, a dative-marked exclamative wh-phrase was used. In In-Situ 
conditions, the dative-marked NP appeared after a topic-marked NP and a nominative-
marked NP. In the Fronted conditions, the dative-marked NP appeared at the beginning 
of the sentence fragment. In Interogative conditions, an interogative wh-phrase was 
used instead of an exclamative wh-phrase. In order to complete the sentence fragments, 
participants had to supply at least two predicates, since there is no predicate that can take 
                                                
70
 Although the target items contained six conditions, we eliminated two conditions from 
the analysis; the eliminated conditions included just an exclamative wh-phrase nante 
wituout any adjective or adverb. Unfortunately, those sentence fragments were apparently 
misinterpreted as a colloquial form of ?what name of N?. Thus, in the following sections, 
the results and analyses are based on twenty sets of items with four conditions each. 
 124 
al thre NPs (case-marked as such) as arguments. Optionaly, they can add an acusative 
marked NP as an argument of the embedded clause. In addition, they must use a licensing 
particle (for exclamative or interogative) in order to license a wh-phrase. The experiment 
lasted about 45 min. 
 
3.2 Results 
3.2.1 Task Acuracy 
Based on the forty-two participants, 840 responses (42 ? 20) were expected to be 
generated. However, nine participants were eliminated in total, since they did not 
complete more than 25% of the target items. The rest of the participants completed the 
target items in average 98.2%. Eight incomplete responses were eliminated from further 
analyses. 
 
Table 1: Experiment 1, counts and percentages of ungrammatical/grammatical 
responses 
Conditions Codable Gramatical Ungramatical 
   N % 
Interogative in-situ 164 162 2 1.2 
Exclamative in-situ 165 153 12 7.3 
Interogative fronting 162 156 6 3.7 
Exclamative fronting 161 123 38 23.6 
Total   58 8.9 
 
Out of 652 codable responses, there were fifty-eight ungramatical responses, as 
ilustrated in Table 1. The number of erors in the Exclamative fronting condition is 
significantly larger than other conditions (?
2
(3) = 60.81, p < 0.01); it could be speculated 
that exclamatives with a fronted wh-phrase are quite dificult for the participants since 
they would be quite rare in naturalistic corpora. Investigation of ungramatical responses 
revealed that two main factors are applicable to most of the erors. The most frequent 
kind of eror was lack of a licensing particle. The second most frequent kind of eror was 
lack of a predicate. In order to complete the fragment gramaticaly, the participant must 
provide at least two predicates. 
 
3.2.2 Licensing Particles 
The next step is to examine the position in which the licensing particle occurs. The result 
is based on the gramatical responses, fifty-eight ungramatical responses being 
eliminated from the codable responses. Table 2 ilustrates where the licensing particle 
occurred: either at the embedded verb or the matrix verb position.
71
 
 
                                                
71
 In the following, we disregard the responses ?both,? in which the participants added the 
licensing particles both at the embedded and the matrix verbs, since the ?both? responses 
do not provide us any insight into the preference of the participants. This does not 
weaken the overal conclustion since there are just a few such responses. 
 125 
Table 2: Experiment 1, distribution of the licensing particles -ka (interogative) 
and -nodaroo (exclamative) in off-line sentence fragment completions. 
Conditions Licensing Particle 
 Embeded Clause Main Clause Both 
 N % N % N % 
Interogative in-situ 159 98.1 3 1.9 0 0 
Exclamative in-situ 151 98.7 2 1.3 0 0 
Interogative fronting 3 21.1 14 73.1 9 5.8 
Exclamative fronting 25 20.3 96 78.0 2 1.6 
 
With respect to the ?embedded? and ?matrix? responses, Fisher Exact tests showed that 
there was no reliable diference betwen two in-situ conditions (p > .40). Both conditions 
showed that an overwhelming number of completions contained a particle in the 
embedded verb position, although this is not gramaticaly required in either case. On 
the other hand, Fisher Exact tests showed that the paterns were significantly diferent in 
the two exclamative conditions (p < .001, 2-tailed). A similar patern emerged in the 
pairwise comparison betwen the two interogative conditions; Fisher Exact tests 
revealed that there were a large number of completions with a licensing particle at the 
embedded verb position in the interogative-in situ condition, while that was not the case 
in the interogative-fronting condition (p < .001, 2-tailed). Finaly, Fisher Exact tests 
showed that there is no reliable diference betwen the two fronting conditions (p > .40). 
Regarding the licensing particle in the exclamative conditions, we observed some 
optionality in whether noda or ka was used as the licensing particle. In the exclamative-in 
situ condition, out of 151 completions with a licensing particle at the embedded verb 
position, ninety-four responses (62.3%) used ka while fifty-seven responses (37.7%) used 
noda. Such optionality was not observed in interogative conditions; the licensing particle 
was always ka. One might suggest that the participants were confused with the distinction 
betwen interogatives and exclamatives. However, a further inspection of the data 
revealed that such confusion was not present at least for the most part. It was clear that 
the licensing particle ka in interogative conditions and exclamative conditions were 
handled diferently. In the following, an additional criterion was added so that al the 
cases in which the participants could be confused are eliminated for further analysis. 
 
Table 3: Experiment 1, presence of particle to at the embeded verb position 
  with to without to 
Conditions particle N % N % 
Exclamative-in situ noda 57 100 0 0 
Exclamative-in situ ka 43 45.7 51 54.3 
Interogative-in situ ka 4 2.5 155 97.5 
 
In al cases where noda was used in the exclamative in situ condition, it was always 
followed by another particle to as sen in Table 3. When ka was used, the data showed 
that it was not always the case that to was present; in only about half of the cases (45.7%), 
was to used. Fisher Exact tests show that the percentage of the presence of to after the 
 126 
particle ka in the interogative in situ condition (97.5%) was quite diferent from the 
exclamative in situ condition (54.3%) (p < .001, 2-tailed). 
The presence of to following the licensing particle ka in the exclamative in situ 
condition sems to be related to whether the predicate can select an interogative 
complement or an exclamative complement with ka and to, as shown in Table 4. 
 
Table 4: Experiment 1, presence/absence of particle to after the licensing particle ka 
in E/IS condition, kinds of predicates 
 without to with to 
Predicates N % N % 
Interogative taking 37 72.5 3 7.0 
Exclamative taking 14 27.5 40 93.0 
 
Out of fifty-one cases where the particle to was absent, predicates that take an 
interogative complement were used in thirty-seven cases (72.5%), whereas predicates 
that take an exclamative complement were used in fourten cases (27.5%). On the other 
hand, out of forty-thre cases where the particle to was present, predicates that take an 
interogative complement were used only in thre cases (7.0%), whereas predicates that 
take an exclamative complement are used in forty cases (93.0%). Fisher Exact tests 
showed that the diference was reliable (p < .001, 2-tailed). One can imagine that the 
cases where to was not used after the particle ka ilustrate that the participants did not 
treat them as exclamatives, which sems reasonable. When al the cases (fifty-one 
responses without to) were eliminated where the participants might have been confused 
with the distinction betwen the interogative and exclamative case, the overal patern 
did not change. Out of 102 responses in the exclamative in situ condition, 100 sentence 
fragments were completed with the licensing particle at the embedded clause (98.0%). 
Taken together, it sems reasonable to conclude that the variations of the licensing 
particles found in exclamative conditions do not show that the participants misunderstood 
the exclamative wh-phrase as interogative, but the data from the predicate show that the 
exclamative alows some variations of the licensing particles. 
 
3.2.3 Where the fronted wh-phrases are interpreted 
Turning to the results in the fronting conditions, Table 2 ilustrates that, regarding the 
position where the licensing particle occurs, there was no reliable diference observed. In 
Experiment 3 of Aoshima, et al (2004), an analysis of the argument structure was 
conducted with respect to the scrambling conditions. This extra work is important since, 
when the licensing particle is placed in the embedded verb position, we can be certain 
that the thematic-role for the fronted phrase originates from the embedded clause due to 
the gramatical requirement that the licensing particle has to c-command the base-
generated position of the wh-phrase. On the other hand, when the licensing particle is 
placed in the matrix verb position, the fronted phrase can receive a thematic-role either 
from the embedded or the matrix predicate, since the licensing particle at the matrix 
clause can c-command the base-generated position of the wh-phrase either in the 
embedded clause or the matrix clause. 
 127 
In the argument structure analysis below, fiften instances of embedded licensing 
from Exclamative fronting condition were eliminated due to the folowing reasons (hence 
only 10 counts in Table 5: 25?15=10). First, those are the cases where ka was used to 
license exclamative wh-phrases, and second, those did not have an extra particle to. 
Recal in the results from the Exclamative In-Situ condition, there was some concern that 
those cases ilustrate that the participants might be confused with the distinction betwen 
interogative and exclamative wh-phrases. 
Secondly, an analysis was conducted to determine how often the fronted wh-
phrases in exclamatives and interogatives were interpreted in the embedded clause. The 
counts were obtained, following Aoshima, et al (2004), by adding (i) the number of 
completions in which the licensing particle occurred in the embedded clause to (i) the 
number of completions in which the embedded predicate can, either obligatorily or 
optionaly, take a dative argument. Importantly, this excludes the trials with the particle 
ka in the embedded clause. The results are shown in Table 5. 
 
Table 5: Experiment 1, fronting conditions; counts and percentage of verbs that 
select a dative-arked argument in completions with main clause licensing particles. 
Conditions Embeded obligatory Embeded optional Main only 
 N % N % N % 
Interogative fronting 23 20.2 25 21.9 6 57.9 
Exclamative fronting 17 17.7 5 5.2 74 7.1 
Total 40  30  140  
 
This ilustrates that there were at most 22 cases (22.9%) where the fronted exclamative 
wh-phrases received their thematic-role in the embedded clause, while there were 48 
cases (42.1%) where the frotned interogative wh-phrases received their theta-role in the 
embedded clause. A Fisher Exact test showed that he diference was significant (p < .001, 
2-tailed). 
 
3.3 Discusion 
3.3.1 In Situ Wh-phrases 
This off-line experiment showed that there is a strong local licensing bias for the in situ 
conditions. When a wh-phrase (either interogative or exclamative) was found in the 
embedded clause, participants added a licensing particle in the embedded verb position. 
No diference was observed betwen the interogative and exclamative wh-phrases in this 
regard. This particular observation of interogative wh-phrases in the embedded clause 
ilustrates that our experiment replicates the results found in Experiment 3 in Aoshima, et 
al (2004) and to certain extent our results conform to those in Miyamoto & Takahashi 
(2002) obtained by on-line experiments. In addition, a very similar tendency was found 
for exclamative wh-phrases. Although exclamatives use diferent wh-phrases and 
diferent licensing particles, the results in this experiment suggest that the presence of the 
exclamative wh-phrase in the embedded clause creates an expectation that there must be 
a licensor at the embedded clause. 
Aoshima, et al (2004), following the conclusion in Miyamoto (2002), note that a 
sequence of a topic-marked NP and a nominative-marked NP is a strong cue for the 
 128 
participants that more than one clause is involved. Also, since there is no single predicate 
that can license al thre NPs (topic, nominative, and dative), it is clear that more than 
one predicate must be supplied. Asuming that the presence of an exclamative wh-phrase 
suggests to the participants that there must be a licensor at an appropriate place, there are 
thre options the participant can choose in order to complete this fragment as a 
gramatical sentence. Option (a) is to place the particle noda at the closest verb position. 
This forces the participants to add another particle to as wel. Then, the participants can 
add an appropriate predicate that can select an embedded exclamative clause. Option (b), 
is to place the particle noda only at the matrix clause. This option does not place the 
particle at the closest verb position. A gramatical sentence can be generated in this 
option since the licensing particle satisfies the c-command requirement of the 
exclamative wh-phrase. Finaly, Option (c) is to place the particle at the embedded verb 
position as wel as the matrix verb position. The particle atached at the embedded verb 
position must be followed by the particle to. 
The results of our experiment suggest that the subjects overwhelmingly choose 
option (a). The choice of (a) over (c) makes sense, under the asumption that adding the 
unnecesary exclamative particle is costly. One licensing particle that can c-command the 
wh-phrase is enough to satisfy the gramatical requirement of the wh-phrase. Hence, 
asuming that declarative sentences are the default, and that adding the exclamative 
particle in absence of the exclamatory wh-phrase is costly, it is quite understandable that 
the participants decided not to choose option (c). 
We stil have to explain the participants? choice of option (a) over option (b). As 
noted before, the participants? preference for lacing the licensing particle at the closest 
verb position is not specific to exclamatives, but is also observed in the in situ 
interogative condition (se also Aoshima et al., 2004). Based on the results of on-line 
experiments, Miyamoto & Takahashi (2002) argue that the presence of the wh-phrase 
creates an expectation of the licensing particle at the earliest gramaticaly legitimate 
position, leading to TME at the embedded verb position. 
The patern observed in our in situ wh-phrase conditions can be interpreted as 
indicating that the same procesing mechanism is underlying interogatives and 
exclamatives, although the details of their materials are slightly diferent from ours (they 
use acusative-marked wh-phrases; we used dative-marked wh-phrases), and the current 
experiment is an of-line experiment. One finding in our study, then, is that the bias for 
placing the local licensing particle is not specific to interogative wh-phrases, but it can 
be extended to exclamative wh-phrases as wel. Taken together, it shows that the 
preference of local licensing is quite strong when procesing not only interogative wh-
phrases but also exclamative wh-phrases. Since the next experiment also targets the 
procesing of the in situ exclamative wh-phrases, we discuss more of the isues related to 
this point. 
 
3.3.2 Fronted Wh-Phrases 
The results from the fronted conditions reveal a slightly diferent picture. Results from 
the fronted interogative wh-phrases are quite the same as sen in Experiment 3 of 
Aoshima, et al (2004). In their experiment, 32.4% of the wh-scrambling conditions were 
completed either having the interogative particle at the embedded position, or both at the 
embedded and matrix position (112/346; se their Table 7). From the argument structure 
 129 
analysis, they showed that 41.0% of the cases where there was a licensing particle at the 
matrix clause had an embedded predicate which can (optionaly or obligatory) take a 
dative argument (96/234; se their Table 8). In total, 61.4% of the fronted wh-phrases are 
interpreted in the embedded clause (208/346). In our experiment, 21.1% of the 
interogative wh-scrambling conditions were completed with the interogative particle at 
the embedded position, (33/156; se Table 2). From the argument structure analysis, 
42.1% of the cases where there was a licensing particle at the matrix clause had the 
embedded predicate which can (optionaly or obligatory) take a dative argument (48/114). 
In total, 55.1% of the fronted wh-phrases are interpreted in the embedded clause (81/147). 
Fisher Exact tests showed that there is no diference betwen our results and theirs. 
On the other hand, our results from the exclamative fronted condition showed that 
only 31.2% of the fronted wh-phrases are interpreted in the embedded clause in total 
(32/106). Therefore, the exclamative fronted condition was not only diferent from the 
interogative fronted condition in our experiment, but also diferent from the results 
obtained with interogatives in Aoshima, et al (2004). In the following, a potential 
acount is proposed for the contrast betwen the fronted interogative and exclamative 
wh-phrases. 
In the interogative fronted condition, the participants first encounter the dative-
marked interogative wh-phrase, followed by two NPs, each marked with a topic and 
nominative marker. The results from Aoshima, et al (2004) show that although it is 
gramaticaly possible to interpret the wh-phrase either at the matrix or the embedded 
clause, the readers predominantly interpret the fronted interogative wh-phrase at the 
embedded clause. Also, a similar patern is observed in our of-line experiment. 
Acording to Aoshima, et al (2004), a dependency is established betwen the fronted wh-
phrase and the gap in the embedded clause, since the embedded verb position is the 
earliest gramaticaly alowed position where the wh-phrase can receive a thematic role 
or the wh-phrase can be licensed by the Q-particle (se more discussion of the Q-marker 
in Aoshima, et al (2005) and Yoshida, et al (2006)). 
Upon encountering the fronted interogative wh-phrase followed by a topic-
marked NP, and since at this point there is no evidence for the parser that there is an 
upcoming embedded clause and the dative-topic is not the canonical order in a clause, it 
sems reasonable to asume that the parser takes it to mean that the wh-phrase is fronted 
from some base-generated position. This leads the parser to posit a gap and the licensing 
Q-particle in the matrix clause, eventualy establishing a dependency betwen the wh-
phrase and the Q-particle. Then, following Miyamoto?s (2002) observation, when the 
nominative-marked NP shows up, it sems that the parser builds a structure for the 
embedded clause. The results from Aoshima, et al (2004) and ours indicate that the parser 
?re-establishes? the dependency by positing a new gap in the embedded clause. 
Having a gap in the most deeply embedded clause is motivated since the posited 
gap can establish the dependency with the licensing Q-particle and satisfy the 
gramatical requirement as soon as possible. At the same time, what the parser has to do 
is to change the type of the embedded clause to interogative, by position the Q-particle at 
the embedded Comp position. Crucialy, there is no need to build any extra projections in 
the structure to determine the scope of the wh-phrase. 
Now, let us consider the case with the fronted exclamative wh-phrase. The 
following diagram ilustrates the situation where the parser encounters the sentence on a 
 130 
word-by-word basis. When the parser encounters the two initial words (a dative-marked 
exclamative wh-phrase and an NP), the parser can posit a functional projection in CP. 
Since the fronted wh-phrase is exclamative, before encountering the actual heads, the 
parser specificaly posits that there is a functional projection, Exclamative Phrase, and its 
overt realization: nodaroo. The parser makes a prediction on the presence of the 
morpheme since they are a gramaticaly required element in order to license the 
exclamative wh-phrase. Again the word order betwen the two phrases ?Wh-DAT + NP-
TOP? informs the parser that the exclamative wh-phrase is dislocated. Crucialy, it sems 
reasonable to asume that the exclamative wh-phrase is taken to be an argument of the 
matrix clause. 
The predominant patern observed in the experiment shows that the fronted 
exclamative wh-phrase is not interpreted in the most deeply embedded clause when the 
next word, NP-NOM, shows up. In other words, the parser keeps its initial hypothesis that 
the sentence initial exclamative wh-phrase is an argument of the matrix verb, and it is 
licensed by the functional projection at the matrix clause (se Figure 1). 
 
Figure 1: Fronted exclamative wh-phrase 
      ExclamativeP      
     ru       
   Wh-Exc-dat       
       IP  Exclamative no-da-roo   
     ru       
    NP-top        
      VP  Infl     
     ru       
      CP   Verb      
             
     ru       
      IP   Comp  to    
             
  NP-nom ru       
      VP   Infl      
             
     ru       
        Verb      
Recal that the parser?s revision of the gap position and re-establishment of the 
dependency, as sen with the fronted interogative wh-phrase, is motivated by the idea 
that the new gap in the embedded clause alows the gramatical requirement to be 
satisfied as soon as possible, since the gap in the embedded clause can form a 
dependency with the licensing particle at the embedded clause. The fact that the fronted 
exclamative wh-phrase is not taken by the parser as an argument of the embedded clause 
suggests that the parser is unwiling to posit the gap in the embedded clause or to re-
 131 
establish the dependency. If the parser posits the gap and the necesary projection in the 
embedded clause, the structure would be something like the one in Figure 2. 
 
Figure 2: Fronted exclamative wh-phrase 
      ExclamativeP       
     ru       
   Wh-Exc-dat       
       IP  Exclamative  no-da (roo)  
     ru       
    NP-top        
      VP  Infl     
     ru       
      CP   Verb      
             
     ru       
   ExclamativeP  Comp  to    
             
     ru       
     IP   Exclamative  no-da-(roo)  
             
  NP-nom ru       
      VP   Infl      
             
     ru       
        Verb      
Unlike the case of interogative wh-clauses, what the parser would have to do after 
recognizing that there is an embedded clause is to posit that there is an extra functional 
projection, eventhough the parser has already posited a gap and established a dependency 
in the matrix clause. This is crucialy diferent from what the parser has to do in wh-
interogatives. With the fronted interogative wh-phrase, in order to re-establish the 
dependency, the parser has to change the overt realization of the embedded Comp from 
the Declarative complementizer to to the Q-particle ka. On the other hand, what the 
parser might be wiling to do in the case of the fronted exclamative wh-phrase is to build 
an extra projection in order to re-establish the dependency so that the gramatical 
requirement of the exclamative wh-phrase is satisfied as soon as possible. Supposing that 
the choice is motivated only by re-establishing the dependency, and given the fact that 
the parser does not decide to posit the new gap, we argue that when the parser has a 
choice whether an extra projection is to be built or not, the parser does not choose to 
create the projection if the whole purpose of doing it is to re-establish the dependency. 
We should emphasize the ?if? clause since it is crucial for explaining another set 
of data. Recal the finding by Miyamoto & Takahashi (2002) that when the interogative 
wh-phrase appears in situ, the parser prefers to encounter the licensing particle ka in the 
closest gramaticaly possible Comp position, though it is gramaticaly possible to 
 132 
place the licensing particle ka at the matrix Comp position. Our data on exclamatives 
indicates that when the exclamative wh-phrase appears after a sequence of two 
nominative marked NPs, it shows a very strong local licensing preference, exactly as sen 
in the case of interogative wh-phrases. Again, in the case of exclamative wh-phrases, it 
is gramaticaly possible to place the licensing particles at the matrix clause. Hence, 
concerning the in situ cases, there is no diference betwen interogative and exclamative 
wh-phrases in terms of how they are procesed. 
For the in situ cases, the parser encounters the exclamative wh-phrase after a 
sequence of the two NPs (NP-TOP and NP-NOM). Our data show that the parser forms a 
dependency betwen the exclamative wh-phrase and the licensing particle in the 
embedded clause. Note that the dependency is formed for the first time at this point. The 
underlying intuition employed to acount for the results of the sentence initial 
exclamative wh-phrases is that building an extra projection is disprefered for re-
establishing the dependency. Since the dependency is established for the first time in the 
in situ cases, building an extra projection is possible and it is done in the most 
economical way, namely in the way that the gramatical requirement is satisfied as soon 
as possible. As argued above, the distinction betwen establishing and re-establishing the 
dependency is crucial. The basic idea is that building an extra structure is not harmful for 
establishing the dependency for the first time, but it is for re-establishing the dependency 
afterwards. 
Based on the results from Experiment 1, an on-line self-paced reading experiment 
was designed in order to investigate the time course of procesing the in situ exclamative 
wh-phrases. 
 
4. Experiment 2: Self-Paced Reading Task 
The aim of this experiment is to examine whether the TME can be observed with 
exclamative wh-phrases on-line, using a self-paced reading task. Given the results 
obtained in the sentence generation task, it is expected that the in situ exclamative wh-
phrases yield a local licensing bias, and the presence of the licensing particles for the 
exclamative wh-phrase at the embedded verb position. This wil be reflected in the 
reading time in the embedded verb region that includes the licensing particles. After the 
reader encounters the in situ exclamative wh-phrase in the embedded clause, the reading 
time of the embedded verb position with the licensing particle is predicted to be shorter 
then that without the licensing particle. 
 
4.1 Participants 
Forty-thre native speakers of Japanese participated in the experiment. Al of them were 
students at Hiroshima University, Japan. Al participants were financialy compensated. 
The experiment lasted about 20 min. 
 
4.2 Materials and Design 
Experimental materials consisted of twenty-four sets of sentences with four conditions 
each (Appendix B for the complete list of items used). The twenty-four sets of items were 
distributed among four lists in a Latin Square design. Each participant saw exactly one of 
the lists intermixed with seventy-two filers in a random order. Target items were 
 133 
manipulated by two factors; by the type of the dative argument in the embedded clause 
(an NP with non-exclamative modifier very vs. exclamative wh), and by the position of 
the particle noda-to or just to (at the embedded vs. matrix verb position). (19) ilustrates 
an example set. 
 
(19) Sample set of experimental conditions for Experiment 2 
 sono-kantoku-wa nintaiduyoi sensyu-ga ? 
 that-manager-TOP patient player-NOM 
 a. Exclamative / noda 
 nante ooku-no fan-ni hidoi waruguti-o ita-noda-to 
 wh-EXC many-GEN fan-DAT awful word-ACC said-EXC-COMP 
 benti-no ura-de gakkarisi-teiru 
 bench-GEN behind-at disappointed-ING 
 ?That manager disappointed, what a lot of fans the patient player said awful 
 words to.? 
 b. Non-exclamative / noda 
 totemo ooku-no fan-ni hidoi waruguti-o ita-noda-to 
 very many-GEN fan-DAT awful word-ACC said-EXC-COMP 
 benti-no ura-de gakkarisi-teiru 
 bench-GEN behind-at disappointed-ING 
 ?That manager disappointed that the patient player said awful words to a lot 
 of fans.? 
 c. Exclamative / to 
 nante ooku-no fan-ni hidoi waruguti-o it-teita-to 
 wh-EXC many-GEN fan-DAT awful word-ACC said-ING-COMP 
 benti-no ura-de kantigaisita-noda 
 bench-GEN behind-at misunderstood-EXC 
 ?What a lot of fans the manager disappointed that the patient player said awful 
 words to!? 
 d. Non-exclamative / to 
 totemo ooku-no fan-ni hidoi waruguti-o it-teita-to 
 very many-GEN fan-DAT awful word-ACC said-ING-COMP 
 benti-no ura-de gakkarisi-teiru 
 bench-GEN behind-at misunderstood-EXC 
 ?That manager disappointed that the patient player said awful words to a lot 
 of fans.? 
 
In the Exclamative conditions, an exclamative wh-phrase is used in an NP marked with a 
dative case -ni. In the Non-exclamative conditions, a modifier for an adjective totemo 
?very? substitutes for the exclamative wh-phrase. In the noda conditions, the exclamative 
licensing particle noda was placed in the embedded verb position, which is 
gramaticaly possible and linearly the closest position for the wh-phrase. In the to 
conditions, just the Declarative complementizer to, not the exclamative particle, appears 
in the embedded verb position, and the exclamative particle noda is placed at the matrix 
verb position. Also in the to conditions, the embedded verb appears in the teiru form in 
order to match the relative length of the embedded verb region. 
 134 
 
4.3 Procedure 
The experiment was conducted on a Del computer running Linger developed at MIT. 
Participants were timed in a phrase-by-phrase self-paced non-cumulative moving-
window reading task (Just & Carpenter, 1982). Most of the sentences were presented on a 
single line. Each trial item was segmented with spaces as shown in Table 6. The various 
particles used in this experiment were presented together with the preceding verbs, for 
those particles are not an independent prosodic word, but a bound morpheme. 
 
Table 6: Experiment 2, segmentation in the self-paced reading task 
1 2 3 4 5 6 
NP-TOP Adj NP-NOM exc-wh Adj NP-DAT 
very Adj NP-DAT 
Adj NP-ACC 
    
7 8 9 10 
V-EXC-COMP 
V-TEIRU-COMP 
NP-GEN NP-at V 
V-NODA 
 
In order to make sure that the participants paid sufficient atention to the experiment, a 
comprehension task was used. After each stimulus sentence, participants saw a yes-no 
comprehension question, which requires comprehension of the sentence. Data points 
were not included in the reading time analyses if the participant did not answer the 
comprehension task correctly. The experimental trials are preceded by instructions and 
thre practice trials. 
 
4.4 Results 
4.4.1 Task Acuracy and Data Analyses 
Some participants and items were eliminated from further analyses due to various reasons. 
Out of forty-thre participants, four participants were eliminated due to the poor acuracy 
(below 70%). Overal acuracy for the experiment is 83.33% without those four 
participants. We also disregarded two other participants, whose reading times are 
considerably longer, compared to others, and one participant, whose reading times are 
considerably shorter, compared to others. As for items, two items were disregarded due to 
their poor acuracy rate (53% and 60%, respectively), and one item due to some 
pragmatic awkwardnes (an unnatural habitual reading arises due to the use of the present 
tense in the embedded clause). The remaining data has been winsorized: data points that 
are 3.5 SD away from the average for each region were eliminated; only 1.0% of the total 
data points were afected. 
 
4.4.2 Reading Time 
The reading time analysis yielded the following results. Regions 7 to 10 were the critical 
regions. 
 
 135 
Table 7: Experiment 2, reading times (ms) and standard erors (in parentheses) per 
region 
Conditions Region 7 Region 8 Region 9 Region 10 
 emb.V NP-gen NP-P mat.V   
a. Exclamative-noda 604 (25) 529 (18) 435 (9) 570 (19) 
b. Exclamative-to 579 (22) 528 (15) 463 (10) 710 (23) 
c. Non-exclamative-noda 593 (23) 514 (16) 448 (11) 608 (20) 
d. Non-exclamative-to 589 (24) 534 (16) 462 (11) 684 (21) 
 
Regarding Region 7 (the embedded verb region), there was no main efect of the object 
type (exclamative or non-exclamative; Fs < 1) nor the complementizer type (declarative 
or exclamative; F
1
(1,35) = 1.17, p > .28; F
2
(1,20) < 1). In Region 8 (the NP-GEN region), 
there was also no main efect of the object type or the complementizer type (Fs < 1). 
At Region 9 (the NP-postposition region), there was a main efect of the 
complementizer type in the item analysis, and marginaly significant main efect of it in 
the participant analysis (F
1
(1,35) = 3.37, p < .075; F
2
(1,20) = 6.43, p < .02), due to the 
longer reading times in the to conditions. There was no main efect of the object type (Fs 
< 1). Pairwise comparisons revealed that, within the exclamative conditions, the to 
conditions were read significantly more slowly than the noda conditions (F
1
(1,35) = 4.48, 
p<.05; F
2
(1,20) = 9.88, p<.01). The same comparison for the non-exclamative conditions 
showed no significant diferences (F
1
(1,35) < 1; F
2
(1,20) = 1.23, p>.25). 
At Region 10 (the matrix verb position), there was no main efect of the object 
type (Fs < 1), but the main efect of the complementizer type was significant (F
1
(1,35) = 
30.14, p < .01; F
2
(1,20) = 19.20, p < .01), showing that conditions with the exclamative 
licensors at the matrix clause were read more slowly. The interaction of the object type 
and the complementizer type was marginaly significant in the participant analysis, but 
not in the item analysis (F
1
(,35) = 3.52, p < .07; F
2
(1,20) = 1.16, p > .25). 
 
4.5 Discusion 
In Experiment 1 of Miyamoto & Takahashi (2002), the condition with the wh-phrase 
where the embedded verb was followed by the Declarative complementizer to was read 
slower than the condition with the Q-particle ka, which is similar to our result. However, 
without the presence of a wh-phrase, the condition where the embedded verb was 
followed by the Q-particle was read slower than the condition with the Declarative 
complementizer, which is comparable to our Non-Exclamative conditions. In our 
experiment, no diference was observed in Non-Exclamative conditions. Regarding the 
results in Miyamoto & Takahashi (2002), they provide two explanations: (a) the 
occurrence of the Q-particle is unexpected given the lack of wh-phrase, and (b) in general, 
interogative sentences are read slower than declarative sentences. One diference 
betwen our exeriment and theirs is that the presence of Q-particle ka without being 
preceded by any wh-phrase unambiguously types a clause as interogative (yes-no 
question), while the presence of the exclamative particle noda without being preceded by 
any wh-phrase does not type a clause as exclamative, but the particle noda indicates that 
some constituent in the structure is focused (Hiraiwa & Ishihara, 2002). Crucialy, the 
clause remains typed as declarative. Such a diference betwen the gramatical 
 136 
properties of the relevant particles sems to contribute to the lack of reading time 
diferences in Non-Exclamative conditions in our experiment. 
The fact that the Exclamative-to condition was read slower than the Exclamative-
noda condition was the expected result. The reading time results in two Exclamative 
conditions showed that the presence of in situ exclamative wh-phrases yields a prediction 
that the licensing particles appear in the closest gramaticaly possible position. The 
reading times in the Exclamative-to condition indicated that the absence of the licensing 
particle was unexpected for the reader; i.e., this ilustrates that the TME is also observed 
with exclamative wh-phrases. Combined with the results obtained from the of-line 
studies, the data suggest that the parser posits the presence of the particle incrementaly. 
When compared with the results in Miyamoto & Takahashi (2002) and Aoshima, et al 
(2004), our results from the on-line study were weak and the efects showed up two 
regions after the reader encountered the crucial embedded verb region. Also, the main 
efect of the object type in that region was significant in the item analysis, but marginaly 
significant in the participant analysis. One speculation about our results is that, compared 
to interogatives, it is possible that there are fewer chances for the native speakers to 
encounter exclamatives. It sems that there are a lot of variations (possibly dialectal) with 
respect to the verbal suffixes that license exclamatives, again when compared to 
interogatives. In our experiment, the licensing particles for exclamatives are set only to 
noda. It could be that for some readers other forms of the licensing particles are more 
acesible. Those isues need to be investigated further in future research. 
Now, recal the patern of the sentence initial exclamative wh-phrases in our 
Experiment 1; the results showed that the sentence initial exclamative wh-phrases were 
predominantly not interpreted in the embedded clause. It was argued that the patern is 
acounted for if the parser does not make a prediction that there are licensing particles at 
the embedded clause. On the other hand, in conditions in which the exclamative wh-
phrase is located in the embedded clause in the sentence completion study and this self-
paced reading study, it was observed that there was a preference by the parser to posit the 
embedded licensing particles when it ses the exclamative wh-phrase in the embedded 
clause. An acount needs to be provided that explains the diference betwen the 
conditions with the sentence initial exclamative wh-phrase and the in situ exclamative 
wh-phrase. 
At the point where the parser encountered the exclamative wh-phrase in the 
embedded clause, it sems plausible to asume that the parser has the representation in 
Figure 3. 
 137 
Figure 3: In situ exclamative wh-phrase 
       CP        
     ru       
              
       IP   Comp      
     ru       
    NP-top        
       VP  Infl     
     ru       
     CP  Verb      
            
       ru       
     IP   Comp  to     
            
   NP-nom ru       
     VP  Infl      
            
       ru       
 ? Wh-exc    Verb      
From the previous experiments, it sems plausible that the parser expects that there is an 
embedded clause, and research on the procesing of the relative clause in the literature 
indicates that the parser expects that the embedded clause is the complement clause. 
Given that, when the parser ses the exclamative wh-phrase, the parser posits the 
presence of the licensing functional heads in the structure. What is crucial is that this is 
the earliest possible stage in which the parser is compeled to posit the licensing particles 
for the exclamative wh-phrase. The gramar alows two possible positions where the 
parser posits the licensing particles: the embedded or matrix Comp position. The 
experimental results from Miyamoto & Takahashi (2002) and Aoshima, et al (2004) and 
the ones from our experiments suggest that the parser makes a prediction that the 
licensing particles wil occur at the embedded Comp position, since the embedded Comp 
position is the earliest point where the gramatical requirement of the exclamative wh-
phrase can be satisfied. 
The crucial point in the in situ cases is that the parser realizes the presence of the 
exclamative wh-phrase for the first time after it recognizes that there is more than one 
clause. In other words, when it tries to establish the relevant dependency, it has already 
made a choice betwen putting the exclamative licensing particle in the embedded clause 
or the matrix clause. Our experimental results indicate that the parser predicts that there is 
an exclamative licensing particle at the embedded verb position, i.e., there is an extra 
functional projection in the embedded clause. The data from the fronted exclamative wh-
phrase suggest that positing this extra projection is possible in case it is the first time the 
parser establishes the relevant dependency. Note that in such a case, the parser is not re-
establishing the dependency. When the parser realizes that positing an extra functional 
projection is required in order to determine the dependency at the earliest point, the 
parser gives up positing that projection. 
 138 
 
 
5. General Discusion 
The discussion above based on the two experiments shows that there is no procesing 
diference betwen interogative and exclamative wh-phrases when they occur in an 
embedded clause. Given the preference of the parser, the relevant licensing particles are 
expected to occur in the embedded CP domain. Now, what needs to be acounted for is 
the procesing diference betwen the fronted and in situ exclamative wh-phrases. When 
the parser ses the embedded exclamative wh-phrase in the embedded clause, the parser 
has to determine that the licensing particles for exclamatives occur overtly, given the 
gramatical requirement for the exclamatives. On the other hand, when the parser ses 
the fronted exclamative wh-phrase and the subsequent sequence of two NPs marked with 
the nominative case-marker, the parser has already posited the overt realization of the 
licensing particles in the structure. As argued above, in order to license the fronted 
exclamative wh-phrase in the embedded clause, the parser has to predict the overt 
realization of the same licensing functional heads in the embedded clause. Since these 
were not the results we got in our experiment, we need to say that the parser does not 
predict whether the licensing functional heads are overtly realized in order to solely 
shorten the dependency. The situation is diferent when the parser ses the exclamative 
wh-phrase in the embedded clause. Here, the overt realization of the licensing heads is 
predicted in order to build a gramatical representation. Without the licensing heads, the 
sentence is ungramatical. With respect to the fronted interogative wh-phrase, after the 
parser ses the second NP marked with the nominative case marker, the parser predicts 
that there is an embedded complement clause in which the overt Comp is also predicted 
independently from the presence of the interogative wh-phrase. Crucialy, the parser 
does not have to build another functional projection to posit a gap in the embedded clause 
We then argue that re-establishing the dependency for the fronted wh-phrase is possible 
by virtue of the fact that the position in which the licensing particle ka occurs happens to 
be Comp where the overt realization of the Declarative complementizer to is guaranted. 
On the other hand, the positions in which the licensing particles for the exclamative wh-
phrase are diferent in the sense that the positions are overtly realized only to satisfy the 
gramatical requirement, such as licensing exclamatives. In short, we argue that, when 
the parser?s preference (such as scope-marking) is at isue, predictions of what form a 
given head is realized are relatively easier than predictions of whether a given head is 
overtly realized or not. 
Obviously, it is very important to determine exactly where such a ranking is 
relevant. The efect is counter-intuitive in the area where the preference of shortening the 
dependency is not at isue. For instance, when the parser realizes that it is dealing with a 
sentence in Japanese, the parser can predict the verb is overtly realized even before it is 
encountered. On the other hand, there is no evidence that the parser predicts the ful 
content of the verb itself. The case we discussed here is, nonetheles, the prediction 
regarding the functional projections, the content of which is extremely limited. The 
alternation betwen the Declarative complementizer to and the Q-particle ka is quite 
specific; so is the form of the licensing particles for exclamatives. 
Taken together, based on a series of experiments on exclamative sentences in 
Japanese, we have shown that TME is observed in exclamatives. Furthermore, we argued 
 139 
that not al functional projections are treated equaly by the parser. Specificaly, we 
suggested that the Comp where the Declarative complementizer to or the Q-particle ka 
occurs is handled in a slightly diferent manner than functional projections such as 
Exclamative Phrase. The diference betwen those two ?types? of functional projections 
emerges in a specific situation where the parser?s preference to shorten the dependency is 
involved. 
 
 
6. Apendix A: items in the sentence completion task A. 
Each of the items in this list represents one ful set of stimuli from Experiment 1. 
Alternative words are given in parenthesis, separated by a slash (/). In fronted conditions, 
the parts in parenthesis are fronted to the beginning of the fragment. 
 
1, f w 
? 
\ x ? 
\ U {r w 	? w ? t }/{s? oh X ^ ? w 	? w ? t } ? 
sono sense-wa gakuse-ga {dono onanoko-ni}/{nante takusan-no onanoko-ni} 
that teacher-top student-nom {which girl-dat}/{wh-exc many-gen girl-dat} 
2, 3 ? x S l U {r w 	H ? ? t }/{s? oT ? M M 	H ? ? t } ? 
tentyo-wa okyaku-ga {dono zyugyoin-ni}/{nante kawai zyugyoin-ni} 
store manager-top customer-nom {which worker-dat}/{wh-exc cute worker-dat} 
3, f w 	? w ? x 
t ? U {r w ? w ? t }/{s? oj M ^ s ? w ? t } ? 
sono onanoko-wa karesi-ga {dono otokonoko-ni}/{nante tisana otokonoko-ni} 
that girl-top boyfriend-nom {which boy-dat}/{wh-exc smal boy-dat} 
4, ? 	t w  ^ ? x f w ? ? ? ? ? ? U {r w ? 
 t }/{s? oT ? M f O s ? 
 t } ? 
kinzyo-no okusan-wa sono serusuman-ga {dono rozin-ni}/{nante kawaisona rozin-ni} 
neighbor-gen house wife-top that salesman-nom {which old man-dat}/{wh-exc por old man-dat} 
5, f w ? ? x ? ? U {r w 	 ? 
? t }/{s? oh X ^ ? w 	 ?
? t } ? 
sono syatyo-wa syain-ga {dono torihikisaki-ni}/{nante takusan-no torihikisaki-ni} 
that president-top worker-nom {which customer-dat}/{wh-exc many-gen customer-dat} 
6, 
? ? x / o U {r w / o ? t }/{s? o{ 	s /o ? t } ? 
kumityo-wa kesatu-ga {dono kesatukan-ni}/{nante hiyowana kesatukan-ni} 
yakuza bos-top police-nom {which policeman-dat}/{wh-exc weak policeman-dat} 
7, f w ? ? H x ? ? U {r w H  t }/{s? o? ? s M H t } ? 
sono katudoka-wa sityo-ga {dono kazoku-ni}/{nante megumarenai kazoku-ni} 
that activist-top mayor-nom {which family-dat}/{wh-exc por family-dat} 
8, 
? ? x ? ? U {r w 
 	{ t }/{s? o??s 
 	{ t } ? 
butyo-wa syatyo-ga {dono hisyo-ni}/{nante munona hisyo-ni} 
manager-top president-nom {which secretary-dat}/{wh-exc incompetent secretary-dat} 
9, f w ? 	 x ? 	 U {r w  ? 
? t }/{s? o? t q h s M  ? 
? t } ? 
sono isya-wa kanzya-ga {dono kangohu-ni}/{nante yakunitatanai kangohu-ni} 
that doctor-top patient-nom {which nurse-dat}/{wh-exc useles nurse-dat} 
10, f w 	 	 x ? 
\ U {r w ? 	$ t }/{s? o${s ? 	$ t } ? 
sono zyosyu-wa gakuse-ga {dono kyozyu-ni}/{nante gankona kyozyu-ni} 
that asistant-top student-nom {which profesor-dat}/{wh-exc stuborn profesor-dat} 
1,  ^ ? x S 	 ; M ^ ? U {r w S l t }/{s? o?? s Sl t } ? 
okusan-wa otetudaisan-ga {dono okyaku-ni}/{nante siturena okyaku-ni} 
wife-top house keper-nom {which guest-dat}/{wh-exc impolite guest-dat} 
12, f w N 	 x ? ? ?? ? ? ? U {r w ^? H t }/{s? o
> w s M ^ ? H t } ? 
sono kasyu-wa purodyusa-ga {dono sakusika-ni}/{nante ninkinonai sakusika-ni} 
that singer-top producer-nom {which lyric writer-dat}/{which unpopular lyric writer-dat} 
13, f w ? ? x ? ? ? ? ? ? U {r w <
e Z t }/{s? o X w < 
e Z t } ? 
 140 
sono syatyo-wa purogurama-ga {dono sitauke-ni}/{nante oku-no sitauke-ni} 
that president-top programer-nom {which subsidiary-dat}/{wh-exc many-gen subsidiary-dat} 
14, 
 	{ x f w 
S ? H U {r w ? ?H t }/{s? o
? s? ?H t } ? 
hisyo-wa sono sezika-ga {dono sisanka-ni}/{nante yuhukuna sisanka-ni} 
secretary-top that politician-nom {which property owner-dat}/{wh-exc rich property owner-dat} 
15, f w ? 0 x ? 	t w 
@ 
? 0 U {r w ? ?h j t }/{s? o
?a s? ? h j t } ? 
sono sakanaya-wa kinzyo-no yaoya-ga {dono kodomotati-ni}/{nante binbona kodomotati-ni} 
that fish dealer-top neighbor-gen vegetable dealer-nom {which children-dat}/{wh-exc por children-dat} 
16, ? 	t w V 0 x ? w ? ? ? ? 0 U {r w G 	 t }/{s? o?? sG 	 t } ? 
kinzyo-no hanaya-wa tonari-no ramenya-ga {dono kisya-ni}/{nante komana kisya-ni} 
neighbor-gen flower dealer-top next dor-gen ramen dealer-nom {which writer-dat}/{wh-exc snoby 
writer-dat} 
17, 	 	 x f w ? 	$ U {r w ? 
\ t }/{s? oyT s?
\ t } ? 
zyosyu-wa sono kyozyu-ga {dono gakuse-ni}/{nante bakana gakuse-ni} 
asistant-top that profesor-nom {which student-dat}/{which stupid student-dat} 
18, f w 
? 
? x ? ?h j U {r w  a t }/{s? o
?s  a t } ? 
sono huhu-wa kodomotati-ga {dono tomodati-ni}/{nante soena tomodati-ni} 
that maried couple-top children-nom {which friend-dat}/{which estranged friend-dat} 
19, f w ? ? ? ? x ? ? ? U {r w ? ? ? ? ? t }/{s? o>? ` M ? ? ? ? ? t } ? 
sono kapuru-wa gaido-ga {dono doraiba-ni}/{nante kimuzukasi doraiba-ni} 
that couple-top guide-nom {which driver-dat}/{wh-exc hard to please driver-dat} 
20, r ? x ? ? U {r w 
\ F t }/{s? o? ^ ` M 
\ F t } ? 
tanin-wa kotyo-ga {dono seto-ni}/{nante yasasi seito-ni} 
clas teacher-top principle-nom {which student-dat}/{wh-exc kind student-dat} 
21, 	Q < x < 
 U {r w ? ? t }/{s? o ^ ? ` U ? ? w ? ? t } ? 
oba-wa hahaoya-ga {dono kodomo-ni}/{nante samisigariya-no kodomo-ni} 
aunt-top mother-nom {which child-dat}/{wh-exc lonely-gen child-dat} 
2, ? ? x ? ? ? ? ? ? U {r w 
? 
 t }/{s? o?/? w 
? 
 t } ? 
bokusi-wa borantia-ga {dono byonin-ni}/{nante tosiyori-no byonin-ni} 
clergyman-top volunter-nom {which patient-dat}/{wh-exc old-gen patient-dat} 
23,  ? x 
? 	 U {r w ? ? ? t }/{s? o? 	?s? ? ? t } ? 
kantoku-wa sensyu-ga {dono fan-ni}/{nante nesina fan-ni} 
manager-top player-nom {which fan-dat}/{wh-exc enthusiastic fan-dat} 
24, ? ? ? ? ??? x ; ? U {r w 	? l t }/{s? o l o M ? 	? l t } ? 
sutyuwadesu-wa kityo-ga {dono zyokyaku-ni}/{nante hutoteiru zyokyaku-ni} 
stewardes-top pilot-nom {which pasenger-dat}/{wh-exc fat pasenger-dat} 
25, ? 4	 x ; ? H U {r w ? ? ? ? ? t }/{s? o	 M ? ? ? ? ? t } ? 
sikisya-wa ongakuka-ga {dono pianisuto-ni}/{nante wakai pianisuto-ni} 
conductor-top musician-nom {which pianist-dat}/{wh-exc young pianist-dat} 
26, : x 
? U {r w 	M? t }/{s? o M T m M 	M ? t } ? 
tuma-wa oto-ga {dono zyui-ni}/{nante ikatui zyui-ni} 
wife-top husband-nom {which veterinarian-dat}/{wh-exc stern veterinarian-dat} 
27, f w  q ^? x ? ? U {r w P
? q ? t }/{s? oO ^ ? X ^ M P 
? q ? t } ? 
sono kokaigin-wa tizi-ga {dono kensetugaisya-ni}/{nante usankusai kensetugaisya-ni} 
that politician-top mayor-nom {which construction company-dat}/{wh-exc suspicious construction 
company-dat} 
28, G 	 x f w ?X _ U {r w 	?  t }/{s? o
? ` M 	?  t } ? 
kisya-wa sono yakuza-ga {dono zyoyu-ni}/{nante utukusi zyoyu-ni} 
reporter-top that gang-nom {which actres-dat}/{wh-exc beautiful actres-dat} 
29, y ?
\ x f w ? 	 U {r w 
? ? t }/{s? o? dh 
?? t } ? 
ryugakusei-wa sono isya-ga {dono hutyo-ni}/{nante yaseta hutyo-ni} 
exchange student-top that doctor-nom {which chief nurse-dat}/{wh-exc skiny chief nurse-dat} 
 141 
30, 3 ? x f w 
b 
 ? U {r w 	? 
Q t }/{s? o> 	s 	?
Q t } ? 
tentyo-wa sono hanbain-ga {dono zyose-ni}/{nante kiyowana zyose-ni} 
store manager-top that worker-nom {which lady-dat}/{wh-exc weak lady-dat} 
 
 
7. Apendix B: items in the self-paced reading task 
Each of the items in this list represents a full set of stimuli from Experiment 2. In the 
Japanese items, alternative words are given in curly brackets, separated by a slash (/). 
English translations are provided only for the non-exclamative conditions (b) and (d). 
The alternative forms of the main clause and embedded verbs are given in curly brackets, 
separated by a slash (/). The words in square brackets correspond to the phrase that is an 
exclamative wh-phrase in the exclamative conditions (a) and (c). Se example (19) for 
one full set of items. 
 


fw ? ? ?? ? ? x 
? g? U 
\s ? o q o? ^A ?w q M_ 	6 Mt 
?g w 

]? m Z? 
\T
?i w i T ?p M h ^q 
?? ? ?w 
j p 
\\?` o Mh { t M ~? ` oM h wi { ^

sono uetoresu-wa ryorityo-ga {nante/totemo} yoryonowarui minarai-ni ryori-no morituke-o 
{tanonda-noda/tanondeita}-to kitin-no ura-de {kobositeita/ihurasiteita-noda} 
that waitres-top chef-nom {wh-exc/very} ineficient aprentice-dat dish-gen garnish-ac 
{asked-exc/asked}-comp kitchen-gen behind-at {complained/spread the word-exc} 
Behind the kitchen, that waitres {complained about/spread the word that} the chef {asking/asked} [a very 
ineficient aprentice] to garnish the dish. 


fw? ? x 
	? 
w 
? 	U 
\s? o qo ? ^? ? M h? 	 t 
?s 
		[? 
\b? w i ` o
M? ^q 

?? w 
g ?q p 
\V l oM h { 	? `h w i{ ^

sono intyo-wa sinzin-no isya-ga {nante/totemo} tosioita kanzya-ni konana syuzyutu-o {suru-
noda/siteiru}-to byoin-no rizikai-de {okoteita/syutyosita-noda} 
that chief doctor-top new doctor-nom {wh-exc/very} old patient-dat dificult operation-ac 
{conduct-exc/conduct}-comp hospital-gen executive meting-at {got angry/exclaimed-exc} 
At the executive meting, that chief doctor {got angry about/exclaimed that} the new doctor 
{conducting/conducted} a dificult operation on [a very old patient]. 


fw_ 	6 Mx 
? ?
 M 
i ? U 
\s ?o q o? ^
?+; s G? t 

y ? 	OG s 
q	?  ?? 

\?T dh w i ? T do M h ^q 
	?? 	U Mw 
'j 	? [ p 
\ lo M ?{ ?? M `h w i{ ^

sono minarai-wa siryobukai toryo-ga {nante/totemo} bukiyona daiku-ni sekininzyudaina 
genbakantoku-o {makaseta-noda/makaseteita}-to sintikuiwai-no utiage-de {ikidoteiru/kantigaisita-noda} 
that aprentice-top thoughtful chief-nom {wh-exc/very} clumsy builder-dat responsible management-ac 
{left-exc/left}-comp building ceremony-gen party-at {is angry/misunderstod-exc} 
At the building dedication, that aprentice {is angry at/misunderstod that} the thoughtful chief 
{leaving/left} [a very clumsy carpenter] responsible for managing the place. 


fw
 	{ x 
? ? 
M 
?	$ U 
\s ? o q o ? ^
? x Ts ? 
\t
	OA s 
?? ? 
\) Q ?w i 
)Q o Mh ^q 
? J w 
q ^p
\pM o M? { ?M ?i w i{ ^

sono hisyo-wa tyuibukai kyozyu-ga {nante/totemo} asahakana gakuse-ni zyuyona sigoto-o {ataeru-
noda/ataeteita}-to gaka-no kaigi-de {nageiteiri/omoikonda-noda} 
that secretary-top careful profesor-nom {wh-exc/very} careles student-dat important job-ac 
{give-exc/gave}-comp department-gen meting-at {is lamenting/misunderstod-exc} 
At the department meting, that secretary {is lamenting/misunderstod that} the careful profesor 
{giving/gave} an important job to [a very careles student]. 


fw] ? x 
k ` M 

??U 
\s? o qo ? ^	 M ? ?t 
?8 w 
?s K? 
\)Q h wi )
Qo M ? ^q 
<? w 
? 	 p 
\p Mo M ?{ ~ ?s l hw i { ^

sono katyo-wa kibisi butyo-ga {nante/totemo} wakai syain-ni tyoki-no yukyukyuka-o {ataeta-
noda/ataeteiru}-to kaeri-no densya-de {nageiteiru/huremawata-noda} 
 142 
that supervisor-top strict manager-nom {wh-exc/very} young worker-dat long-gen paid leave-ac 
{gave-exc/give}-comp return-gen train-at {is lamenting/spread the word-exc} 
On the train home, that supervisor {is lamenting/spread the word that} the strict manager {giving/gave} [a 
very young worker] paid leave. 


fw N N	x 
?a ? s 
? ?? ? ?? U 
\s ? o qo ? ^
?
Qwq M ?? ? ?q ? t 
? ?
?? ? w 

? ;? 
\?T ` hw i ?T ` oM ? ^q 
? ?	t w 
? 	 	?p 
\9V ` oM ? { ? M 
?i w i{ ^

sono enkakasyu-wa mazimena manezya-ga {nante/totemo} hyobanowarui ibentogaisya-ni konsato-no 
senden-o {iraisita-noda/iraisiteiru}-to zimusyo-no tyusyazyo-de {gekidositeiru/omoikonda-noda} 
that enka singer-top serious manager-nom {wh-exc/very} disreputable ad company-dat concert-gen ad-ac 
{asked-exc/ask}-comp ofice-gen parking-in {is furious/misunderstod-exc} 
In the ofice parking lot, that enka singer {is furious about about/misunderstod that} the serious manager 
{asking/asked} [a very disreputable ad company] to advertise the concert. 


fw? ? ?? ? ?? x 
?? ? ?w 
?? ? ?? ? U 
\s? o q o? ^O ^?X ^M? ?? ?
?? ? t 
	? `M ? ?? ? w 

M ^? 
\? dh w i ? d oM ? ^q 
? ?? ? w 
? 0p 
\? ` o
Mh { 
?? `h w i{ ^

sono myuzisyan-wa beteran-no manezya-ga {nante/totemo} usankusai purodyusa-ni atarasi 
arubamu-no seisaku-o {makaseta-noda/makaseteiru}-to terebikyoku-no gakuya-de {hungaisiteita/hinansita-
noda} 
that musician-top experienced-gen manager-nom {wh-exc/very} od producer-dat new album-gen produce-
ac {asked-exc/asked}-comp TV station-gen gren rom-at {got angry/railed about-exc} 
In the gren rom at the TV station, that musician {got angry about/railed about the fact that} the 
experienced manager {asking/asked} [a very od producer] to produce the new album. 


fw? ? x 
	? ? $s 

?
\h jU 
\s? o qo ? ^.M ? ?t 
?` M 

:? ? 
\? Q hw i 
?Q o Mh ^q 
? ? w 
	? ?q ^p 
\t M? l oM h { ?? M` h wi { ^

sono kotyo-wa zyosikitekina sensetati-ga {nante/totemo} osanai zido-ni muzukasi sugaku-o 
{osieta-noda/osieteita}-to gako-no syokuinkaigi-de {ihateita/kantigaisita-noda} 
that principal-top responsible teachers-nom {wh-exc/very} young student-dat dificult math-ac 
{taught-exc/taught}-comp schol-gen meting-at {exclaimed/misunderstod-exc} 
At the schol meting, that principal {exclaimed/misunderstod that} the responsible teachers 
{teaching/taught} dificult math to [very young children]. 


fw? ? ?x 
N ?s 
?3 ? U 
\s ?o q o? ^
?M 	? 
t
? w 
%? ? 
\? ? hw i 
?? o M? ^q 
? 3 w 
q ^p
\?M o M? { 	?a h wi { ^

sono ginkoin-wa kenzituna sitentyo-ga {nante/totemo} haraguroi-ni syonin-ni tagaku-no yusi-o 
{mitometa-noda/mitometeiru}-to siten-no kaigi-de {odoroiteiru/sinzita-noda} 
that bank worker-top conservative chief manager-nom {wh-exc/very} bad company-dat a lot-gen debt-ac 
{acepted-exc/acepted}-comp branch-gen meting {is surprised/misunderstod-exc} 
At the branch meting, that bank worker {is surprised about/misunderstod that} the conservative chief 
manager {alowing/alowed} [a very bad campany] to borow a lot of money. 


fw- ? 	x 

Y? f Os 
S 	n? ^?U 
\s? o qo ? ^h X ^?w 
?? t 
? ? s 
? d
MN ? 
\? ? hw i ? ? oM ? ^q 
15"w 
	B ?? p 
\
C` oM h { t M? ? wi { ^

sono hogosya-wa syozikisona omawarisan-ga {nante/totemo} takusan-no kodomo-ni kyoryokuna 
kakusezai-o {hirometa-noda/hirometeiru}-to PTA-no atumari-de {kokuhatusiteita/iharu-noda} 
that parent-top respectable policeman-nom {wh-exc/very} many-gen children-dat strong drug-ac 
{spread-exc/spread}-comp PTA-gen meting-at {pointed out/claimed-exc} 
At the PTA meting, that parent {pointed out/claimed that} the respectable policeman {giving/gave} strong 
drugs to [very many children]. 


fw ? x 
? ^`M 

?	U 
\s ? o q o? ^? Qw q M
l t 
? ?? w 
?? ? ? 
\
l
hw i 
l oM h ^q 

P
? w 
?? ? p 
\K V? o M? { 	? a hw i { ^

sono kantoku-wa yasasi sensyu-ga {nante/totemo} gyoginowarui kankyaku-ni timu-no guzu-o 
{kubata-noda/kubateita}-to ensesaki-no hoteru-de {akireteiru/sinzita-noda} 
that manager-top kind player-nom {wh-exc/very} il-manered spectator-dat team-gen gods-ac 
{provided-exc/provided}-comp expedition-gen hotel-at {is amazed/mistok-exc} 
 143 
At the hotel, that manager {is amazed about/wrongly mistok the fact that} the kind player 
{providing/provided} team gods to [a very il-manered spectator]. 


fw ? x 
? 1?M 

?	U 
\s ? o q o? ^ Xw 
?? ? t 
{ r M 
q ?? 
\tl h w
i t lo M h ^q 
?? ? w 
j p 
\U l T? ` oM ? { ? ?M ` hw i { ^

sono kantoku-wa nintaizuyoi sensyu-ga {nante/totemo} oku-no fan-ni hidoi waruguti-o {ita-noda/iteita}-
to benti-no ura-de {gakarisiteiru/kantigaisita-noda} 
that manager-top patient player-nom {wh-exc/very} many-gen fan-dat bad words-ac {told-exc/told}-comp 
bench-gen behind-at {is disapointed/misunderstod-exc} 
Behind the bench, that manager {is disapointed about/misunderstod that} the patient player 
{swearing/swore} at [very many fans]. 


fw ? ?? x 
? 0 s 
 ? U 
\s ?o qo ? ^	2 s ?? ? ?? t 
k ` M 

G? 
\) Q h
wi )Q h ^q 
 	?w 
? ?? ? p 
\U lT ? `o M ? 8 4` o M? w i ^{ 

sono kyudansyatyo-wa kanyona kantoku-ga {nante/totemo} yusyuna pitya-ni kibisi batu-o {ataeta-
noda/ataeta}-to kyuzyo-no burupen-de {gakarisiteiru/kitaisiteiru-noda} 
that owner-top patient manager-nom {wh-exc/very} great pitcher-dat harsh penalty-ac {gave-exc/gave}-
comp bal park-gen bulpen-at {is disapointed/expected-exc} 
In the bulpen of the balpark, that owner of the team {is disapointed about/ses that} the patient manager 
{giving/gave} a harsh penalty to [a very talented pitcher]. 


fw? ? ?? ? x 
?s 
? ?? ? ?? U 
\s ? o q o? ^h X^? w	?  t 
? 
	B w 
?
?? ? 
\T ? iw i T ? pM h ^q 
? ?? ? w 
? Q ?p 
\y? ? pM ? { ? r` h wi { ^

sono kameraman-wa yunona asisutanto-ga {nante/totemo} takusan-no zyoyu-ni syasinsyu-no moderu-
o {tanonda-noda/tanondeita}-to sutazio-no hikaesitu-de {kuyandeiru/gokaisita-noda} 
that cameraman-top eficient asistant-nom {wh-exc/very} many actres-dat photo album-gen model-ac 
{asked-exc/asked}-comp studio-gen rom-at {regret/misunderstod-exc} 
In the studio gren rom, that cameraman {regrets/misunderstod that} the eficient asistant 
{asking/asked} [very many acreses] to be models in a photo album. 


fw? ? q? x 
? ? s 
 ? U 
\s ?o qo ? ^U? w qM 
?? 0 t 
G ? w 
	;M ? 

\^dhw i ^do Mh ^q 
? 	t w 
q ?p 
\|l X ? `h { ? ` hw i { ^

sono tyonaikaityo-wa zityokuna syuhu-ga {nante/totemo} garanowarui tekiya-ni maturi-no tetudai-o 
{saseta-noda/saseteita}-to kinzyo-no kaigo-de {bikurisita/uwasasita-noda} 
that local oficial-top honest mayor-nom {wh-exc/very} bad stret vendor-dat festival-gen help-ac 
{let-exc/let}-comp neighborhod-gen meting-at {surprised/gosiped-exc} 
At the neighborhod meting, that local oficial {was surprised about/gosiped that} the honest mayor 
{asking/asked} [a very bad stret vendor] to help with the festival. 


fwG ? 
\x
T? ` M 
? ? ?? U 
\s ? o q o ? ^
?
Qwq M 
? 	$ t 
 	2s 
a? 

\
*
? `h w i 
* 
? `o M h ^q 
Z? ? w 

? p 
\U l T? ` oM h { > ?m Z hw i { ^

sono daigakuse-wa tanomosi sidokyokan-ga {nante/totemo} hyobanowarui kyozyu-ni yusyuna 
tomodati-o {suisensita-noda/suisensiteita}-to kenkyusitu-no mae-de {gakarisiteita/kimetuketa-noda} 
that student-top reliable profesor-nom {wh-exc/very} disreputable profesor-dat talented friend-ac 
{recomended-exc/recomended}-comp ofice-gen front-at {was disapointed/aserted-exc} 
In front of the ofice, that student {was disapointed about/aserted that} the reliable profesor 
{recomending/recomended} a talented friend to [a very disreputable profesor]. 


fw? ? x 

? ?s 

	{U 
\s? o q o ? ^G ? s 
	?
? t 
GV s 
 ? ? 
\) Qh w i 
)Q o M? ^q 
? ? w 
q ^p
\?l ` oM ? { ? r `h w i{ ^

sono syatyo-wa someina hisyo-ga {nante/totemo} daizina torihikisaki-ni okina songai-o {ataeta-
noda/ataeteiru}-to yakuin-no kaigi-de {situbosita/gokaisita-noda} 
that president-top smart secretary-nom {wh-exc/very} important customer-dat big los-ac 
{caused-exc/caused}-comp executive-gen meting-at {is disapointed/misunderstod-exc} 
At the executive meting, that president {is disapointed about/misunderstod that} the smart secretary 
{causing/caused} a big los to [a very important customer]. 


fw	?  x 
? ?? w 
? U 
\s ? o q o? ^?s 
? t 
?s ^w 
? ?? 
\T? i w
i T ?p M h ^q 
?? ? ?w 
? p 
\K V? o M? { 84 ` hw i { ^

 144 
sono zyoyu-wa zituryokuha-no kantoku-ga {nante/totemo} yumena danyu-ni zikaisaku-no wakiyaku-o 
{tanonda-noda/tanondeita}-to rokebasu-no naka-de {akireteita/kitaisita-noda} 
that actres-top skilful-gen director-nom {wh-exc/very} famous actor-dat next movie-gen suporting 
actor-ac {asked-exc/ask}-comp broadcast van-gen inside-at {is amazed/anticipated-exc} 
In the broadcast van, that actres {is amazed at/anticipated that} the skilful director {asking/asked} [a very 
famous actor] to be a suporting actor in the next movie. 


fw? ? x 
	? `M 

?<U 
\s? o q o ? ^$ { s	? 
?t 
?8 w 
?? 
\q ? mZ h w
i q ?m Z h ^q 
?S ? w 
? 	
?q p 
\? 	? `o M ?{ &l h wi { ^

sono syatyo-wa atarasi buka-ga {nante/totemo} gankona torihikisaki-ni tyoki-no keyaku-o {torituketa-
noda/torituketa}-to nendomatu-no kabunusisokai-de {kansinsiteiru/negata-noda} 
that president-top new orker-nom {wh-exc/very} dificult customer-dat long term-gen contract-ac 
{made-exc/made}-comp end of year-gen stock holder meting-at {is impresed/wished-exc} 
At the year-end stock holder meting, that president {is impresed by/wished that} the new orker 
{making/had made} a long-term contract with [a very dificult customer]. 


fw? ? ?? ? ?x 

\?> s 
? ? ?? U 
\s ? o q o? ^h X^? w? ? ?t 
?
? 
? ?
?	? ? ? 
\I `h w i I `o M h ^q 
?? ? ?? w 
q	? p 
\? ? `o M h{ tM ? lh w
i{ ^

sono manezya-wa namaikina aidoru-ga {nante/totemo} takusan-no fan-ni tyokusetu sainsikisi-o 
{watasita-noda/watasiteita}-c konsato-no kaizyo-de {kandositeita/ihata-noda} 
that manager-top insolent idle-nom {wh-exc/very} many-gen fan-dat directly autograph-ac 
{gave-exc/gave}-comp concert-gen place-at {was moved/claimed-exc} 
At the concert, that manager {was moved by/claimed that} the insolent idol {giving/gave} an autograph 
directly to [very many fans]. 


fw$ 	B 	x 
?? w 
^HU 
\s ? o q o? ^h X^? w 
? 	t 

 M 
? ?? 
\)Q h w
i ) Qo M ? ^q 
	Z
[? w 
$ 	B q^p 
\t l oM h { ? ?` h wi { ^

sono hensyusya-wa mumei-no saka-ga {nante/totemo} takusan-no dokusya-ni hukai kando-o {ataeta-
noda/ataeteiru}-to syupansya-no hensyukaigi-de {iteita/kotyosita-noda} 
that editor-top unknown-gen writer-nom {wh-exc/very} many-gen readers-dat dep impresion-ac 
{gave-exc/gave}-comp publisher-gen editorial meting-at {mentioned/exagerated-exc} 
At the meting at the publisher?s ofice, that editor {mentioned/exagerated the claim} that the unknown 
writer made a strong impresion on [very many readers]. 


fw? ? x 
? 	?s 
	? 	U 
\s? o q o ? ^h X ^?w 

? t 
	? w 
? ? W? 
\?h ? `
hw i ? h ?` o M? ^q 
? q w 
? ? ?p 
\?? ` oM h { 
? ;` h wi { ^

sono bokusi-wa nesina sinzya-ga {nante/totemo} takusan-no hitobito-ni kokoro-no yasuragi-o 
{motarasita-noda/motarasiteiru}-to kyokai-no baza-de {kandositeita/sendensita-noda} 
that clergy-top enthusiastic congregation-nom {wh-exc/very} many-gen people-dat heart-gen peace-ac 
{provided-exc/provided}-comp church-gen bazar-at {was moved/anounced-exc} 
At the church bazar, that clergyman {was moved by/anounced that} the enthusiastic congregation 
{providing/provided} peace of mind to [very many people]. 


fwh H x 
h ?w 
	
U
\s? o q o ? ^h X ^?w 
? ? ?? ? t 
P 
?w 
^
??
\

?
?l hw i 
? 
?lo M? ^q 
2 a qw 
q 	?p 
\9V ` oM h { ? r` o Mh w i{ ^

sono gaka-wa garo-no syuzin-ga {nante/totemo} takusan korekuta-ni nisemono-no sakuhin-o {uriharata-
noda/uriharateiru}-to tenrankai-no kaizyo-de {gekidositeita/gokaisiteita-noda} 
that painter-top galery-gen manager-nom {wh-exc/very} many colector-dat fake-gen product-ac 
{sold-exc/sold}-comp exhibition-gen place-at {got angry/misunderstod-exc} 
At the art exhibition, that painter {got angry at/misunderstod that} the manager of the art galery 
{having/had} sold fake art work to [very many colectors]. 


fw? ? x 
U ?\s 

M U 
\s ? o q o? ^h X^? w 
	? ? t 

? w 
? 	Z ? 
\? D
`o ` ?l h wi ?D ` o` ? lh ^q 
v ? w 
G ? p 
\S r? M oM ? { ? ?M ` hw i { ^

sono rikisi-wa gankona oyakata-ga {nante/totemo} takusan-no sindesi-ni sinya-no gaisyutu-o 
{kyokasitesimata-noda/kyokasitesimata}-to zasi-no kizi-de {odoroiteiru/kantigaisita-noda} 
that sumo wrestler-top stuborn bos-nom {wh-exc/very} many-gen new aprentice-dat mid night-gen 
stayout-ac {alowed-exc/alowed}-comp magazine-gen article-at {is surprized/misunderstod-exc} 
 145 
In the magazine article, that sumo wrestler {is surprised at/misunderstod that} the stuborn sumo bos 
{alowing/alowed} [very many new aprentices] to stay out until midnight. 
 
 
 
 
 146 
 
CHAPTER 7: CONCLUSION 
 
In this thesis, we discussed a variety of empirical facts about exclamatives in English and 
Japanese. By employing Rizi?s (1997) proposal of left periphery structures, we argued 
that Japanese exclamatives involve Mood phrase, Focus phrase, and Finite phrase, and 
that the head of each functional projection is overtly realized. This work showed not only 
that the left / right periphery analysis is a powerful tool which can shed light on the rich 
empirical facts of exclamatives, but also that it can guide us toward discovering further 
empirical facts about exclamatives. Therefore one contribution of this work is in 
providing facts and generalizations regarding exclamative. We further touched upon the 
isue of how the human sentence procesor handles exclamatives. With a few 
experiments, we examined what mechanisms the parser has to have, given a theory of 
phrase structure that contains very complex properties. The conclusion we have reached 
is that the parser is equipped with a mechanism that does not treat al the functional 
projections in the exactly same way. 
 
 
 147 
 
REFERENCES 
 
Abels, Klaus. 2003. Succesive Cyclicity, Anti-locality, and Adposition Stranding. 
Doctoral disertation, University of Connecticut. Storrs, CT. 
Abney, Steven Paul. 1987. The English noun phrase in its sentential aspect. Doctoral 
disertation, MIT, Cambridge, MA. 
Almeida, Diogo. A. d. A. & Masaya Yoshida. 2005. Sluicing in Brazilian Portuguese. 
Unpublished manuscript, University of Maryland, College Park. 
Aoshima, Sachiko, Colin Philips & Amy Weinberg. 2004. Procesing filer-gap 
dependencies in a head-final language. Journal of Memory and Language 51, 
1:23?54. 
Aoshima, Sachiko, Colin Philips & Masaya Yoshida. 2005. The source of the bias for 
longer filer-gap dependencies in Japanese. Paper presented at The 18th Annual 
CUNY Conference on Human Sentence Procesing, University of Arizona, 
Tucson. 
Baltin, Mark & Paul M. Postal. 1996. More on Reanalysis Hypothesis. Linguistic Inquiry 
27, 1:127?145. 
Beck, Sigrid. 1996. Quantified structures as bariers for LF movement. Natural Language 
Semantics 4, 1: 1?56. 
Bennis, Hans. 1995. The meaning of structure: The wat voor construction revisitied. 
Linguistics in the Netherlands 1995, 25?36.  
Bhat, Rakesh & James Yoon. 1991. On the Composition of COMP and Parameters of 
V2. In The proceding of West Coast Conference on Formal Linguistics 10, ed. 
by Dawn Bates. 41?52. Stanford, CA: CSLI Publications. 
Bresnan, Joan. 1970. On complementizers: Toward a syntactic theory of complement 
types. Foundations of Language 6, 3:297-321. 
Bresnan, Joan W. 1972. Theory of complementation in English syntax. Doctoral 
disertation, MIT. 
Bresnan, Joan W. 1973. Syntax of comparative clause construction in English. Linguistic 
Inquiry 4, 3:275?343. 
Cheng, Lisa L.-S. 1991. On the typology of wh-questions. Doctoral disertation, MIT, 
Cambridge, MA. 
Chomsky, Noam. 1973. Conditions on transformations. In A Festschrift for Morris Hale, 
eds. by Stephen R. Anderson & Paul Kiparsky, 232?286. New York, NY: Holt, 
Rinehart and Winston. 
Chomsky, Noam. 1977. On wh-movement. In Formal syntax, eds. by Peter Culicover, 
Thomas Wasow, & Adrian Akmajian, 71?132. New York: Academic Pres. 
Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: MIT Pres. 
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In Step by step: Esays on 
minimalist syntax in honor of Howard Lasnik, eds. by Roger Martin, David 
Michaels & Juan Uriagereka, 89?155. Cambridge, MA: MIT Pres. 
Chomsky, Noam. 2001. Derivation by Phase. In Ken Hale: A life in language, ed. by 
Michael Kenstowicz, 1?52. Cambridge, MA: MIT Pres. 
Chung, Sandra, Wiliam A. Ladusaw & James cCloskey. 1995. Sluicing and logical 
form. Natural Language Semantics 3, 3:239?282. 
 148 
Cinque, Guglielmo. 1990. Types of A?-dependencies. Cambridge, MA: MIT Pres. 
Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: Oxford University 
Pres. 
Conroy, Anastasia. 2005. The semantics of how come: A look at how factivity does it al. 
In University of Maryland Working Papers in Linguistics 14. College Park, MD: 
Department of Linguistics, University of Maryland. 
Corver, Norbert. 1990. The syntax of left branch extractions. Doctoral disertation, 
Tilberg University. Tilberg, Holland. 
Corver, Norbert. 1991. Evidence for DegP. In the Procedings of the North East 
Linguistic Society 21. ed. by Tim Sherer, 3?47. Amherst, MA: GLSA 
Publications. 
Corver, Norbert. 1997. The internal syntax of the Dutch extended adjectival projection. 
Natural Language and Linguistic Theory 15, 2:289?368. 
Corver, Norbert. 1997. Much-support as a last resort. Linguistic Inquiry 28, 1:119?164. 
Craenenbroeck, Jeroen. van. 2004. Elipsis in Dutch dialect. Unpublished doctoral 
disertation, Leiden University, Leiden. 
Crain, Stephen & Janet Dean Fodor. 1985. How can gramars help parsers? In Natural 
language parsing, eds. by David R. Dowty, Lauri Kartunen, & Arnold M. 
Zwicky, 94?128. Cambridge: Cambridge University Pres. 
Cresti, Diana. 1995. Extraction and reconstruction. Natural Language Semantics 3, 1:79?
122. 
Culicover, Peter W. 1999. Syntactic nuts: Hard cases, syntactic theory, and language 
acquisition. Oxford: Oxford University Pres. 
d?Avis, Franz-Josef. 2002. On the interpretation of wh-clauses in exclamative 
environment. Theoretical Linguistics 28, 1:5-31. 
Deguchi, Masanori & Yoshihisa Kitagawa. 2002. Prosody and wh-questions. In 
Procedings of the Thirty-second Annual Meting of the North-Eastern Linguistic 
Society, ed. by Masako Hirotani, 73?92. Amherst, MA: GLSA Publications. 
den Dikken, Marcel. 2003. On the morphosyntax of wh-movement. In Multiple wh-
fronting, eds. by Cedric Boeckx & Kleanthes K.Grohmann, 77?98. Amsterdam: 
John Benjamins. 
den Dikken, Marcel & Anastasia Giannakidou. 2002. From hel to polarity: ?Agresively 
non-D-linked'? wh-phrases as polarity items. Linguistic Inquiry 33, 1:31?61. 
de Vincenzi, Marica. 1991a. Filer-gap dependencies in a null subject language: 
Referential and nonreferential WHs. Journal of Psycholinguistic Research 20, 
3:197?213. 
de Vincenzi, Marica. 1991b. Syntactic parsing strategies in Italian. Dordrecht, The 
Netherlands: Kluwer Academic Publisher. 
Eliott, Dale. 1971. The gramar of emotive and exclamatory sentences in English. 
Doctoral disertation, Ohio State University: Columbus, OH. [also appeared in 
Ohio State Working Papers in Linguistics 8, eds. by Dale Eliot, Michael Geis, 
Alexander Grosu, Bary Novel, Ann Zwicky and Arnold Zwicky, ii?110. 
Eliott, Dale. 1974. The gramar of emotive and exclamatory sentences in English. 
Foundations of Language 11, 231?246. 
Emonds, Joseph E. 1985. A unified theory of syntactic categories. Dordrecht: Folis 
Publications. 
 149 
Fitzpatrick, Justin. 2005. The whys and how comes of presupposition and NPI licensing 
in questions. Procedings of the 24th West Coast Conference on Formal 
Linguistics. eds. by John Alderete, Chung-hye Han & Alexei Kochetov, 138?145. 
Somervile, MA: Cascadila Pres. 
Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, MA: MIT Pres. 
Frampton, John. 1991. Relativized minimality, A review. The Linguistic Review 8, 1: 1?
46. 
Fuji, Tomohiro & Hajime Ono. 2005. English wh-exclamatives and T-to-C movement. 
Manuscript submited for publications. 
Fuji, Tomohiro & Hajime Ono. 2006. Wh-exclamatives and functional heads in Japanese. 
Unpublished manuscript. (University of Maryland, College Park) 
Fukui, Naoki. 1988. LF extraction of naze: Some theoretical implications. Natural 
Language and Linguistic Theory 6, 4: 503?526. 
Fukui, Naoki & Yuji Takano. 1998. Symmetry in syntax: Merge and demerge. Journal of 
East Asian Linguistics 7, 1: 27?86. 
Giannakidou, Anastasia. 1999. Afective dependencies. Linguistics and Philosophy 22, 
4:367?421. 
Grimshaw, Jane. 1979. Complement selection and the lexicon. Linguistic Inquiry 10, 
2:279?326. 
Grimshaw, Jane. 1997. Projection, heads, and optimality. Linguistic Inquiry 28, 3:373?
422. 
Gutierez-Rexach, Javier. 1996. The semantics of exclamatives. In Syntax at sunset, eds. 
by Edward Garet & Felicia Le. 146?162. Los Angeles, CA: UCLA. 
Gutierez-Rexach, Javier. 2001. Spanish exclamatives and the interpretation of the left 
periphery. In Romance languages and linguistic theory 1999: selected papers 
from ?Going Romance? 1999, eds. by Yves D'Hulst, Johan Rooryck & Jan 
Schroten. 167?194. Amsterdam: John Benjamins. 
Hagstrom, Paul. 1998. Decomposing questions. Doctoral disertation, MIT, Cambridge, 
MA. 
Hamblin, Charles Leonard. 1973. Questions in Montague English. Foundations of 
Language 10, 1: 41?53. 
Hankamer, Jorge & Ivan Sag. 1976. Dep and surface anaphora. Linguistic Inquiry 7, 
3:391?426. 
Hara, Yurie. 2006. Gramar of knowledge representation: Japanese discourse items at 
interfaces. Doctoral disertation, University of Delaware. 
Harada, Kazuko I. 1972. Constraints on WH-Q Binding. In Studies in Descriptive and 
Applied Linguistics: Bulletin of the Summer Institute in Linguistics, vol. V, 
180?206. Tokyo: International Christian University. 
Hegarty, Michael. 1992. Adjunct extraction and chain configurations. Doctoral 
disertation, MIT, Cambridge, MA. 
Heycock, Caroline. 1995. Asymmetries in Reconstruction. Linguistic Inquiry 26, 4: 547?
570. 
Hiraiwa, Ken. 2000. On nominative-genitive conversion. In A few from building E39: 
MIT Working Papers in Linguistics 39, eds. by Elena Guerzoni & Ora 
atsushansky, 66?124. Cambridge, MA: MITWPL. 
 150 
Hiraiwa, Ken. 2002. Nominative-genitive conversion revisited. In Japanese/Koran 
Linguistics 10, eds., by Noriko M. Akatsuka & Susan Strauss, 546?558. Stanford, 
CA: CSLI Publications 
Hiraiwa, Ken & Shin-ichiro Ishihara. 2002. Mising links: Cleft, sluicing, and "no da" 
construction in Japanese. In Procedings of HUMIT 2001: MIT Working Papers 
in Linguistics 43, eds. by Tania Ionin, Hejeong Ko & Andrew Nevins, 35?54. 
Cambridge, MA: MITWPL. 
Hornstein, Norbert & Amy einberg. 1981. Case theory and preposition stranding. 
Linguistic Inquiry 12, 1:55?91. 
Huang, C.-T. James. 1982. Logical relations in Chinese and the theory of gramar. 
Doctoral disertation, MIT, Cambridge, MA. 
Huang, C-T. James & C.-S. Luther Liu. 2001. Logophoricity, atitudes and ziji at the 
interface. In Long distance reflexive (Syntax and semantics, vol. 33), eds. by Peter 
Cole, Gabriela Hermon & C.-T. James Huang, 141?196. New York, NY: 
Academic Pres. 
Huang, C-T. James & C.-C. Jane Tang. 1991. The local nature of the long-distance 
reflexive in Chinese, In Long-distance anaphors, eds. by Jan Koster & Eric 
Reuland, 263?282. Cambridge: Cambridge University Pres. 
Inoue, Kazuko. 1976. Henkeibunpoo to Nihongo [Transformational Grammar and 
Japanese]. Tokyo: Taishukan. 
Ishihara, Shinichiro. 2002a. Syntax-phonology interface of wh-constructions in Japanese. 
In Procedings of TCP 2002, ed. Yukio Otsu, 165?189. Tokyo: Hituzi Syobo. 
Ishihara, Shinichiro. 2002b. Invisible but audible wh-scope marking: Wh-constructions 
and deacenting in Japanese. In Procedings of WCFL 21. eds. Line Mikkelsen 
& Chris Potts, 180?193. Somervile, MA: Cascadila Pres. 
Ishii, Yasuo. 1991. Operators and empty categories in Japanese. Doctoral disertation, 
University of Connecticut, Storrs. 
Jackendoff, Ray. 1977. Constraints on Phrase Structure Rules. In Formal Syntax, eds. by 
Peter Culicover, Thomas Wasow & Adrian Akmajian. 249?283. San Diego, CA: 
Academic Pres. 
Johnson, Yuki. 2003. Modality and the Japanese language. Ann Arbor, MI: Center for 
Japanese Studies, University of Michigan. 
Josephs, Luis. S. 1976. Complementation. In Syntax and semantics, vol. 5, Japanese 
generative gramar, ed. by Masayoshi Shibatani, 307?369. New York, NY: 
Academic Pres. 
Just, Marcel Adam, Patricia A. Carpenter & Jacqueline D. Woolley. 1982. Paradigms and 
proceses in reading comprehension. Journal of Experimental Psychology: 
General 111, 2:228?238. 
Kamide, Yuki & Don C. Mitchel. 1999. Incremental pre-head atachment in Japanese 
parsing. Language and Cognitive Proceses 14, 5:631?662. 
Kaplan, Tamar & John Whitman. 1995. The category of relative clauses in Japanese, with 
reference to Korean. Journal of East Asian Linguistics 4, 1: 29?58. 
Kartunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Philosophy 1, 
1: 3?44. 
Katada, Fusa. 1991. The LF representation of anaphora. Linguistic Inquiry 22, 2: 287?
314. 
 151 
Kennedy, Christopher. 1997. Comparison and polar opposition. In Procedings of SALT 
7, ed. by Aaron Lawson. 240?257. Ithaca, NY: CLC Publications. 
Kennedy, Christopher. 2000. Polar oppositions and the ontology of ?degres?. Linguistics 
and Philosophy 24, 1: 33?70. 
Kennedy, Christopher & Merchant, Jason. 2000. Atributive Comparative Deletion. 
Natural Language and Linguistic Theory 18, 1:89?146. 
Kiparsky, Paul & Carol Kiparsky. 1970. Fact. In Progres in Linguistics, eds. by Manfred 
Bierwisch & Karl Erich Keidolph, 143?173. The Haugue: Mouton. 
Kishimoto, Hideki. 2001. Binding of indeterminate pronouns and clause structure in 
Japanese. Linguistic Inquiry 32, 4: 597?633. 
Kis, Katalin. (ed.) 1995. Discourse configurational languages. New York: Oxford 
University Pres. 
Kitagawa, Yoshihisa. 2005. Wh-scope puzzles, In Procedings of the Thirty-fifth Annual 
Meting of the North-Eastern Linguistic Society. Amherst, MA: GLSA 
Publications. 
Kroch, Anthony. 1989. Amount quantification, referentiality, and long wh-movement. 
Unpublished manuscript, University of Pennsylvania. 
Kuno, Susumu. 1972. Functional sentence perspective: A case study from Japanese and 
English. Linguistic Inquiry 3, 3: 269?320. 
Kuno, Susumu. 1973. The structure of the Japanese language. Cambridge, MA: MIT 
Pres. 
Kuroda, Shige-Yuki. 1965. Generative gramatical studies in the Japanese language. 
Doctoral disertation, MIT, Cambridge, MA. 
Kuroda, Shige-Yuki. 1973. Where epistemology, style, and gramar met: a case study 
from Japanese. In A festschrift for Morris Halle, eds by Stephen R. Anderson, & 
Paul Kiparsky, 377?391. New York, NY: Holt, Rinehart and Winston. 
Lasnik, Howard. 1999. On feature strength: Thre minimalist approaches to overt 
movement. Linguistic Inquiry 30, 2:197?217. 
Lasnik, Howard. 2005. Review of Jason Merchant, The Syntax of Silence. Language 81, 
2:259?265. 
Lasnik, Howard & Mamoru Saito. 1984. On the nature of proper government. Linguistic 
Inquiry 15, 2: 235?289. 
Lasnik, Howard & Mamoru Saito. 1992. Move ?. Cambridge, MA: MIT Pres. 
Le, Felicia. 1994. Negative polarity licensing in wh-questions: The case for two 
licensors. Master thesis, UCLA. 
Lipt?k, Anik?. 2001. On the syntax of wh-items in Hungarian. Doctoral disertation, 
Universiteit Leiden. 
Lobeck, Anne. 1995. Elipsis: Functional heads, licensing, and identification. Oxford: 
Oxford University Pres. 
McCawley-Akatsuka, Noriko. 1978. Epistemology and Japanese syntax. In Papers from 
the 14th Regional Meting of the Chicago Linguistic Society, eds. by Donka 
Farkas, Wesley Jacobson and Karol Todrys, 272?284. Chicago Linguistic Society, 
University of Chicago. 
McGloin, Naomi & Hiroko Terakura. 1978. On the asertive predicate No Desu in 
Japanese. In Papers from the 14th Regional Meting of the Chicago Linguistic 
 152 
Society, eds. by Donka Farkas, Wesley Jacobson & Karol Todry, 285?296. 
Chicago Linguistic Society, University of Chicago. 
May, Robert. 1985. Logical Form. Cambridge, MA: MIT Pres. 
elvold, Janis. 1991. Factivity and Definitenes. ore Papers on Wh-Movement (MIT 
Working Papers in Linguistics 15), eds. by Lisa Cheng and Hamida Demirdache. 
97?117. Cambridge, MA: MITWPL. 
Merchant, Jason. 2001. The syntax of silence. Oxford: Oxford University Pres. 
erchant, Jason. 2002. Swiping in Germanic. In Studies in comparative Germanic syntax, 
eds. by J.-W. C. Zwart & W. Abraham, 289?315. Amsterdam: John Benjamins. 
Miyamoto, Edson T. 2002. Case marker as clause boundary inducers in Japanese. Journal 
of Psycholinguistic Research 31, 4:307?347 
Miyamoto, Edson T. & Shoichi Takahashi. 2002. The procesing of wh-phrases and 
interogative complementizers in Japanese. In Japanese / Korean linguistics 10, 
eds., by Noriko Akatsuka, Susan Strauss, & Bernard Comrie, 62?75. Stanford, 
CA: CSLI Publications. 
Motomura, Mitsue. 2003. The thematic roles of sentential to/ko complements in 
Japanese/Korean. In Japanese / Korean Linguistics 11, ed. by Patricia M. Clancy, 
439?454. Stanford, CA: CSLI Publications. 
Murasugi, Keiko. 1990. Noun phrases in Japanese and English: A study in syntax, 
acquisition and learnability. Doctoral disertation, University of Connecticut, 
Storrs. 
Nakau, Minoru. 1973. Sentential complementation in Japanese. Tokyo: Kaitakusha. 
Nelson, Nicole. 1997. The structure of exclamatives in English. Unpublished manuscript, 
Rutgers University. 
Nishigauchi, Taisuke. 1986. Quantification in syntax. Doctoral disertation, University of 
Masachusets, Amherst. 
Nishigauchi, Taisuke. 1990. Quantification in the theory of grammar. Dordrecht: Kluwer 
Academic. 
Nishigauchi, Taisuke. 1999. ?Point of view? and phrase structure. Theoretical and 
Applied Linguistics at Kobe Shoin 2: 49?60. 
Nishigauchi, Taisuke. 1999b. Quantification and wh-constructions. In A handbook of 
Japanese linguistics, ed. by Natsuko Tsujimura, 268?296. New York: Blackwel. 
Obenauer, Hans Georg. 1994. Aspects de la syntaxe A-bare. Doctoral disertation, 
Universit de Paris VII, Paris. 
Oda, Toshiko. 2002. Exclamatives and negative islands. In Procedings of Chicago 
Linguistics Society 38 (volume 2), eds. by Mary Andronis, Erin Debenport, Anne 
Pycha & Keiko Yoshimura, 97?109. 
Oda, Toshiko. 2003. Wh-movement of exclamatives. Nanzan Working Papers in 
Japanese Language 10, 85?123. 
Ono, Hajime. 2002a. An emphatic particle DA and exclamatory sentences in Japanese. In 
Georgetown Working Papers in Theoretical Linguistics 2 [Proceding of the 
Workshop on the Syntax-Semantics Interface in the CP-domain], eds. by Simon 
Mauck & Jenny Mitelstaedt, 211?246. Department of Linguistics, Georgetown 
University. 
 153 
Ono, Hajime. 2002b. Exclamatory sentences in Japanese: A preliminary study. In The 
procedings of the third Tokyo Conference on Psycholinguistics, ed. by Yukio 
Otsu, 305?326. Tokyo: Hituzi Syobo Publ. 
Ono, Hajime. 2003. Surprise, Surprise!: (Un)availability of Multiple Exclamatives, Paper 
presented at Symposium on Indeterminate Pronouns; the Annual Meting of 
English Linguistics Society in Japan, Shizuoka Prefectural University, Japan. 
Ono, Hajime. 2004. Why doesn't English Alow Multiple Exclamatives? Paper presented 
at the 27th GLO Colloquium 2004, Aristotle University of Thesaloniki, Grece. 
Ono, Hajime & Tomohiro Fuji. 2006. English wh-exclamatives and the role of T-to-C in 
wh-clauses. In University of Maryland Working Papers in Linguistics 14, 164?
189. College Park, MD: Department of Linguistics, University of Maryland. 
Ono, Hajime. 2002. Exclamatory Sentences in Japanese. In The Procedings of the Third 
Tokyo Conference on Psycholinguistics. ed. by Yukio Otsu. 305?326. Hituzi 
Syobo Publisher, Tokyo. 
Ono, Hajime, Masaya Yoshida, Sachiko Aoshima, & Colin Philips. (to appear). Real-
time computation of Japanese exclamatives and the strength of locality biases in 
sentence comprehension. Cognitive Studies: Bulletin of the Japanese Cognitive 
Science Society 13, 3. 
Pesetsky, David. 1987. Wh-in-situ: Movement and unselective binding. In The 
Representation of (In)definitenes, eds. by Eric J. Reuland & Alice G.B. ter 
Meulen. 98?129. Cambridge, MA: MIT Pres. 
Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge, MA: MIT Pres. 
Pesetsky, David & Esther Torrego. 2001. T-to-C movement: Causes and consequences. 
In Ken Hale: A life in language, ed. by Michael Kenstowicz, 355?426. Cambridge, 
MA: MIT Pres. 
Reis, arga. 1999. On sentence types in German: An enquiry into the relationship 
betwen gramar and pragmatics. Interdisciplinary Journal for Germanic 
Linguistics and Semiotic Analysis 4, 2:195?236. 
Riemsdijk, Henk. van. 1978. A case study in syntactic markednes: The binding nature of 
prepositional phrases. Dordrecht: Foris. 
Rizi, Luigi. 1990. Relativized minimality. Cambridge, MA: MIT Pres. 
Rizi, Luigi. 1997. The fine structure of the left periphery. In Elements of grammar, ed. 
Liliane Haegeman, 281?337. Dordrecht: Kluwer. 
Rosen, Carol. 1976. Gues what about? In Papers from the 6th meting of the North 
Eastern Linguistics Society, eds. by Alan Ford, John Reighard & Rajendra Singh. 
205?211. Montreal: Montreal Working Papers in Linguistics. 
Ross, John Robert. 1967. Constraints on variables in syntax. Doctoral disertation, MIT, 
Cambridge, MA. 
Ross, John Robert. 1969. Gues who? In Papers from the fifth regional meting of the 
Chicago Linguistic Society, eds. by Robert I. Binnick, Alice Davison, Georgia 
Gren & Jery L Morgan. 252?286. Chicago, IL: Chicago Linguistic Society, 
University of Chicago. 
Ross, John Robert. 1984. Inner islands. In Procedings of the 10th Annual Meting of the 
Berkeley Linguistics Society, eds. by Claudia Brugman & Monica acaulay. 258?
265. Berkely, CA: Berkeley Linguistics Society. 
 154 
Saito, Mamoru. 1989. Scrambling as semanticaly vacuous A?-movement. In Alternative 
conceptions of phrase structure, eds. by Mark. R. Baltin & Anthony S. Kroch. 
182?200. Chicago: University of Chicago Pres. 
Sano, Masaki. 2002. What derives long-distance scope: The case of only in English and 
dake in Japanese. Ritsumeikan Bungaku 576: 293?302. 
Schafer, Wolfram. 2002. Kakari Musubi, Noda-constructions, and how 
gramaticalization theory mets formal gramar. In Japanese / Korean 
Linguistics 10, ed. by Noriko Akatsuka, 320?333. Stanford, CA: CSLI 
Publications. 
Sels, Peter. 1987. Aspects of logophoricity. Linguistic Inquiry 18, 3: 445?479. 
Shimoyama, Junko. 2001. Wh-constructions in Japanese. Doctoral disertation. 
University of Masachusets, Amherst. 
Speas, Margaret. 2000. Person and Point of View in Navajo. In WCFL papers in honor 
of Ken Hale, ed. by Eloise Jelinek, xx?xx. Cambridge, MA: MIT Pres. 
Sprouse, Jon. 2005. The acent projection principle: Why the hel not? In Procedings of 
the 29th Penn Linguistic Colloquium (UPenn working papers in linguistics 12-1) 
(pp. xx?xx). Philadelphia, PA: Department of Linguistics, University of 
Pennsylvania. 
Stowe, Laurie. A. 1986. Parsing Wh-constructions: Evidence for on-line gap location. 
Language and Cognitive Proceses 1, 3:227?245. 
Svenonius, Peter. 1994. The structural location of the atributive adjective. In The 
Procedings of the Twelfth West Coast Conference on Formal Linguistics, eds. by 
Eric Duncan, Donka Farkas and Philip Spaelti, 439?454. Stanford, CA: CSLI 
Publications. 
de Swart, Henri?te. 1992. Intervention efects, monotonicity and scope. In Procedings 
of the Second Conference on Semantics and Linguistic Theory (The Ohio State 
University Working Papers in Linguistics no. 40), eds. by Chris Barker & David 
Dowty. 387?406. Columbus, OH: OSU Department of Linguistics. 
Tada, Hiroaki. 2002. On the lexical categorial features of functional categories. The 
Procedings of the Twentieth National Conference of the English Linguistic 
Society of Japan, 69?72. 
Takahashi, Daiko. 1993. Movement of wh-phrases in Japanese. Natural Language and 
Linguistic Theory 11, 4:655?678. 
Takahashi, Daiko. 2002. Determiner raising and scope shift. Linguistic Inquiry 33, 
4:575?615. 
Takeda, Kazue. 1999. Multiple headed structure. Doctoral disertation. University of 
California, Irvine. 
Tenny, Carol. To appear. Evidentiality, experiencers, and the syntax of sentience in 
Japanese. Journal of East Asian Linguistics. 
Tenny, Carol & Margaret Speas. 2003. Configurational properties of point of view roles. 
In Asymetry in grammar, ed. by Anna Maria Di Scuilo, 315?34. Amsterdam: 
John Benjamins. 
Tokieda, Motoki. 1950. Nihon Bunpoo: Koogo hen. Tokyo: Iwanami. 
Torrego, Esther & Juan Uriagereka. 2002. Parataxis. In Derivations: Exploring the 
dynamics of syntax, ed. by Juan Uriagereka, 253?265. London, England: 
Routledge. 
 155 
Uchibori, Asako. 2000. The syntax of subjunctive complements: Evidence from Japanese. 
Doctoral disertation. University of Connecticut, Storrs. 
Uchibori, Asako. 1997. Opacity and subjunctive complements in Japanese. In Japanese / 
Korean Linguistics 6, eds. by Ho-min Sohn & John Haig. 399?414, Stanford, CA: 
CSLI Publications. 
Ura, Hiroyuki. 2000. Checking theory and grammatical functions in Universal Grammar. 
Oxford: Oxford University Pres. 
Uriagereka, Juan. 1995. An F position in Western Romance. In Discourse 
configurational languages, ed. by Katalin ? Kis, 153?175. New York: Oxford 
University Pres. 
Vilalba, Xavier. 2001. The right edge of exclamative sentences in Catalan. Catalan 
Working Papers in Linguistics 9, 119?135. 
Vilalba, Xavier. 2003. An exceptional exclamative sentence type in Romance. Lingua 
113:713?745. 
Watanabe, Akira. 1992a. Subjacency and S-structure movement of wh-in-situ. Journal of 
East Asian Linguistics 1, 3: 255?291. 
Watanabe, Akira. 1992b. Wh-in-situ, subjacency, and chain formation. MIT Ocasional 
Papers in Linguistics #2. MITWPL. 
Watanabe, Akira. 1993, Larsonian CP recursion, factive complements, and selection The 
Procedings of the North East Linguistic Society 23. ed. by Amy Schafer. 523?
537. GLSA Publications: Amherst, MA. 
Watanabe, Akira. 1996. Nominative-genitive conversion and agrement in Japanese: A 
cross-linguistic perspective. Journal of East Asian Linguistics 5, 4: 373?410. 
Watanabe, Akira. 2003. Wh and operator constructions in Japanese. Lingua 113, 4?6: 
519?558. 
Yatsushiro, Kazuko 1996. Case, Scope and Feature Movement. In Formal Approaches to 
Japanese Linguistics 2 (MIT Working Papers in Linguistics 29), eds. by 
Masatoshi Koizumi, Masayuki Oishi, & Uli Sauerland. 319?335, MITWPL, 
Cambridge, MA. 
Yoshida, Keiko & Tomoyuki Yoshida. 1996. Question Marker Drop in Japanese, ICU 
Language Research Bulletin 11, 37?54. 
Yoshida, M., Aoshima, S., & Philips, C. (2004). Relative clause prediction in Japanese. 
Paper presented at The 17th Annual CUNY Sentence Procesing Conference, 
University of Maryland, College Park. 
Zanuttini, Rafaela & Paul Portner. 2003. Exclamative clauses: At the syntax-semantics 
interface. Language 79, 1: 39?81.