The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier* Yoshiki Ogawa

Transcription

1 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier* Yoshiki Ogawa Abstract In Japanese root clauses, most verbs must be paired with the morpheme -tei to form a complex predicate in order to denote a present state. However, in certain subordinate clauses including the relative clause, the presuppositional clause of a cleft sentence, and the comparative clause, this restriction is exempted. Semantic restrictions are also imposed on verbs, and the verbs which permit the simple present tense in the relevant syntactic environment are limited to those which denote a potentially everlasting state or a potentially everlasting and repeated motion. In this article, I will explain both of these syntactic and semantic restrictions by making a single proposal that the UG accommodates a phonetically empty universal quantifier that quantifies over time, called U-OP. Along with this proposal, I assume that the U-OP is subject to a formal licensing condition which is quite similar to other phonetically empty operators and that the verbs in the simple present tense in Japanese involve an empty position that corresponds to the -tei morpheme and that can function as a variable for the U-OP. We will also argue that the proposed U-OP is similar but different from Chierchia s (1995) generic operator. Keywords: simple present tense, aspect, universal quantification over time, null operators, genericity, stage-level/individual-level distinction

2 Yoshiki Ogawa 1. Introduction It is hard to identify the semantics of a tense morpheme attached to a verb, since a single form often refers to more than one meaning: a verb paired with the past tense morpheme denotes not only an event or state that held in the past but also a counterfactual context (but see Iatridou 2000), and a verb paired with the present tense morpheme does not always denote an event or state that holds at the point of speech time. Thus, in English, the simple present tense attached to a stative verb denotes a state that holds at the point of speech time, whereas the one attached to a non-stative verb can only denote a generic or habitual property attributed to the referent of the subject: (1) a. John knows Mary. b. John walks in the garden. c. Dogs bark. It is also known that English and Japanese differ with respect to whether the simple present tense on a stative verb can denote a state that holds in the real world at the point of speech time (henceforth, referred to as a present state ). Thus, the Japanese sentence in (2a) can receive a modal interpretation that a certain state may be realized in the future, whereas it cannot denote a present state. In order to denote a present state, the morpheme -tei (more accurately, the complex of two morphemes -te and -i), which is used to denote a progressive or perfective aspect elsewhere, must be added, as in (2b): (2) a. John-ga Mary-o sir-u (koto) John-Nom Mary-Acc know-nonpast (fact) John will come to know Mary./*John knows Mary.

3 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier b. John-ga Mary-o sit-tei-ru (koto) John-Nom Mary-Acc know-tei-nonpast (fact) John knows Mary. Some literature has pointed out that, in Japanese, some verbs in the simple present tense can denote the present state if they occur in a relative clause (cf. Takahashi 1973, Kitahara, et. al. 1981, Kinsui 1994). 1 However, to the best of my knowledge, no one has observed that sir-u, which is glossed as know-nonpast, can denote the present state of knowing if it occurs in a relative clause, as in (3b). Much less, no generalization on this construction-specific phenomenon including sir-u has been reached yet: (3) a.*suu-nin-no hito-ga sono ziken-no sinsou-o sir-u few-cl-gen man-nom the case-gen truth-acc know-nonpast A few people know the truth about the case. b. Sono ziken-no sinsou-o sir-u hito-to hanasi-o si-you. the case-gen truth-acc know-nonpast man-with talk-acc do-shall Let's talk with the man who knows the truth about the case. Suppose, at odd with the traditional classification in the Japanese linguistics (cf. note 3), that the verb siru know belongs to the stative verbs in the sense of Vendler (1967), since many stative verbs manifest the same behavior as siru in (3). However, a comprehensive examination of the Japanese verbs will demonstrate that not all stative verbs show the same pattern as siru and that some action verbs also behave as in (3). Here arise at least three questions. First, what is the common semantic property of the set of verbs that show the same contrast as in (3)? Second, why is the simple present tense allowed in the relative clause but not in a simple clause in

4 Yoshiki Ogawa Japanese? Third, why is this restriction imposed in Japanese, but not in English? To the best of my knowledge, no one has challenged and provided a solution to all of these questions in the field of the generative grammar. In order to answer these questions, I propose that universal grammar accommodates a phonetically empty universal quantifier (U-OP) that can be adjoined to a clause and that it is syntactically licensed by satisfying both of the following two properties: (i) binding an empty position in the verbal complex consisting exclusively of the verb stem and the present tense morpheme, and (ii) being c-commanded by another semantically contentful operator. We will argue that these two properties of the U-OP will both be motivated on the grounds independent of their temporal properties. This article is organized as follows: in section 2, we will propose a descriptive generalization about the verbs in the simple present tense that can denote a present state in a relative clause. In section 3, we will introduce a set of theoretical devices and account for the generalization. Specifically, it is proposed that the Japanese verbs in the simple present tense must introduce a temporal variable that can be bound by the U-OP. In section 4, we will illustrate four more syntactic environments in which the simple present tense can denote a present state, and argue that the facts immediately follow from the proposed mechanism. In section 5, we will provide a morphological motivation for the U-OP in Japanese. In section 6, we will show that the presence of the U-OP in Japanese receives another, and more robust, syntactic motivation. In section 7, we will discuss two remaining issues that lie in the proposed framework. In section 8, we will discuss why, in English, the simple present tense on a stative verb can denote a present state, even in a simple clause. In section 9, we will discuss three alternative analyses of the proposed generalization that do not rely on the U-OP. And section 10 is a conclusion of this article.

5 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier 2. The Descriptive Generalization In this section, we will identify what kind of verbs can be paired with the simple present tense to denote a present state in a relative clause The Simple Present Tense in a Simple Clause in Japanese Vendler (1967) divides verbs in English into the following four different classes in terms of their aspectual properties: stative verbs, action verbs, accomplishment verbs, and achievement verbs. They are exemplified as below: (4) a. stative verbs: know, have, believe, love, etc. b. action verbs: run, walk, eat, talk, etc. c. accomplishment verbs: draw, break, open, create, etc. d. achievement verbs: win, reach, get, attach, etc. We have seen that stative verbs in Japanese cannot be paired with the simple present tense to denote a present state. The following sentences show that not only stative verbs but also action verbs, accomplishment verbs, and achievement verbs can only receive a modal interpretation if they are paired with the simple present tense: (5) a. John-ga kouen-o aruk-u (koto) John-Nom park-acc walk-nonpast fact John will/can walk in the park. b. John-ga e-o kak-u (koto) John-Nom picture-acc draw-nonpast fact John will/can draw a picture.

6 Yoshiki Ogawa c. John-ga siai-ni kat-u (koto) John-Nom game-at win-nonpast fact John will/can win the game. Therefore, we can conclude that the following descriptive generalization about the morphology-semantics correspondence exists in Japanese, irrespective of whether the verb is stative verb, action verb, accomplishment verb, or achievement verb: (6) If a verb is directly followed by the present tense morpheme -ru, it can receive a modal (habitual or future) interpretation but not a present state interpretation The Simple Present Tense in a Relative Clause in Japanese In a simple clause in Japanese, (6) seems to hold almost without exception. 2 However, when the four classes of verbs paired with the simple present tense are placed in a relative clause, they do not behave uniformly. First, compare (7a-d): (7) a. Sono ziken-no sinsou-o sir-u hito-to hanasi-o si-you. 3 the case-gen truth-acc know-nonpast man-with talk-acc do-shall Let's talk with the man who knows the truth about the case. b. Kouen-de utaw-u hito-to hanasi-o si-you. park-loc sing-nonpast man-with talk-acc do-shall Let's talk with the man who is singing in the park. c. Kouen-de e-o kak-u hito-to hanasi-o si-you. Park-Loc picture-acc draw-nonpast man-with talk-acc do-shall Let's talk with the man who is drawing a picture in the park.

7 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier d. Kyou-no siai ni kat-u hito-to hanasi-o si-you. today-gen game-dat win-nonpast man-with talk-acc do-shall Let's talk with the man who is winning today's game. The verb siru know in (7a) is a stative verb (although this assumption is controversial; see note 3), the verb utawu sing is an action verb, the verb kaku draw is an accomplishment verb, and the verb katu win is an achievement verb. And only (7a) is acceptable in the present state interpretation. (If we replace the simple present tense in (7a-d) by the progressive -teiru form, we always obtain an acceptable sentence.) The contrast observed between (7a) and (7b-d) is also observed between verbs like sinziru believe, hokoru boast, wakaru understand, aru be/have, kankeisuru relate, kakawaru relate, tigawu differ, zokusu belong on the one hand, and verbs like naguru hit, sawaru touch, migaku polish, noru ride (so far, verbs of contact, a subclass of action verbs), yaku burn, yomu read, kawu buy, hakobu bring (so far, accomplishment verbs), hakkensuru discover, kizuku notice, mitukeru find, arawareru appear, kieru disappear, toutyakusuru arrive (so far, achievement verbs), on the other hand. These facts appear to indicate that, in a relative clause, only stative verbs can be paired with the simple present tense to denote a present state. However, a more comprehensive examination of Japanese verbs demonstrate that this is not an accurate generalization. Levin and Rappaport (1994) refer to verbs such as hang, sit, and stand, which specify the position of an entity that bears a particular spatial configuration with respect to that position, as verbs of spatial configuration. They distinguish three different types of noncausative meanings and one type of causative meaning which a particular verb of spatial configuration in English has, and refer to each of the three types of noncausative meanings as the maintain position sense (agentive), the assume position sense (agentive), the simple position sense (nonagentive), respectively (ibid.:127). The

8 Yoshiki Ogawa three different types of noncausative meaning can be illustrated, with the verb stand, as follows: 4 (8) a. Yvenne stood alone (in the hallway). (the maintain position sense) b. Yvenne stood up/down. (the assume position sense) c. The statue stood *(on the pedestal). (the simple position sense) Given Vendler's classification, stood in (8a) and (8c) is a stative verb, whereas stood in (8b) is an accomplishment verb. The Japanese translations of these three different meanings of the verb stand are not morphologically identical. Thus, we have (9a-c) corresponding to (8a-c): (9) a. John-ga genkan-ni hitoride tat-tei-ta (koto) John-Nom hallway-at alone stand-tei-past fact b. John-ga tat-ta (koto) John-Nom stand-past fact c. Douzou-ga daiza-noueni tat-tei-ta (koto) statue-nom pedestal-on stand-tei-past fact Among these three sentences, (9a) and (9c), but not (9b), accompany the morpheme -tei. This fact shows that in Japanese, as well as in English, a single verb of spatial configuration can be used as either stative verb or accomplishment verb, and that tatu is a stative verb only in (9a) and (9c) (pace Kindaichi (1950), who would analyze tatu in (9a,c) as momentary verbs or the forth kind of verbs rather than stative verbs). Now, if any stative verb in the Japanese relative clause could be paired with the simple present tense to denote a present state, the verb tatu should be paired with the simple present tense to denote a present state, whether it has the maintain

9 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier position sense or in the simple position sense. However, the fact is that the verb tatu in the simple position sense, but not the one in the maintain position sense, can be paired with the simple present tense to denote a present state. Thus, we observe the following contrast: 5 (10) a.*genkan-ni hitoride tat-u hito-to hanasi-o si-you. hallway-at alone stand-nonpast man-with talk-acc do-shall Let's talk with the man who is standing alone in the hallway. b. Oka-noueni tat-u yamagoya-ni ik-ou. hill-on stand-nonpast cottage-to go-shall Let's go to the cottage that stands on the hill. Note that the verb tatu is a stative verb in both (10a) and (10b). Nevertheless, (10b) but not (10a) is acceptable in the present state interpretation. The only difference between (10a) and (10b) is that the state of someone standing in the hallway is a temporal state that can change in the next moment, whereas the state of a cottage standing on the hill continues permanently, unless some accident such as an earthquake destroys it. In other words, Kratzer (1989) will analyze tatu in (8a) as a stage-level stative predicate and tatu in (8b) as an individual-level stative predicate. This is shown by the following contrast: (11) a. John is standing/*stands in the hallway. b. A cottage stands/*is standing on the hill. Most verbs of spatial configuration are ambiguous between the stage-level and individual-level stative predicates. However, when we examine relative clauses in which verbs of spatial configuration are in the simple present tense, they are acceptable only when they receive an individual-level stative interpretation. Some

10 Yoshiki Ogawa examples are given below: (12) a. Mati-no tyuusinbu-o turanuk-u kansen douro-o hasir-ou. town-gen center-acc run-nonpast main road-acc drive-shall Let s drive the main road that runs through the center of the town. b. Tohoku-tihou-ni yokotawar-u Ouu Sanmyaku-o koe-you. Tohoku-district-Loc lie-nonpast Ouu mountains-acc cross-shall Let s cross over the Ouu mountains that lie in the Tohoku District. c. Fuji sanroku-ni hirogar-u Mikatagahada daiti-o Fiji fothill-loc spread-nonpast Mikatagahara plateau-acc aruk-ou. walk-shall Let s walk around the Mikatagahara plateau that spreads on the foothills of Mt. Fuji. d. Ganzen-ni sobie-ru Unzen Fugendake-ni nobor-ou. before-eye-loc soar-nonpast Unzen Fugendake-Loc climb-shall Let s climb the Unzen Fugendake that soars before our eyes. e. Taiheiyou-o ganka-ni nozom-u kono Pacific Ocean-Acc below-eye-loc command-a-view-nonpast this yasiki-o kaw-ou. manshion-acc buy-shall Let s buy this manshion that commands a view of the Pacific Ocean below. f. Umai hanashi-no kage-ni hisom-u otosiana-ni kiotuke-you. juicy story-gen shadow-loc lurk-nonpast pit-to care-shall Let be careful for a pit that lurks behind a juicy story.

11 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier g. Daidokoro to ima-o sikir-u husuma-o kitchen and living room-acc partition-nonpast fusuma-acc torinozok-ou. get rid of-shall Let s get rid of the fusuma that partitions the kitchen and the living room. h. Ano oka-noueni nemur-u Taro-no itai-o hori-das-ou. that hill-on sleep-nonpast Taroo-Gen body-acc dig-up-shall Let s dig up Taro s body that sleeps under the ground of that hill. i. Kono mizuumi-ni sum-u kyodai seibutu-o tukamae-you. this lake-loc live-nonpast huge creature-acc catch-shall Let s catch the huge creature that lives in this lake. j. Sin'yaku kaihatu-no kagi-o nigir-u kono new medicine develop-gen key-acc hold-nonpast this kagakubussitu-o hukaku kenkyuu-siyou. chemical-acc deeply study-shall Let s make a deep study of this chemical that holds a key to the development of a new medicine. k. Tokyo-no 60 kiro kita-ni iti-suru Omiya-o Tokyo-Gen 60 kilometer north-loc locate-nonpast Omiya-Acc mezas-ou. head-for-shall Let s head for the Omiya that locates 60 kilometers to the north of Tokyo. l. Tanpakusitu-o ooku hukum-u kono syokuhin-o tabe-you. protein-acc much contain-nonpast this food-acc eat-shall Let s eat this food that contains much protein.

12 Yoshiki Ogawa m. Kigi-ga oisiger-u miturin-nonaka-o susum-ou. 6 tree-nom thrive-nonpast jungle-inside-acc step forward-shall Let s step forward the jungle in which trees thrive. n. Fukusuu-no ryouiki-ni matagar-u mondai-ni trikum-ou. plural-gen field-dat spread-nonpast issue-dat challenge-shall Let s challenge an issue that spreads over more than one field. o. Eiyou-ni tom-u kono syokuhin-o tabe-you. nutrition-to abound-nonpast this food-acc eat-shall Let s eat this food that abounds nutritions. p. 9-hon-no asi-o mot-u kono tako-o 9-Cl-Gen tentacle-acc have-nonpast this octopus-acc nigasou. release-shall Let s release this octopus that have nine tentacles. Even when the same verbs of spatial configuration as in (12) are used, a present state interpretation is incompatible with a stage-level interpretation of the predicates: 7 (13) a.*hanzaisya-ga hisom-u akiya-ni sen'nyuusi-you. Criminal-Nom lurk-nonpast empty house-to sneak-into-shall Let s sneak into the house where a criminal is lurking. b.*enkai-o sikir-u Taro-o turedas-ou. feast-acc manage-nonpast Taro-Acc take out-shall Let s take out Taro, who is managing the feast. c.*kouen-no benti-de nemur-u Taro-o okos-ou. Park-Gen bench-loc sleep-nonpast Taro-Acc wake-shall Let s wake up Taro who is sleeping on the bench of the park.

13 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier d.*asoko-de uma-ni matagar-u hito-to hanasi-o si-you. over there-loc horse-on ride-nonpast man-with talk-acc do-shall Let s talk with the man who is riding a horse over there. e.*hanako-no te-o nigir-u otoko-no kao-o mi-you. Hanako-Gen hand-acc hold-nonpast man-gen face-acc look-shall Let s look at the face of the man holding the hands of Hanako. f.*taro-no kaban-o mot-u syounen-to hanasi-o si-you. Taro-Gen bag-acc hold-nonpast boy-with talk-acc do-shall Let s talk with the boy who is holding Taroo s bag. On the basis of these observations, we could reach the following generalization: (14) In a relative clause, only individual-level stative verbs can be directly followed by the present tense morpheme to denote a present state. However, there is another class of verbs that can occur with the simple present tense in a relative clause to denote a present state. They are clearly not stative verbs. This class of verbs includes nagareru flow/run, mawaru revolve, kizamu chop, kurasu lead a life, matu wait, yureru shake, tenmetusuru blink, mawu whirl, awadatu bubble, and uzumaku swirl. Some examples containing these verbs are given below: (15) a. Si-nai-o nagare-ru Hikawa-ni damu-o tukur-ou. town-in-acc flow-nonpast Hikawa-Loc dam-acc construct-shall Let s make a dam in the Hikawa that runs through this town.

14 Yoshiki Ogawa b. Tikyuu-no mawari-o mawar-u tuki-ni rokettoi-o toba-sou. earth-gen round-acc revolve-nonpast moon-to rocket-acc send-shall Let s send a rocket to the moon that revolves round the Earth. c. Tetudou-no husetu-o mat-u zimoto zyuumin-ni railroad-gen building-acc wait-nonpast local resident-to hanasi-o kik-ou opinion-acc listen to-shall Let's listen to the opinion of the local residents who are waiting for the (planned) building of the railroad. d. 100 nen-kan seikakuni toki-o kizam-u kono 100 yeay-for accurately time-acc chop-nonpast this furikodokei-ga boku-no takaramono da. pendulum clock-nom my-gen treasure is This pendulum clock that has kept good time for 100 years is my treasure. e. Kono mati-ni kuras-u hitobito-ni yume-o this town Loc lead-a-life-nonpast people-dat dream-acc atae-ta-i. give-want-nonpast I want to give a dream to the people who lead a life in this town. f. Soyokaze-ni yure-ru huurin-ga breeze-for shake-nonpast Japanese wind bell-nom dasi-tei-u oto-wa suzusisa-o motarasu-darou. emit-nonpast sound-nom coolness-acc bring-will The sound made by the Japanese wind bell that is shaking for the breeze will bring us coolness.

15 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier g. Oki-de tenmetu-suru toudai-no hikari-no-okage-de offing-loc blink-nonpast lighthouse-gen light-gen-thanks-to minato-no houkou-ga wakaru-darou. harbor-gen direction-nom see-will We will see the direction to the harbor thanks to the light of the lighthouse that blinks in the offing. h. Hankagai-noue-o maw-u karasu-o nantoka Amusement quarter-acc whirl-nonpast crow-acc somehow oihara-ou. disperse-shall Let s disperse the crows that whirl over the amusement quarter. Although these Japanese verbs are clearly instances of action verbs, they differ from verbs like naguru hit, yomu read, and migaku polish in that they can all denote some constant motion of the referent of their subject that can last forever in an externally determined local area, unless some external force blocks it. Thus, a river s running through a town involves the constant motion of its water which lasts forever along a geographically determined course on the land, and the moon's rotation round the earth is a constant motion along a physically determined orbit and which lasts forever. The local resident's waiting for the construction of the railroad may not last forever, because the railroad may open in near future or because the plan itself may be canceled; but we can easily expect that the railroad does not open forever, without the plan being canceled, due to the lack of the budget or the governmental policy. Anyway, the action of waiting is constant and has no inherent endpoint. A pendulum clock's keeping good time is shown by the constant swing of the pendulum, which can be repeated forever unless the clock breaks down. As such, the motions indicated by these verbs share some property of constancy and potentially eternal continuation. If we following Smith (1983) in

16 Yoshiki Ogawa assuming that states are stable situations that do not simply change, then we could say that the relevant running, rotating, and waiting motions are a kind of states (cf. van Voorst 1988:22). Some of the verbs in (15) can also be used to denote an event which is not everlasting. In these cases, however, in contrast to (15a-c), the verbs nagareru, mawaru, and kizamu cannot be paired with the simple present tense to denote a present state: (16) a.*kareha-ga suu-mai nagare-ru ogawa-no hotori-ni dry leaf-nom some-cl flow-nonpast creak-gen beside-at tat-ou. stand-shall Let s stand beside the creak on which some leaves are flowing down. b.*tokuisaki-o mawar-u Taro-ni denwa-o customer-acc go the rounds of-nonpast Taro-Dat telephone-acc si-you. do-shall Let s call Taro who is going the rounds of his customers. c.*daidokoro-de houtyou-de yasai-o kizam-u kitchen-loc kitchen knife-with vegetable-acc chop-nonpast hito-ni hanasikake-te-wa-ikenai. man-to speak-comp-top-must-not You must not speak to a person who is chopping vegetables with a kitchen knife in the kitchen. Although the verbs exemplified in (15) and those in (16) are both plausibly classified into action verbs, (15) and (16) are different in one crucial respect: the

17 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier events denoted by the verbs in (15) can be everlasting and repeatable unless some external (natural) force blocks them, the events denoted by the verbs in (16) have a pragmatically determined endpoint which is given by the referent of the object and which the referent of the subject arrives at even if no external (natural) force blocks the event(s). Thus, a dry leaf s flowing on a river will end if the river reaches the sea; Taro's going the rounds of his customers will end if he finish visiting all of his customers; someone s chopping a vegetable will end if the vegetable is divided into sufficiently small pieces. We have seen that the simple present tense can also be paired with a verb to denote a present state even if the verb is action verb, if the motion denoted by the verb is everlasting unless some external force blocks it, that is, potentially everlasting. Combining these observations with (14), we propose the final version of our generalization as follows: (17) In a relative clause, a verb in the simple present tense can denote a present state if and only if the verbal complex denotes either a state or a constant and repeatable motion of the referent of its subject that is potentially everlasting. 8 Before closing this section, let us point out two important clarifications about the description in (17) that turn out crucial in the following discussion. The first is concerned with the possible interpretations of sentences such as (18): (18) a.(*)ren ai syousetu-o kak-u syousetuka-wa love story-acc write-nonpast novelist-top ikiikisi-tei-ru. lively-be-nonpast The novelist who is writing love stories is lively.

18 Yoshiki Ogawa b.(*)hune-nouede katuo-o tur-u ryousi-wa ship-on bonito-acc fish-nonpast fisher-top ikiikisi-tei-ru. lively-be-nonpast The fisher who is fishing bonitos on a ship is lively. In (18a,b), just as in (7c), a present state interpretation is unavailable (or, at least, unforced). We can relate this fact to the fact that the verbs kaku write and turu fish are accomplishment verbs, whose motions have a lexically determined endpoint. However, they are acceptable in a generic or modal (future or habitual) interpretation. This fact does not conflict with our generalization, since neither the generic nor the modal interpretation requires that the state or motion is going on at the point of speech time. In fact, in the generic and modal interpretations, any type of verb can be paired with the simple present tense, and this holds not only in a relative clause but also in a simple clause. Hence, in the following discussion, readers should be careful not to confuse the present state interpretation on the one hand and the generic or modal interpretation on the other hand (see section 7.2 for a relevant discussion). Second, we stated (17) as a generalization about Japanese. However, the English translations of some of the verbs in (7a), (10b), and (15) can be used with the simple present tense to denote a present state. Consider (19) and (20): (19) a. The man (that) knows the rumor b. The statue (that) stands on the pedestal (20) a. Nomadism (that) runs in his blood b. The earth (that) revolves round the sun Given these facts, one might wish to restate (17) as a cross-linguistic

19 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier generalization and seek to find an independent reason for why the same types of verbs in English can be paired with the simple present tense to denote a present state in a non-relative clause as well as in a relative clause. However, this is not a plausible way toward a solution of the problem at hand, for two reasons. First, the English translations of the other verbs in (15) cannot be paired with the simple present tense to denote a present state: (21) a. The crow (that) is whirling/*whirls in the sky b. The breeze (that) is shaking/*shakes the flags c. The green lights (that) are blinking/*blink on the surface of the harbor Second, in contrast to the first point, the examples in (19) snd (20) do not have a counterpart with the progressive morpheme. These two facts will suffice to shows that we should provide different treatments for the minimal pair in Japanese that involves a sentence with the simple present tense and the one with the V-tei-ru form on the one hand and the minimal pair in English which involves a sentence with the simple present tense and the one with the progressive form on the other, even if the two are largely overlapping phenomena. For this reason, for the moment, we will limit our focus to Japanese, and return to some questions about English and/or universal grammar in sections 8 and 10. Readers should also refer to Giorgi and Pianesi s (1997:chapter 4) work on the simple present tense, Giorgi and Pianesi answer the question why the non-stative verbs in English cannot denote a present state as in (22): (22) a. In English, (unlike in Italian,) non-stative verbs are always marked as perfective, in order to distinguish them from nouns in the absence of overt conjugations.

20 Yoshiki Ogawa b. Perfective predicates denote closed events. c. Closed events cannot be simultaneous with punctual events. d. Speech event is punctual. e. Hence, the non-stative verbs in English cannot denote a present state According to this logic, they have to explain why stative verbs do not have to be marked as perfective. For this question, they assume, following Chierchia (1995), that statives and habituals have a quantificational feature associated with the generic operator, and argue that the presence of the quantificational feature exempts them from holding a perfective feature, and as a result, they do not denote a closed event. Although I do not stand on a particular position for or against the assumption in (22a,b), I will argue against Chierchia's (1995) proposals about statives in section 7.2. If Chierchia's argument collapses, Giorgi and Pianesi's (1997) analysis depending on it will also have to be modified. But I will not discuss the intricate issue in this article, leaving it for future research. Now that we have identified our main concern in this article as the simple present tense in Japanese, the questions to be solved are the following: (i) why is the semantic restriction imposed on verbs?, and (ii) why is the syntactic restriction imposed on their occurrence? We will introduce a set of theoretical devices that enable us to answer these questions. 3. Proposals 3.1. Theoretical Constructs One important aspect of the generalization in (17) is that the present tense morpheme does not always receive a present state (or everlasting) interpretation and that it is more often associated with a generic or modal interpretation. This is an instance of the lack of one-to-one correspondence between the (morpho-)phonetic form and the meaning of an expression that is one salient

21 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier property of human language. Throughout the history of generative grammar, the mismatch between form and meaning has been one of the strongest motivations for postulating a phonetically invisible theoretical construct. For example, consider the fact that, in (23), a single phonetic form is combined with two interpretations as shown in (23a,b): (23) Max saw the lady with the binocular. a. Max saw the lady by means of the binocular. b. Max saw the lady who possessed the binocular. This fact has been explained by associating (23) with two different phrase structures which are phonetically invisible and whose informal notations are as follows: (24) a. Max [ VP saw [ DP the lady] [ PP with the binocular]]. b. Max [ VP saw [ DP the [ NP lady] [ PP with the binocular]]]. Along a similar vein, when we want to resolve the lack of one-to-one correspondence between the present tense morphology and its interpretations, it may be reasonable to postulate some phonetically empty theoretical construct(s), as far as it enables us to explain the relevant set of data that is otherwise unexplained. Another important aspect described in (17) is that, if the present tense morphology associated with a particular class of verbs in a relative clause has a present state interpretation, then it always has another interpretation, which I will call the everlasting interpretation. One may consider these two interpretations associated with a single phonetic form to be a case of lexical ambiguity. However,

22 Yoshiki Ogawa another possibility is that the present state interpretation is a special case of the everlasting interpretation. If we pursue this possibility, we can assume that the present state interpretation does not directly follow from the existence of the present tense morpheme but is actually derived from the holding of the everlasting interpretation, by some entailment relation. A similar construal can be obtained when a universal quantifier interacts with an existential quantifier in a simple clause, as in the following example: (25) Every girl called some boy. a. Each member of the set of girls called one different boy. b. There was a single boy (not necessarily specific) who was called by all the members of the set of girls. In a standard theory of logical form, the two interpretations in (25a,b) are derived from two different hierarchical arrangements of the logical operators, as shown below: (26) a. [ x, x a girl [ y, y a boy [x called y]]] b. [ y, y a boy [ x, x a girl [x called y]]] An alternative construal is that (25) is not ambiguous but the interpretation in (25b) is a special case of the one in (25a) (cf. Reinhart 1979). According to this construal, we can assume that the only logical form mapped from (25) is (26a) (but see Chierchia and McConnel-Ginet (1990: ) for an argument against this view). Along the same line of reasoning, we can assume that there is only one logical form for the two different interpretations of sentences like (3b), (7a), and (12a-p), and that the present state interpretation is a special case of the everlasting interpretation.

23 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier A third important aspect of the generalization is that the simple present tense in the Japanese relative clauses can have a present state interpretation for two apparently unrelated types of verbs: the first type is a individual-level stative verb, and the second is an action verb denoting a constant motion that can be everlasting. However, if we restate the two types of sentences by means of a logical representation, there loom up a few common properties between them. Compare (27a,b) and (28a,b): (27) a. The statue stands on the pedestal. b. For every t, t = time [the statue stand on the pedestal at t] (28) a. The moon revolves round the earth. b. For every t, t = time [the moon revolve round the earth at t] (27b) and (28b) share two significant properties, as summarized below: (29) a. In the nuclear scope, there is a logical variable which denotes a a time, denotes a a time, which is bound by a universal quantifier. b. The predicate in the nuclear scope takes two thematic arguments, one of which is theme and the other of which is location. Moreover, among Vendler's (1967) classification of the four different types of verbs, only stative verbs (including stand in (27a)) and action verbs (including revolve in (28a)) can be modified by duration adverbs such as for one hour. See Kaneko and Endo (2001:106) and Kusumoto (2001b:5) for a relevant discussion. Given the two properties shared by the individual-level stative verbs and the action verbs that denote a constant and everlasting motion, it will be reasonable to identify the phonetically empty theoretical construct for understanding the form-meaning mismatch in question as the phonetically empty universal quantifier

24 Yoshiki Ogawa that quantifies over time. Therefore, let us propose that, in Japanese, the empty universal quantifier can be adjoined to a clause whenever its verb receives the everlasting interpretation. If such an operator is adjoined to the clause, at least two formal requirements must be satisfied: first, it must bind a variable in order to avoid vacuous quantification; second, since it is a kind of null operator, it must conform to some extra requirement that is irrelevant to phonetically realized operators. (Note that both of these are not semantic requirements but formal or syntactic requirements.) As for the latter requirement, it is comparable to the requirement that, in the parasitic gap construction, the null operator that occurs at the left edge of an adjunct clause be c-commanded by the phonetically realized and semantically contentful Wh-operator. Combining these conjectures, let us propose (30a-f), as the first approximation: (30) a. A phonetically empty universal quantifier that quantifies over time (henceforth, the U-OP) can be adjoined to any clause whose verb is in the simple present tense. b. In Japanese, all the verbs in the simple present tense must introduce a phonetically empty temporal variable. c. For the sake of formal licensing, the U-OP must bind a temporal variable. d. For the sake of formal licensing, the phonetically empty temporal variable must be c-commanded by some operator that quantifies over time (e.g. the U-OP, a generic operator, a habitual operator, or a future modal operator). e. If the U-OP binds the temporal variable, the everlasting interpretation of the verb is obtained.

25 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier f. The present state interpretation is derived as a special case of the everlasting interpretation The Asymmetry between Simple Clauses and Relative Clauses Now, let us see how the proposals in (30) interact to derive a well-formed sentence with the everlasting interpretation and the present state interpretation. An example that can receive the relevant interpretations is (10b), repeated below as (31a): (31) a. Oka-noueni tat-u yamagoya-ni ik-ou. hill-on stand-nonpast cottage-to go-shall Let s go to the cottage that stands on the hill. b. [ NP [ CP Wh-OP i [ IP U-OP [ IP t i oka-noueni tatu-<e>]]] yamagoya i ]-ni ikou. Given the proposal in (30a), the structure of (31a) is shown in (31b), where the empty Wh-operator (represented as Wh-OP ) has been located in [Spec, CP] and is coindexed with the relative head, and the U-OP is adjoined to IP (the adjunction site will turn out crucial in the following discussion). 9 In (31b), since the verb tatu is exclusively paired with the present tense morpheme, given (30b), it must introduce a temporal variable indicated as <e>. Given (30c), in this structure, the Wh-OP binds its trace t as a variable, and the U-OP binds <e> as a variable. For these reasons, the two operators are exempted from vacuous quantification. Given (30d) and (30e), the sentence in (31a) in which the temporal variable <e> is c-commanded by the U-OP receives the everlasting interpretation. Given (30f), since (31a) receives the everlasting interpretation, the present state interpretation is also available here, as a special case of the former interpretation. Now, let us return to a fourth important aspect described in (17). The point is

26 Yoshiki Ogawa that the simple present tense in Japanese receives a present state interpretation and the everlasting interpretation if it is contained in a relative clause but not if it is in a simple clause. Compare (31a) with (32a), whose structure is (32b): (32) a. Yamagoya-ga oka-noueni tat-u (koto) Cottage-Nom hill-on stand-nonpast fact b. [ NP [ IP U-OP [ IP yamagoya-ga oka-noueni tatu-<e>]] (koto)] (32a) can receive a generic interpretation or a future modal interpretation but it cannot receive a present state interpretation or the everlasting interpretation. Why does such an asymmetry emerge between relative clauses and simple clauses? One notable difference between the two types of clauses is that the former but not the latter involve the movement of a Wh-OP to [Spec, CP] (Chomsky 1977). Considering the fact that the relevant asymmetry stems from the presence or absence of the Wh-OP and that the Wh-OP is a kind of operator, let us propose the formal licensing requirements on the U-OP, as in (33a,b) ((33a) is essentially identical with (30c)): (33) The Formal Licensing Requirements on the Empty U-OP: a. The U-OP must bind a temporal variable introduced by the verbal complex that exclusively consists of the verb stem and the present tense morpheme. b. The U-OP must be c-commanded by some semantically non-vacuous operator that is located in a scope position in overt syntax. The second requirement in (33b) is not an ad hoc stipulation proposed just in order to account for the relevant asymmetry between relative clauses and simple

27 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier clauses, but it can be subsumed under a general requirement on null operators. For example, in the parasitic gap construction, the null operator that moves inside the adjunct clause must be c-commanded by the coindexed Wh-OP that has been overtly moved to [Spec, CP] ((34a,b) are taken from Contreras 1993:3): (34) a. Which articles i did John file t i [OP j without reading e j ]? (i = j) b. *Who filed which files i [OP j without reading e j ]? (i = j) In (34a), the null OP is c-commanded by the coindexed Wh-phrase that is moved to [Spec, CP] in overt syntax. On the other hand, in (34b), the null OP is not c-commanded by the coindexed Wh-phrase, which remains in-situ in overt syntax. The c-command requirement in the overt syntax is not limited to the null OP that occurs in the parasitic gap construction but holds for almost all the null OPs, including the null OPs in the tough construction, in the relative clause, and in the comparative construction. Given these facts, we can take (33b) as being derivable from a general condition on the phonetically empty operators. 10 Given (33b), we can provide a natural account for the contrast between (31a) and (32a). First, consider (32a). Here, there is no Wh-movement involved. Hence, if the U-OP is adjoined to the clause, it cannot be c-commanded by any semantically non-vacuous operator, and hence it cannot satisfy the formal licensing requirement in (33b). Next, consider (31a). Here, the U-OP is c-commanded by the Wh-OP that is overtly moved to [Spec, CP]. One might say that the Wh-OP does not have an inherent semantic content. But it is given some semantic content by being bound by the relative head (cf. Chomsky (1986:85) for strong binding ). Hence, the U-OP is c-commanded by a semantically non-vacuous operator that is located in a scope position in overt syntax, and satisfies the formal licensing requirement in (33b).

28 Yoshiki Ogawa It should be recognized that there are a number of mismatches between the formal licensing requirement on the null OP in the parasitic gap construction and that on the U-OP. The most salient difference between them is that the formal licenser of the null OP in the parasitic gap construction is coindexed with the semantically non-vacuous operator, whereas the U-OP is never coindexed with the Wh-OP. Note that, although the null OP in (34b) is c-commanded by an overtly moved Wh-phrase, the two operators are NOT coindexed. That the coindexing requirement is imposed on the null OP in the parasitic gap construction is crucially different from the way the U-OP is licensed. Arguably, however, a number of ways to motivate (33b) on independent grounds are conceivable. One such way is to look at how the Wh-OP in the relative clause in (31b) is licensed: strong binding. Chomsky (1986:85) suggests that the strong binding from the head of a relative clause to the Wh-OP is different from ordinary binding and is more similar to predication in that the relative clause is predicated of its head. Predication is a kind of asymmetric c-command relation. Then, (33b) will follow if we assume that the U-OP must undergo strong binding by a semantically non-vacuous operator. Another way to motivate (33b) on independent grounds is to relate the U-OP to the negative polarity items (NPI) such as any. The NPI must be c-commanded by an operator in overt syntax (Progovac 1988). The most salient licenser is the negative operator as in (35a), but the interrogative Wh-phrase as in (35b) and the relative clause Wh-OP as in (35c) can also license it. Furthermore, the modal operator as in (35d), the universal quantifier as in (35e), the conditional if as in (35f) (which can be translated with universal quantification), and the comparative than as in (35g) (which Chomsky (1977) argues to induce overt Wh-movement of empty operator) can also license the NPI ((35e,f) are taken from Okuno and Ogawa 2002:8):

29 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier (35) a. In addition, don't forward any warning about a new virus. b. How shall we cover religion on any beat By David Crumm, Free Press Religion Writer? c. Find lists of famous people who were born or died on any date, religious observances, holidays, religious history. d. You can always contact us with any questions. e. Every man who insulted anyone should apologize. f. If Mary saw anyone, she will let us know. g. Balloons cause more childhood deaths than any other toy! Semantically as well, the U-OP is similar to the NPI, since sentences like You can unsubscribe at any time arguably involve universal quantification over time in the modalized world. If we can assume that the U-OP is a phonetically empty parallel of the NPI, albeit the categorial difference, then the c-command requirement on the U-OP in (33b) will follow in a straightforward manner (but see Okuno and Ogawa 2001:ch.1~4). Among the two possibilities to fill the apparent gap between the null OP in the parasitic gap construction and the U-OP, the latter alternative is not specifically referring to the fact that the U-OP is phonetically empty but focus on the fact that the U-OP cannot take the widest scope for some reason. Since it is far beyond the scope of this article to determine which possibility is closer to the truth or whether a third possibility must be envisaged, we will stop the discussion about the formal licensing requirement in (33b) and conclude this subsection by simply assuming that (33b) is one of the necessary conditions for the U-OP to be formally licensed. We have so far discussed why, in Japanese, some types of verbs can be in the simple present tense to denote a present state in a relative clause, but not in a simple clause. We have argued that this is because of the following four reasons:

30 Yoshiki Ogawa (i) verbs in the simple present tense must introduce a temporal variable, (ii) the relevant sentences must introduce the U-OP for the sake of the everlasting interpretation, (iii) the U-OP, once introduced, must be formally licensed by binding the temporal variable and by being c-commanded by a semantically non-vacuous operator, one of which is the Wh-OP, and (iv) the present state interpretation is a special case of the everlasting interpretation that is obtained when the temporal variable is bound by the U-OP The Restriction on Verb Types The postulation of the U-OP that quantifies over time in (30a) and of the empty temporal variable introduced by the simple present tense in (30b) can also provide a straightforward account for the fact that in the same relative clause in Japanese, if the verb is a stage-level stative verb or an action verb that does not denote a constant motion or an accomplishment verb or an achievement verb, then the simple present tense combined with it cannot denote a present state interpretation. Thus, (38a) is ill-formed in the relevant interpretation. If the U-OP is adjoined to the clause, (38a) will have the syntactic representation in (38b). If we adopt Parsons (1990) framework of event semantics, (38b) will be mapped into the logical representation in (38c): (38) a. *Genkan-ni hitoride tat-u hito-to hanasi-o si-you. hallway-at alone stand-nonpast man-with talk-acc do-shall Let's talk with a man who is standing alone in the hallway. b. [ NP [ CP WH-OP i [ IP U-OP j [ IP t i genkan-ni hitoride tatu-<e j >]]] hito] c. [ x, x = man [ t, t = time [ e [Standing (e) & Object (e, x) & On (e, hallway) & Hold (e, t)]]]] 11 In (38b) the Wh-OP c-commands the U-OP and the U-OP binds <e> as its

31 The Simple Present Tense in Japanese and the Phonetically Empty Universal Quantifier variable. Hence, the U-OP satisfies both of its formal licensing requirements in (33). However, (38a) is ruled out as unacceptable, since its logical representation in (38c) is semantically anomalous in terms of our knowledge about the world. In (38c), since the temporal variable t is quantified over by the universal quantifier, it must receive the everlasting interpretation. However, the action of someone s standing on the hallway is temporally delimited and even if someone is standing on the hallway at a time, he or she may move to another place at the next moment, with a rather high probability. Hence, we cannot adjoin the U-OP to a clause whose predicate denotes someone s standing somewhere. On the other hand, if the U-OP is not adjoined, some other operator that quantifies over time, such as the generic/habitual operator or the modal operator, must be adjoined to IP; otherwise, the temporal variable associated with the simple present tense remains unbound, leading to a sentence with a free variable. In (35a), however, the adjunction of the generic operator or the modal operator to the relative clause is semantically anomalous, because we now cannot see someone that stands in the hallway only generically or someone that stands there in an irrealis world. For these reasons, neither the U-OP, the generic/habitual operator, nor the modal operator can be adjoined to the relative clause in (38a). In the absence of a possible licenser, the temporal variable is left unbound. This is why (38a) is unacceptable Licensing of the Simple Present Tense in Non-Relative Clauses 4.1. The Cleft Construction In section 3, we have argued that, for some limited sets of verbs, the simple present tense is compatible with them in a relative clause but not in a simple clause, because the former but not the latter involves the Wh-OP in a scope position and that the c-commanding of the U-OP by the Wh-OP makes the former formally licensed. Given this proposal, it is predicted that, if there is another environment in