Japanese pseudoclefts

Examples (1a-b) both consist of a grammatical subject (XP1) and its predicate (XP2) with a copula between them; they differ from each other only in the form of XP1. It is common to assume that only the construction in (1a) is categorized into the class of pseudoclefts (e.g., Higgins, 1979). But I follow Akmajian (1970) among others, who regards the construction in (1b) as a type of pseudocleft as well. Although I discuss this analysis and the difference between (1a-b) in Section 2, I note here that (1a-b) are respectively called a wh-cleft and th-cleft because their XP1 starts with wh and th, respectively. In Japanese, sentences like (2) are often referred to as a pseudocleft. The structure of (2) reflects the analysis of Hoji (1990).2,3,4

In the literature, de is often referred to as either a postposition (e.g., Nakayama, 1988) or copula in the sense of an element necessary to establish a predication relation (e.g., Bloch, 1946;Nishiyama, 1999).Importantly, whichever analysis turns out to be correct, it is reasonable to assume that de is a functional category serving to accommodate the two elements.Therefore, this paper assumes that de is a RELATOR. 5s for -ar, I assume in the same lines with Nishiyama (1999) and Urushibara (1993) that it is a verb, to which a tense marker is attached.On the basis of these assumptions, the schematic structure of so-called Japanese pseudoclefts can be represented roughly, as in (4).
(4) Schematic structure of so-called Japanese pseudoclefts XP 1 and XP 2 originate in the specifier and complement of de in RP head, and so-called pseudoclefts in Japanese derive when the XP 1 moves to Spec TP.I eventually argue in Section 3 that so-called Japanese pseudoclefts are th-clefts corresponding to (1b) in English.
The goal of this paper is to demonstrate that so-called pseudoclefts in Japanese deserve being called pseudoclefts.More specifically, I will demonstrate that they are th-clefts.The paper contributes to the linguistic literature basically in two respects.First, this is, to my knowledge, the first literature that clearly mentions why so-called pseudoclefts in Japanese can be assumed to be pseudoclefts that correspond to pseudoclefts in other languages.Despite the term pseudocleft being well known in the Japanese literature, Japanese pseudoclefts have not captured much attention, and their existence has been assumed without much justification.For example, although Hiraiwa and Ishihara (2012) call sentences like (2) pseudoclefts, it has not been clarified in what sense they are pseudoclefts, or whether they share some properties with pseudoclefts in other languages.In fact, as recently as 2014, Park (2014) says, "it is a historical accident that sentences such as [(2)] are called pseudoclefts."He mentions this when he uses the term pseudocleft for Japanese and Korean for the first time.Importantly, he mentioned this to clarify that Japanese and Korean do not have a construction corresponding to English pseudoclefts, contrary to my claim.Therefore, in the rest of this paper, I consistently use the term pseudocleft to refer to constructions satisfying the definition of pseudoclefts in the next section, and 'conventional' pseudocleft to refer to so-called pseudoclefts in Japanese until it is demonstrated that they are indeed pseudoclefts corresponding to pseudoclefts in other languages.
Second, this paper also contributes to the study of other syntactic phenomena.For one, the conventional pseudoclefts that I discuss in this paper are a particular type of copular sentence,6 so this paper contributes to the study of copular sentences in general.This is because conventional pseudoclefts turn out to be pseudoclefts, and potentially carries fertility for subsequent investigations of Japanese pseudoclefts.It follows that the findings of Japanese pseudoclefts in future research can contribute to the literature on copular sentences as well as pseudoclefts, as it is said that "the analysis of pseudoclefts plays an important role in the evaluation of proposal for the treatment of copular sentences," (Partee, 1998, p. 2).Also, it has been proposed that elliptical constructions known as sluicing and stripping in Japanese are derived from a cleft/conventional pseudocleft (e.g., Fukaya and Hoji, 1999;Hiraiwa and Ishihara, 2012) and that Malagasy sluicing is derived from a pseudocleft (Potsdam, 2007).Thus, the conclusion that conventional pseudoclefts are indeed pseudoclefts enables the study of the sluicing/stripping to be conducted more efficiently because the authors can now take account of various properties of pseudoclefts proposed for many languages.
The organization of this paper is as follows.Section 2 first provides the definition of pseudoclefts and some other related terminologies.Section 3 then demonstrates that conventional pseudoclefts in Japanese are th-clefts based on that definition.Finally, Section 4 concludes.

Definition of pseudoclefts
The definition of the term pseudocleft is often not very clear in the literature even though various 'characteristics' of the construction are discussed.Thus, I propose a clear definition of the construction in this section.Essentially, a crucial property of pseudoclefts is that they involve a type of relative construction in XP 1 .So I first briefly discuss a definition of relative constructions, and some terminologies having to do with the constructions.Then, the following section defines pseudoclefts.

Relative constructions.
A definition of relative constructions is given in (5). 75) Definition of relative constructions (Lehmann, 1986, p. 2) "A relative construction is a construction consisting of a nominal (or a common noun phrase, in the terms of categorical grammar), which may be empty, and a subordinate clause interpreted as attributively modifying the nominal.The nominal is called the head and the subordinate clause the In (6), the word in bold is the head, and the embedded clause modifies the head as a relative clause.Given that the head is outside the relative clause, this type of relative construction is called a head-external relative.The position within the relative clause that corresponds to the head is empty as indicated by [e].I call this sort of position a gap.As far as what has been mentioned about (6a-b) so far is concerned, they are identical.But they differ in whether the relative clause involves that or which.In this paper, I call them relative markers (e.g., Romaine, 1980) as elements that typically demarcate the beginning of relative clauses in VO languages (e.g.Herrmann, 2003).8The head of relative constructions does not always appear outside the relative clause.While it sometimes appears inside the relative clause (also known as a head-internal relative), there is a case where an overt head cannot be found.The latter type of construction is known as a free relative; an example is given in (7).Sentence (7) corresponds to (6a-b).Just like (6a-b), the complement of bought is empty, and a relative clause starting with the relative marker what exists.But there is no overt element between the matrix verb and relative clause.Thus, ( 7) is justified as a free relative.

Pseudoclefts.
In light of the definition of relative constructions in Section 2.1, I define pseudoclefts as in (8), in the same spirit as Collins (1991).9(8) Definition of pseudoclefts a. Pseudoclefts consist of XP 1 and XP 2 , where XP 1 is a relative construction with a gap.b.For each pseudocleft, there is a non-clefted sentence consisting of XP 2 and materials in the relative clause that follows the relative marker.
th-cleft (9a-b) are both copular sentences where what Herman bought or the thing which Herman bought refers to XP 1 and that tarantula refers to XP 2 .In (9a), XP 1 is a free relative whose relative marker is what.In (9b), XP 1 is a head-external relative, whose head is thing followed by a relative marker which.Thus, (9a-b) satisfies (8a).I define wh-clefts as pseudoclefts with a free relative in XP 1 , and th-clefts as pseudoclefts with a head-external relative in XP 1 .Thus, (9a-b) are candidates for a wh-cleft and th-cleft, respectively, at this moment.
Essentially, the satisfaction of (8b) indicates the intimate relation between the gap in XP 1 and XP 2 .Thus, pseudoclefts like (9a-b) look as if they are derived by 'cleaving' a non-clefted sentence like (10a) and (10c) into XP 1 and XP 2 in (10b) and (10d).In other words, these sentences have the property of clefting, and thus sentences like (9a-b) are called pseudo'clefts.' On the other hand, the satisfaction of (9a) ensures that they are 'pseudo'clefts, as opposed to clefts.An example of English clefts that corresponds to pseudoclefts in ( 9) is given in (11), for the sake of comparison.

(11) Cleft
It is that tarantula that Herman bought.

Existence of pseudoclefts in Japanese
In this section, I demonstrate the existence of pseudoclefts in Japanese in reference to the definition of pseudoclefts in (8).It turns out that Japanese pseudoclefts have hybrid features of English wh-clefts and thclefts (i.e., XP 1 has the properties of both types of English pseudoclefts).I argue that they are more similar to English th-clefts.
Section 3 consists of three subsections.Section 3.1 first shows that Japanese conventional pseudoclefts, which are exemplified here again as ( 12), involve a head-external relative in XP 1 and thus satisfy (8a).
3.1.XP 1 = head-external relative.The goal of Section 3.1 is to demonstrate that XP 1 in sentences like ( 12) is a head-external relative.More specifically, I propose (13) as the structure of the XP 1 in question.The claim in ( 13) is that no is a complementizer and that it projects a relative clause modifying the head of the relative construction, which is phonologically null.Given that there is no overt head, the structure in (13) has the property of free relative.But the structure does not involve a wh-item, and instead includes an overt complementizer, which overtly appears in th-clefts in English.Thus, I will assume that XP 1 in Japanese conventional pseudoclefts is a head-external relative clause.It should be pointed out clearly that the structure in (13) is novel.Consider, for example, ( 14), where Hoji's (1990) and Hiraiwa and Ishihara's (2012) structures of conventional pseudoclefts are given.10(14) a. (Hoji, 1990, (225a) My analysis of the whole structure of conventional pseudoclefts will be provided later.Here, we focus on the difference in the structures of XP 1 among (13), (14a) and (14b).First, a unique property of the XP 1 in (14a) is the category of no.As mentioned above, although Hoji (1990) does not clarify the exact identity of no, he assumes it is a nominal element.This is in contrast to the analysis of no in ( 13) and (14b), where it is assumed as a complementizer.As for the XP 1 in (14b), its unique property is that XP 1 is a CP, contrary to the XP 1 of ( 13) and (14a) being an NP.Given those differences among (13-14), two hypotheses need to be proven in demonstrating that ( 13) is the correct structure of XP 1 ; (a) no is a complementizer although various lexical and functional items are realized as no in Japanese, and (b) XP 1 is nominal in nature although the NP head is not spelled out overtly.To this end, Section 3.1 is divided into three subsections.Section 3.1.1shows that no in Japanese serves as a genitive case, pronoun, or complementizer, and eliminate the possibility that no in Japanese conventional pseudoclefts is a genitive case marker.Section 3.1.2demonstrates that the relevant no is a complementizer.Section 3.1.3argues that there is a null NP head in XP 1 .

'
No' ̸ = Genitive case marker.It has been proposed that various lexical and functional items are realized as no in Japanese such as genitive case, pronoun and complementizer (e.g., Murasugi, 1991).Consider first a genitive case marker no in (15).( 15) Taroo-no Taro-GEN pen pen 'Taro's pen' (Murasugi, 1991, (1), p. 21) A genitive marker no is inserted between two nominal phrases, and the nominal phrase with no modifies the other nominal phrase.Also, a remarkable difference between no as a genitive marker and other two kinds of nos is that no in the genitive construction has long been in existence unlike the others (Fujino, 2013).
Next, consider a pronoun no in (16a-b).
(16) a. siro-i white-NPST no one 'the one which is white' (Murasugi, 1991, (87a), p. 72) b. hasit-tei-ru run-STV-NPST no one 'the one which is running' (Murasugi, 1991, (87d), p. 72) Roughly speaking, no as a pronoun corresponds to the English indefinite pronoun one.But unlike the one in English, the Japanese pronoun no requires a modifier.Thus, it is always modified by an AP as in (16a) or relative clause as in (16b).11On the other hand, pronoun no and its associated noun never modifies another noun unlike an NP with genitive no.Thus, pronoun no is distributed differently from genitive case marker no.C 'Did Taro go back?' (Murasugi, 1991, (ii), p. 100) Sentence ( 17a) is a cleft construction.It is known that Japanese clefts and conventional pseudoclefts are similar.Consider a pseudocleft corresponding to (17a).
The one Yamada met was Russell.' (Murasugi, 1991, (142a), p. 93) Sentences (17a) and ( 18) differ only in whether Russell is case-marked or not.Abstracting away from the exact analysis of Japanese clefts, I note that no in (17a) is widely accepted as a complementizer and XP 1 is a CP (e.g., Hoji, 1990;Kizu, 1999;Hiraiwa and Ishihara, 2012). 13As for no in (17b), this is a question particle used in colloquial speech, and it corresponds to a more well-known question particle ka in formal speech.In this way, no in Japanese is used as a genitive case, pronoun, or complementizer.
The rest of this section is devoted to excluding the possibility that no in conventional pseudoclefts is a genitive case marker.A crucial piece of evidence for this claim comes from the different realization of a genitive case marker from that of pronouns and complementizers in the Toyama dialect of Japanese.In this dialect, although genitive case is still realized as no, no as a pronoun or complementizer in Tokyo dialect is realized as ga (19a-c).
(19) a. siroi white ga one 'the one which is white' (Murasugi, 1991, (86a), p. 72) b. hasit-tei-ru run-STV-NPST ga one 'the one which is running' (Murasugi, 1991, (86d) (Murasugi, 1991, (147a), p. 95) Notice that no in ( 18) is replaced by ga in (20).The null hypothesis would be that no in conventional pseudoclefts in the Tokyo dialect would also be either a pronoun or complementizer rather than a genitive case marker, as in the Toyama dialect. 15In fact, the same distribution of no and ga is reported for the Kochi dialect of Japanese as well (Takeda, 1999).Therefore, all else being equal, it is reasonable to eliminate the possibility that the no in question is a genitive case marker.
3.1.2'No' = complementizer.Having established that no in conventional Japanese pseudoclefts is not a genitive case marker, I argue in this section that it is more reasonable to assume it is a complementizer rather than a pronoun.The first piece of evidence comes from the availability of the use of honorific expressions in Japanese conventional pseudoclefts.It has been proposed that pronoun no cannot refer to an individual who is socially superior to the speaker (e.g., Harada, 1976).Compare first (21a) with (21b).The difference between (21a-b) is only whether the relative clause head is an R-expression hito or pronoun no.Note that both sentences are grammatical.With this in mind, compare next (22a) with (22b).
13 Some arguments for this claim will be given below. 14 Murasugi (1991) claims in her footnote 49 that no in (17b) is also replaced by ga in Toyama dialect. 15Since ga in the Toyama dialect is construed as a pronoun, no in the Tokyo dialect seems to be a pronoun, as Murasugi (1991) argues.However, the substitution of no with an R-expression leads to ungrammaticality unlike no in (16a-b).So it seems more plausible to assume it marks genitive case.However, this does not necessarily mean that ga is a genitive case as well.This is because the genitive case no can appear overtly.
(22) (Murasugi, 1991, (149) The difference between (22a-b) is also reduced only to whether the relative clause head is hito or no.But (22a-b) differ from (21a-b) in that they involve the honorific expression orare.Accordingly, (22b) is ungrammatical unlike (21b).Thus, it is assumed that the ungrammaticality is attributed to the incompatible use of no with an honorific expression.
Given this restriction of the use of pronoun no, it is predicted that if no in Japanese conventional pseudoclefts is a pronoun, conventional pseudoclefts cannot involve an honorific expression.But this prediction is not borne out, as shown in ( 23), where the sentence structure reflects the assumption that no is a pronoun.desu is 'The one who is eating there is Prof. Tanaka.' (Murasugi, 1991, (150), p. 96) Note that ( 23) is grammatical even though no is used with orare.Therefore, sentences like ( 23) indicate that no in Japanese conventional pseudoclefts is not a pronoun, either.
In addition, there is a positive evidence to support that no in conventional pseudoclefts is a complementizer.Murasugi (1991) investigates the head-external relatives in child speech, and shows that no as a complementizer is spelled out as in (24). 1624) a. buta-san pig-DIM tata-ite-ru hit-STV-NPST (*no) C taiko drum 'the drum that the piggy is playing' (Murasugi, 1991, (20a) C wanwa doggie 'a doggie that is holding a flower' (Murasugi, 1991, (23a), p. 214) In ( 24), no appears between the relative clause and head.At this moment, there is a possibility that no in (24) is not a complementizer, but either a genitive case marker or pronoun.But Murasugi (1991) eliminated this possibility.First, she refers to data in the Toyama dialect, which is similar to (24).
(25) (Murasugi, 1991, (15) Note that in Toyama dialect, the child speech involves ga (either pronoun or complementizer) instead of no (genitive case) between the relative clause and head.The null hypothesis would be that no in (24) is at least not a genitive case marker, but either a pronoun or complementizer.This is important data.Without (25), one may claim that no in ( 24) is just the same type of overgeneralization as the second language learners' overgeneration of no such that they insert genitive no whenever one element modifies the other.
As for the possibility of the no in question being a pronoun, this is not very likely either, given that there is no referent for no in the contexts where (24a-b) are uttered. 17But Murasugi (1991) eliminates the possibility of no being a pronoun in a different way; that is, if no is a pronoun, no as a genitive case marker needs to be inserted between the no and relative clause head.Given that children never fail to insert the no as a genitive case in the relevant position (Murasugi, 1991), it is reasonable to assume that no in ( 24) is not a pronoun.It is then most plausible to assume that no as a complementizer is involved in relative constructions, although it is not overtly realized in adult speech. 18Therefore, the data in child speech also supports that no in Japanese conventional pseudoclefts is a complementizer.It follows that the correct structure of XP 1 is either Hiraiwa and Ishihara's (2012) or my structure rather than Hoji's structure.In (26), each author's XP 1 structure is repeated from (13-14).( 26 Hoji (1990) In the next section, I demonstrate that (26b) fares better than (26a).
3.1.3Arguments for the presence of a phonologically null noun in XP 1 .Given that no in Japanese conventional pseudoclefts is a complementizer, a straightforward assumption is that XP 1 is a CP, as in (26a).This is because Japanese is a strictly head-final language, and no appears at the right edge of XP 1 .In this section, however, I argue for the presence of the null noun projecting XP 1 as a nominal expression.To this end, I first demonstrate that XP 1 in Japanese conventional pseudoclefts is nominal in nature.The following subsection then provides supporting arguments for the presence of a null noun.
XP 1 = nominal phrase.In this section, I argue that XP 1 in Japanese conventional pseudoclefts is a nominal phrase based on three diagnostics.Although the focus of this section is on Japanese conventional pseudoclefts, cleft data are also presented.This is for making clear the contrast between XP 1 in conventional pseudoclefts and clefts; that is, even though they have the same superficial forms, XP  (Kizu, 1999, (31b), p. 98) One of the diagnostics is the availability of the XP 1 substitution.It is predicted that if XP 1 in (27a) is a nominal, it can be substituted by a semantically similar head-external relative with an overt head.On the other hand, such a substitution is not necessarily available for the XP 1 in (27b).This is because XP 1 in (27b) is a CP.With this in mind, consider (28a-b). 17Since the contexts for these utterances are relatively long, I refer the reader to (Murasugi, 1991, pp. 212-214). 18 Murasugi (1991) proposes an IP hypothesis such that each relative clause in Japanese is an IP (TP).For example, she supports this claim based on sentences that are ungrammatical due to the violation of the Empty Category Principle formulated by Lasnik and Saito (1984).I refer the reader to Section 3.4 in Murasugi (1991) for the relevant discussion.But her argument cannot necessarily be carried over to relative clauses in conventional pseudoclefts.Besides, crucially, although she does not differentiate pseudoclefts from clefts and call both types of sentences clefts, she does assume that no in conventional pseudoclefts (e.g., ( 18)) is a complementizer.Thus, my conclusion that no in conventional pseudoclefts is a complementizer does not contradict Murasugi's analysis. 19At of at-ta is an allomorph of ar-.The phonological shape of at is caused by the following affix in the process of a sound change known as onbin.In (28a-b), XP 1 is substituted by a head-external relative headed by hito.As predicted above, (28a) is grammatical whereas (28b) is ungrammatical.Thus, the availability of the XP 1 substitution suggests that XP 1 in conventional pseudoclefts is a nominal.
Another diagnostic to identify the syntactic category of XP 1 is the availability of nominative-genitive conversion (NGC).It has been reported that the subject of the prenominal sentential modifier can be not only nominative case-marked but also genitive case-marked (e.g., Harada, 1976;Watanabe, 1996).Consider first (29), where the subject in (28a) is genitive case-marked.In (28a), the sentence Kare-ga deat-ta is modifying the noun hito.Thus, NGC is available for (28a) as in (29).On the other hand, Kare in (30) cannot be genitive case-marked, because the sentence does not modify any noun.Note that (31a) is grammatical while (31b) is reasonably ungrammatical because its XP 1 is a CP.This indicates that in (31a), there is a noun that Kare-no deat-ta modifies.Thus, the availability of the NGC also indicates that XP 1 in Japanese conventional pseudoclefts is a nominal.In fact, the NGC does support the existence of the null noun in XP 1 in Japanese conventional pseudoclefts, which is modified by [kare-no deat-ta no].
The third diagnostic is the availability of using the combination of a numeral quantifier (NQ) and classifier (CL) in XP 1 .As it is known that a particular CL is chosen based on the properties of entities denoted by its host noun, a NQ-CL combination is used with a noun (e.g., Amazaki, 2005).Thus, it is predicted that an NQ-CL can be used in XP 1 of conventional pseudoclefts but not in XP 1 of clefts.With this in mind, consider (32).Whereas an NQ-CL go-nin can be used in (32a), it cannot in (32b).Therefore, the availability of an NQ-CL also indicates that XP 1 in Japanese pseudoclefts is a nominal.Also, this diagnostic indeed supports for the presence of a null noun too.This is because morphological form of a CL is determined by its host noun.
Attested examples of the null head of relative constructions.So far, I have demonstrated that XP 1 in Japanese conventional pseudoclefts is a nominal based on three diagnostics.Also, two of them supported the existence of a null noun heading the nominal.This line of analysis such that null noun is assumed to exist despite its not being spelled out overtly is reminiscent of the discussion about English free relatives.As we saw previously, English free relatives such as (33a) look similar to indirect questions such as (33b).Crucially, however, free relatives and indirect questions are different in their distributions.As shown in (33a), free relatives can be the complement of verbs like buy that is subcategorized for direct object NPs (e.g., Bresnan and Grimshaw, 1978).Thus, the free relative in (33a) is considered a nominal unlike that in (33b), which is a CP.As for the position of a wh-item in free relatives, the most widely adopted hypothesis is called COMP Hypothesis, which was first introduced by Kuroda (1968) and supported by many authors (e.g., Hirschb ühler, 1978;Groos and Van Riemsdijk, 1981;Rooryck, 1994).On the COMP Hypothesis, the wh-item occupies Spec CP.But since the relative clause is considered a nominal phrase, it is assumed that there is a phonologically null noun in the head position of the relative construction.Note that this line of analysis is similar to my analysis of Japanese pseudoclefts, although what is overtly realized in Japanese pseudoclefts is a CP head rather than a specifier of CP.Therefore, it is not implausible to assume for a phonologically null noun in XP 1 of Japanese conventional pseudoclefts. 20oreover, further supporting evidence for the presence of a null noun in Japanese conventional pseudoclefts comes from data in Classical Japanese.Interestingly, head-external relatives in Classical Japanese allow their head to be phonologically null.To begin with, consider (34) to confirm that Classical Japanese has head-external relative constructions, just like in Modern Japanese.In ( 34), the bracketed phrase is the relative clause with a gap indicated by e, and there is a head of the relative construction to the right of the relative clause.One thing that is different between relative constructions in Classical and Modern Japanese is the conjugation form of the verb in relative clause.In Modern Japanese, there is no morphological distinction between the conclusive form and adnominal form; the conclusive form marks the end of sentence, and the adnominal form indicates that a clause with a verb of its form serves as a sentential modifier.Since Classical Japanese distinguishes those two forms, the verbs in the relative clauses in (34a-b) have the adnominal form na-ru and hoyu-ru instead of their conclusive counterparts na-ri and hoyu, respectively.
[e i kaku thus mi-tutu look-as 'looking at (the scenery) i that is thus there' (Tosanikki, 10th century) b. [e i sugurete exceptionally tokimeki favor tama-u] receive-ADN pro i ari-keri be-PRF 'there was (a person) i that e i exceptionally received the favor (of the emperor)' (Genzi Monogatari, 11th century) Note that there is no overt nominal expression following the relative clauses in (35a-b) although the referent of those expressions can be identified from the context.It is assumed that one reason behind the possibility of null nominals in Classical Japanese is the existence of the adnominal form.This is because a verb in adnominal form ensures the presence of a nominal that is modified by the clause including that verb.By analogy, it might be the case that in conventional pseudoclefts in Modern Japanese, complementizer no plays a similar role to indicate the presence of a null nominal.Irrespective of the validity of this hypothesis, however, a crucial point of ( 35) is that relative head could be phonologically null.Therefore, my assumption for a null nominal in conventional pseudoclefts is not implausible.Thus, I submit the structure of XP 1 of Japanese conventional pseudoclefts in (36).
(36) [ NP [ CP . . .e . . .no] e] 3.1.4More on XP 1 .I presented (37a) as an example of Japanese conventional pseudoclefts.The structure of the sentence reflects Hoji's (1990) analysis.But having demonstrated that no is a complementizer, and there is a phonologically null noun after no, (37a) is now represented as in (37b).
(37) a. (= (2)) [ (koto) fact 'the fact that the one John bought is this apple' Before moving to Section 3.2, there are two more things I would like to discuss about XP 1 in Section 3.1; identity of the gap in XP 1 and its coindexation.First, the gap in XP 1 of Japanese conventional pseudoclefts does not seem a pro, contrary to Hoji's analysis.One piece of evidence for this claim is that it is never overtly spelled out, as (38) shows that overt pronoun sore(-o) 'it-ACC' cannot surface in the gap position.Another piece of evidence against the gap being a pro comes from the reflexive connectivity effect between XP 1 and XP 2 .It has been proposed that a reflexive in XP 2 can be coreferential with the subject in XP 1 (e.g., Higgins, 1979). 22Consider an example of the reflexive connectivity effect in English pseudoclefts below.
(39) a.The one John i blamed yesterday was himself i .
b.The one John i blamed yesterday was him i *.
Sentences (39a-b) are different from each other only in terms of XP 2 .What is interesting about ( 39) is that the XP 2 that can be coindexed with John in XP 1 is himself, and not him.This is so despite the fact that John does not c-command XP 2 superficially.
The availability of the coindexation in ( 39) is known as a connectivity effect because the XP 2 in (39) can be interpreted as if it is in the gap in XP 1 .If the XP 2 in (39) is inserted in the gap position, the TPs of the XP 1 are those in (40a-b), respectively.
(40) a. John i blamed himself i yesterday.
b. John i blamed him i * yesterday.
In (40a-b), the availability of the coindexation is reasonable because himself satisfies Condition A and him violates Condition B of the Binding Theory.Since (40a-b) have the same results as those in (39a-b) in the availability of the coindexation, pseudoclefts like (39) are known to exhibit the connectivity effect. 23n light of (39-40), consider first the Japanese counterpart of ( 39) in ( 41).(41a) involves a reflexive pronoun zibun-zisin 'self-self' (e.g., Katada, 1991), and (41b) involves a personal pronoun kare, which is proposed to abide by Condition B of the Binding theory (e.g., Noguchi, 1997).Just like (39a-b), (41a-b) differ only in whether XP 2 is a reflexive or personal pronoun.Note also that what can be coreferential with the subject of the embedded clause in XP 1 (i.e., Taroo) is zibun-zisin, and not kare.Thus, in Japanese conventional pseudoclefts too, it is a reflexive pronoun that can refer to the subject of the relative clause in XP 1 .Also, this result is consistent with the availability of the coindexation in (42), which corresponds to (40).too, what can be coreferential with Taroo is zibun-zisin.Therefore, it can be said that Japanese conventional pseudoclefts also exhibit the reflexive connectivity effect, which is a crucial property of English pseudoclefts.

M. Harada
Having demonstrated that in Japanese conventional pseudoclefts, XP 2 is coindexed with the gap in XP 1 , where the gap can be coindexed with the subject, I point out that this fact undermines Hoji's analysis of the gap being a pro.This is because if the gap is a pro, it violates Condition B of the Binding Theory, and sentences like (41a) must be ungrammatical, contrary to the fact.Thus, the availability of the use of a reflexive in XP 2 that refers to the subject in XP 1 casts doubt on the claim that the gap in XP 1 is a pro.Therefore, in this paper, I just indicate the gap in XP 1 with e as an empty category. 24 Secondly, although no is coindexed with XP 2 and the gap in XP 1 in Hoji's (1990) analysis (37a), it does not seem to be the case that null noun is coindexed with those two elements as in (37b).Consider first (43) to confirm that an R-expression can be used in XP 2 .This is not a welcome result for a hypothesis that the null noun is coindexed with XP 2 .This is because the grammaticality of ( 43) is not affected by the violation of Condition C of the Binding Theory even though the null noun c-commands XP 2 .On the other hand, if the null noun is not coindexed with XP 2 , the grammaticality of ( 43) is not surprising at all.Given this, it is worth noting again that XP 2 is coindexed with the gap in XP 1 .What this means is that the gap in XP 1 is coindexed with XP 2 , which cannot be assumed to be coindexed with the null noun.Therefore, it is plausible to assume that the null noun is not coindexed with the gap in XP 1 as well as with XP 2 .In fact, this is an expected hypothesis if conventional pseudoclefts are indeed th-clefts.This is because Collins (1991) also proposes that the head of the relative construction in XP 1 of English th-clefts is not coindexed with either the gap or XP 2 .This can be confirmed by his schematic structure of English th-clefts (44).
(44) (Collins, 1991, p. 27): The Although he assumes that the gap in XP 1 and XP 2 are coindexed, there is no element coindexed with the head, just as I mentioned about Japanese data.
To sum up, Section 3.1 demonstrated that XP 1 in Japanese conventional pseudoclefts is a head-external relative.In fact, it turned out it differs from XP 1 of English th-clefts only in that the relative marker is always a complementizer and the head is always phonologically null.In this way, Japanese conventional pseudoclefts have the defining properties of pseudoclefts in (8a).

Presence of non-clefted counterparts of Japanese conventional pseudoclefts.
In this section, I demonstrate that Japanese conventional pseudoclefts have the defining properties of pseudoclefts in (8b) as well; that is, they have their non-clefted counterparts.First, consider the conventional pseudoclefts in (45) again.45) has its non-clefted counterpart. 24One of the strong arguments for the claim of the gap being a pro is the lack of subjacency effect in Japanese relative clause.But it has been proposed that Japanese head-external relatives are indeed subject to the subjacency effect in some constructions (e.g., Inoue, 1976;Hasegawa, 1981;Ishizuka, 2009).Thus, there is a claim that Japanese head-external relatives also involve A movement.However, those constructions that are proposed to trigger the subjacency violation do not actually trigger the violation if vocabularies and contexts are carefully designed.For this reason, although I do not deny the possibility that Japanese relatives also involve A movement, it seems premature to accept that possibility now.Therefore, I just assume that there is an empty category in XP 1 .Sentence (46a) consists of XP 2 and materials in the relative clause that follow the relative marker in ( 45) and (46b).In other words, ( 45) and (46b) look as if they are derived by 'cleaving' a non-clefted sentence in (46a) into XP 1 and XP 2 .Thus, sentences like (45) also has a property of clefting.Therefore, given that Japanese conventional pseudoclefts satisfy (8a-b), it is reasonable to claim that they are indeed pseudoclefts.Now that Japanese conventional pseudoclefts are pseudoclefts, I provide its schematic structure below in (47).
(47) Schematic structure of Japanese pseudoclefts [ NP [ CP . . .e i . . .no] e]-case XP 2i de ar- As shown in (47), Japanese pseudoclefts consist of three major components; XP 1 , which is the first boxed phrase in (47), XP 2 , and de ar-, which, roughly speaking, functions like English be.XP 1 is a head-external relative, whose head is phonologically null, unlike English th-clefts.However, I demonstrated that it is reasonable to assume its existence, and that no is a complementizer, which could appear in English th-clefts as well.Therefore, I conclude sentences with the structure in (47), which have the two defining properties of pseudoclefts in (8), are pseudoclefts corresponding to English th-clefts.

Summary.
In this section, I justified the presence of pseudoclefts in Japanese in light of the definition of pseudoclefts in (8a-b).More specifically, I argued that Japanese conventional pseudoclefts are th-clefts.
This section largely divided into two parts.Section 3.1 first demonstrated that XP 1 of Japanese pseudoclefts is a head-external relative clause, and thus Japanese pseudoclefts satisfy (8a).I argued that no, which can be realized as a genitive case, pronoun or complementizer in Japanese, is a complementizer in Japanese pseudoclefts.Given this conclusion, it is counterintuitive that XP 1 is an NP or head-external relative because there is no overt noun head in XP 1 .However, I showed that the XP 1 is nominal in nature, and provided data in English, Modern Japanese and Classical Japanese to support for the presence of a phonologically null noun that projects a relative construction.Thus, it is not implausible to assume that XP 1 of Japanese pseudoclefts is a head-external relative.Section 3.2 constructed an argument that Japanese pseudoclefts satisfy the other defining property of pseudoclefts as well.That is, they have their non-clefted counterparts.Therefore, the satisfaction of (8ab) indicates that Japanese conventional pseudoclefts are indeed pseudoclefts that correspond to English pseudoclefts.

Conclusion
This paper examined whether the constructions that are often referred to as pseudoclefts in Japanese (which I called conventional pseudoclefts) are indeed pseudoclefts, and if so, in what sense they are pseudoclefts.With this purpose of the paper, Section 2 first defined pseudoclefts.This was important especially because much of the literature on pseudoclefts does not provide a clear definition of the construction.After it was demonstrated that pseudoclefts consist of two defining properties, Section 3 then discussed how Japanese conventional pseudoclefts satisfy those defining properties.Concretely, Section 3.1 clarified an ambiguous syntactic category of no in Japanese conventional pseudoclefts as a complementizer, and provided several pieces of evidence for my claim regarding why XP 1 should be considered a nominal expression or headexternal relative.This section was related to one the two defining properties of pseudoclefts.Then, Section 3.2 demonstrated that Japanese conventional pseudoclefts also have their non-clefted counterparts, which is the other defining property of pseudoclefts.In this way, the paper concluded that Japanese conventional pseudoclefts, which satisfy two defining properties of pseudoclefts, are indeed pseudoclefts.
The conclusion of this paper now enables us to delve into various properties of Japanese pseudoclefts in reference to pseudoclefts in other languages.Also, similarities and differences between Japanese and other languages' pseudoclefts in future research will yield insights into the research of pseudoclefts and other associated fields of study.
[relative clause].The attributive relation between head and [relative clause] is such that the head is involved in what is stated in the clause."With (5) in mind, consider two examples of English relative constructions in (6).(6) a. John buys the thing [that Herman bought [e]].b.John buys the thing [which Herman bought [e]].
) It was those students i that he came across.' he came across were those students.' those students i that he came across.'