This paper is concerned with cleft constructions and reconstruction effects in English and Japanese. Japanese cleft constructions involve two different syntactic dependencies, movement and deletion. This assumption explains facts that have not been reported in the literature. The reflexive pronoun in (la) and the reciprocal pronoun in (lb) in the focus phrase can be bound either by the higher subject or by the lower subject in the presupposition. In clear contrast, the lower subject in Japanese cleft constructions cannot bind anaphors in the focus phrase. In (2), only the higher subject can bind the anaphors in the focus phrase. What explains the contrast between (1) and (2)? We argue that an operator in Japanese moves from the position adjoined to the lower clause (tk in (3)), not from the thematic gap position (ek). It is shown that the dependency (ii) in (3) stems from movement, and (i) from deletion. Since Opk (or the focus phrase associated with it) reconstructs only to the position of tk, the anaphor can only be bound by the higher subject, Sallyi-Nom.

A variety of linguistic evidences have been to support the assumption that depictives and resultatives are different in their thematic structures with a main predicate, despite they appear identical on the surface. Both depictives and resultative contain a subject NP, a verb, an object NP, and an adjective following the object NP, as shown in (1)-(2). However, the thematic relation between the verbs and the adjectives are different. In depictives, the adjective characterizes the object NP in relation to the action or process described by the verb. Thus, (la) means: 'I ate the beef and at the time I ate it, it was raw.' The NP is characterized at the time of the action of the verb. In the resultatives, the final adjective characterizes the state of the object NP, a state which results from the action or process described by the verb. Thus, (2a) means: I caused the window to be clean by wiping it; the adjective describes the final state of the NP. (1) a. I ate the beef raw. b. I ate the food cold (depictives) (2) a. I wiped the window clean. b. I kicked the door open. (resultatives) Recently it has been proposed that the argument structure is reflected by sentence accentuation (Schmerling 1976, Gusenhoven 1983, Selkirk 1984). The main claim in the focus theory (e.g., accent percolation theory) is that in a focus constituent consisting of a head and an argument, the accent is realized on the argument, everything else being equal (Gussenhoven 1983, Selkirk 1984). Uhmann (1991) proposes that if focus is assigned to a constituent, all the phonological phrases of that constituent bear an accent. She also points out that a head and an argument form a single phonological phrase, whereas a head and an adjunct form a separate phonological phrase, Following Uhmann (1991), I assume that a pitch accent is the evidence for phonological units and manifests focused words or constituents. The accentual differences between head-argument and head-adjunct are clearly shown by Gussenhoven (1992) as follows: when a head-argument structure is in focus, as in (3), the accent falls on the argument, tent, while when a head-adjunct structure is in focus, as in (4), an accent is realized on the head smoked and the adjunct tent, which are in separate phonological phrases. Likewise, both gerookt 'smoked', and tent 'tent' are accented in Dutch. The phonological phrase is represented by parenthesis, the focus structure is by bracket and the accented words are capitalized. (3) a. John [ ( stayed in the TENT) l b. John [ ( in the TENT gebleven)] F (4) a. John [(SMOKED)(in the TENT)] F b. John [(heeft in the TENT) (gerookt)]F (Gussenhoven 1992: 94) Does the distinction between an argument and an adjunct exist? My goal in this paper is to investigate the issue of how the theory of focus applies to the different types of secondary constructions, namely depictive and resultative constructions. The experiments are conducted to examine the statuses of resultative and depictive constituents in terms of their focus marking.

In this paper, I will discuss the semantics of the Japanese so-called "Counter-Equi NP" (henceforth, CENP) Construction, given in (I). (1) CENP Construction Keisatsu-wa [ doroboo-ga nige-ru ]-tokoro-o tsukamae-ta. police-TOP burglar-NOM escape-PRES-occasion-ACC arrest-PAST 'The police arrested a burglar on the occasion during which he/she was escaping.' The CENP construction is similar to the so-called "internally headed relative clause" (henceforth, IHRC) construction, given in (2), in that, in the CENP construction, an NP within the embedded tokoro-clause is interpreted as an argument of the matrix verb. (2) IHRC Construction Keisatsu-wa [ doroboo-ga nige-ru ]-no-o tsukamae-ta. police-TOP burglar-NOM escape- PRES-NO-ACC arrest-PAST 'The police arrested a burglar on the occasion during which he/she was escaping.' In both (1) and (2), the embedded subject doroboo 'burglar' is interpreted as an object of the matrix verb tsukamaer 'arrest'. In this paper, I argue that the noun tokoro semantically has a dual function. To be more specific, it is a generalized quantifier over an entity and at the same time an event when it combines with the tokoro-clause, adopting Srivastav's (1991) generalized quantifier approach to Hindi correlatives.

Introduction (1) English tough-constructions (TCs, e.g., Bill is easy to please, Mary is hard to work with) have caused considerable controversy about their correct analysis and this controversy can be roughly described by two main types of analyses. The first type, which can be called "strong connectivity" analysis, assumes syntactic connectivity between the tough-subject (i.e., the subject of the main clause in a tough-sentence) and the gap in the infinitival phrase/clause. Thus, studies along this line argue that the tough-subject is not the true subject of the tough-adjective but is generated as the object of a verb or preposition in the infinitival phrase/clause and moved to the subject position (Postal 1971, Postal & Ross 1971, Berman 1973, Postal 1974, Comrie & Matthews 1990, etc.)(2). This position is supported by the well-known fact that typical toughadjectives such as easy, hard, and difficult generally have no or little semantic effect on their subjects. On the other hand, the second type, which can be called "weak connectivity" analysis, argues that there is no syntactic connectivity between the tough-subject and the gap and that toughadjectives subcategorize for an infinitival phrase with a gap. Thus, early transformational studies such as Ross (1967: 231), Akmajian (1972), and Lasnik & Fiengo (1974) say that the object of the complement of a tough-adjective has been deleted. Government-Binding (GB) theory, for example, Chomsky (1977: 102-110, 1981: 308-314), proposes the movement of an empty operator that binds the trace in the object position and is coindexed with the subject. Furthermore, although no movement of an empty operator is posited, the analysis of HeadDriven Phrase Structure Grammar is similar to that of GB theory in that it does not assume syntactic connectivity. Thus, Pollard & Sag (1994) analyze TCs as a lexical fact about some special predicates and assume that such predicates as easy, difficult, take, and cost subcategorize for infinitive complements containing an accusative NP gap which is coindexed with the subject. Even though the second type of approach has long been more favored by current syntactic frameworks, it is not clear whether there is sufficient empirical evidence to support this more dominant, second type of analysis ( cf. Jones 1983 )(3). Since synchronic linguistics doesn't seem to be able to resolve this controversy one way or the other, what then does the diachrony of the relevant parts of English grammar say about the analysis of TCs? Does diachrony argue for any particular position?

This paper examines the stages of development in the acquisition of Dutch main stress by children. In the early stages of prosodic development, there are strong restrictions on word size. These restrictions can be explained by the high ranking of alignment constraints (McCarthy & Prince 1993) over faithfulness (McCarthy & Prince 1995). Moreover, the truncation and early prominence patterns are due to adherence to Strict Layering (Selkirk 1984) which emerges from the constraint ranking. The subsequent violation of Strict Layering (Weak Layering, Ito & Mester 1992) at later stages arises from constraint re-ranking. This analysis provides a unified account of child language and adult grammar. The development of the child's grammar moves from unmarked to marked structure (Demuth 1997) building on the units of the Prosodic Hierarchy (Selkirk 1980). Working within the Optimality Framework (Prince & Smolensky 1993), using well-attested constraints, the relationships between stages of acquisition are explained through constraint re-ranking. The paper is organized as follows. First I provide a general overview of the stages of stress acquisition. I then discuss on each stage individually focussing first on the size restrictions found in the early stages of development before turning to stages 3 and 4.

This paper explores the nature of possessive relations in Universal Grammar, and their involvement in a syntactic process known as Possessor Raising or, as it is usually called in the Relational Grammar (RG) tradition, Possessor Ascension. Possessor Raising can be defined as the transformation that takes the D-structure possessor of a direct object in the sentence and assigns to it a surface grammatical relation (GR) to the verb of the sentence.