東北大学大学院文学研究科・文学部 英語学研究室



Hiroaki Emoto (2010) Ellipsis and Movement in Phase Theory


This dissertation, focusing on ellipsis and movement phenomena, attempts to refine the notion of phase in phase theory developed in the current Minimalist Program (Chomsky (2000, 2001, 2004, 2006, 2008)). This framework assumes that phases are the minimal units of linguistic computation. Syntactic operations, such as Merge, Move and Agree, apply at the phase level, and at the end of each phase, part of syntactic structure is simultaneously transferred to phonological and semantic components, thereby becoming inaccessible to further syntactic operations, so that the burden of computation is considerably reduced.

Our attempts to refine the concept of phase are divided into the following two parts. In the first part, we explore the issue of whether there are operations other than Merge, Move and Agree that apply at the phase level.We propose that Ellipsis is a phase-based operation, and demonstrate how this proposal can account for ellipsis phenomena. In particular, we argue that vP-ellipsis (VP-deletion) is triggered by the Silence feature on the phase head C, which is “inherited” by T (Chomsky (2008)).The proposal, together with the Phase Impenetrability Condition (Chomsky (2001)), can correctly account for the possible/impossible environment for vP-ellipsis in infinitival clauses. Furthermore, we show that noun phrases and sentences are parallel with respect to Ellipsis. This lends additional support to the hypothesis of the parallelism between noun phrases and sentences and the claim that DP is a phase. Finally, it is shown that out proposal and Chomsky’s assumption that syntactic derivation, semantic interpretation and phonetic interpretation apply at the same cycle/ phase, can account for some semantic phenomena which are relevant to the interaction between Ellipsis and interpretation.This provides a strong support for the notion of phase (Chapter 2).

In the second part, we argue for an even more restrictive view of movement than the conventional phase-based view that movement applies at the phase level. We propose that movable elements are restricted to phases. We show that this proposal can correctly account for the possibility or impossibility of movement in various constructions, including topicalization, cleft sentences, and pseudo-cleft sentences. In addition, we investigate the consequences of our proposal for (a) relative clauses, (b) the Left Branch Condition, (c) movement of passive and unaccusative vPs, (d) movement of small clauses to the subject position, and (e) movement of participle perception verb complements.Our claim that movable elements are restricted to phases requires us to reexamine the previous analyses of (a) and (b), since they involve movement of non-phases, and thus we attempt to offer an alternative analysis which conforms to our claim. We also show that, although (c)-(e) seem to involve movement of non-phases, these phenomena are not problems for our analysis, but rather support it (Chapter 3).

Satoru Kanno (2010) Finiteness and Specificity in the Minimalist Program


This dissertation investigates the syntactic and semantic roles of the features of phases. First, I explore the syntactic roles of the features of phases. I propose the determining condition of the phasehood of CP, which states that CP is a phase when C has both an Agree feature and a Tense feature. The proposal also states that CP is a non-phase when C does not have both of them or C does not have either of them. I argue that control complements have C that does not have both of them, that raising complements has C that lacks a Tense feature, and that tensed infinitives in Turkish have C that does not have an Agree feature. Therefore, these three types of clauses contrast with respect to Move and Agree with finite clauses whose C has both an Agree feature and a Tense feature, so that it works as a phase.

Next, I investigate the semantic roles of the features of phases. The proposal is that one way of checking an Agree feature excludes some of the possible interpretations of noun phrases. Focusing on indefinites, only a specific interpretation is assigned when they enter into an agreement relation with movement. It has been claimed that in the Germanic languages, semantic interpretations are determined by syntactic positions of subjects and objects. However, there is a growing number of papers that point out that forms of case and agreement morphemes correlate with the semantic interpretations. I argue that the main factor is positional differences of subjects and objects. I extend this claim to the internal structure of DPs.

Taichi Nakamura (2010) Aspects of the Theory of Merge


This dissertation explores within the Minimalist Program theoretical and empirical consequences of the theory of Merge proposed by Chomsky (2004, among others). We propose that the four types of Merge the theory allows are fully utilized in linguistic computation to fulfill several needs of the Conceptual-Intentional interface: (i) External Set-Merge forms a specifier-head-complement configuration and encodes a θ-theoretic (selectional) relation; (ii) External Pair-Merge produces a host-adjunct configuration and encodes a modification relation; (iii) Internal Set-Merge applies at non-final landing sites of movement and provides positions dedicated to no specific interface need such as reconstruction positions; (iv) Internal Pair-Merge applies at final landing sites of movement and provides discourse-related or scope positions.

Chapter 2 discusses empirical consequences of our proposed theory of Merge including an adjunction-based unification of subject and adjunct islands, a novel account of (Anti-)Freezing Effects, a substitution-based unification of Japanese scrambling and successive cyclic movement, and a PF-based characterization of EPP.

Chapter 3 examines interactions between our proposed theory of Merge and the theory of Agree, focusing on Rizzi’s (1990) Relativized Minimality principle. We address a problem concerning intervention effects and their repair by movement. Our solution to this problem by elaborating the chain mechanism not only simplifies the Minimality principle to much a greater degree but also offers a new perspective on the principle.

Chapter 4 investigates the issue of whether relative clauses are all Pair-Merged adjuncts or not. We consider what we call determiner-headed free relative clauses (DHFR), e.g. This is the what you do against the Rams, and show that the properties of DHFRs are easily accounted for by the adjunction-free, raising analysis of relative clauses but not by the adjunction-based, matching analysis. We thus support the view that the raising and matching analyses are both necessary to account for a wide range of properties of relative clauses.

Takahiro Tozawa (2009) Suntax of Subordinate Clauses in English: A Minimalist Approach


This thesis is an investigation of the resultative construction within the framework of X' Semantics. I propose that "resultative phrases" are modifiers in the semantic structure of verbs: "resultative phrases" added to sentences based on verbs of change of state are modifiers of the Stage predicate in the semantic structure of these verbs and "resultative phrases" added to sentences based on activity verbs are degree modifiers specifying the degree to which the activity denoted by the verb has been carried out. On this approach, there are two kinds of "resultative phrases": "resultative phrases" as modifiers of the Stage predicate in the semantic structure of change of state verbs and "resultative phrases" as degree modifiers. The former are universally found since change of state verbs are universally found. This distinction is supported by Romanian resultative constructions. Romanian allows only "resultative phrases" that function as modifiers of the result states lexicalized by change of state verbs, and does not allow the addition of XPs that function as degree modifiers. This descriptive generalization can be accounted for by the assumption that Romanian has "resultative phrases" as modifiers of the Stage predicate, but not "resultative phrases" as degree modifiers, that is, Romanian has no operation of "resultative phrase" addition as English does.

The resultative construction has been assumed to be constrained by Tenny's (1987, 1994) Single Delimiting Constraint (or Goldberg's Unique Path Constraint), which asserts that only one delimiting per verb phrase is allowed: since "resultative phrases" act as delimiters of the events denoted by verbs, two resultative phrases cannot cooccur. However, there are examples in which two resultative phrases, a resultative phrase and a directional phrase, can cooccur, contrary to the predictions of this constraint. In order to account for this fact, I propose that resultative phrases denoting a change of state are added as modifiers to the semantic structure of verbs, while resultative phrases denoting goals, namely directional phrases, are added as a result of the application of semantic conflation. Though these two resultative phrases are really delimiters, one of them delimits a change of state aspect of the event, while the other delimits a directed motion aspect of that event. Thus, these two kinds of delimiters delimit different aspects of one and the same event. To this case, the Single Delimiting Constraint cannot apply.

Mika Takahashi (2003) Optimality in Syntactic Derivations


This dissertation attempts to provide an answer to the questions of "what is an optimal derivation?" in the Minimalist Program (Chomsky 1998, 1999, 2001), focusing on the role of locality conditions. Specifically, I argue that optimality involves Barriers-type locality, but does not involve Relativized Minimality (RM)-type locality.

The main arguments for the claim above are divided into the following three sets: (i) purported RM effects in head movement, raising and wh-islands can be deduced from Barriers-type locality conditions or some independently motivated principles. This type of argument receives support from superraising in English, Niuean and Georgian and superpassive in Japanese (chapter 3). (ii) In phase theory put forth by Chomsky (1998, 1999, 2001), I address a new problem concerning the assignment of P- and EPP- features to a phase head. My solution to this problem will make it possible to deduce the so-called Superiority effects from the latest version of Barriers-type locality conditions. This analysis can give a principled account to some multiple wh questions which are problematic for RM-type locality conditions (chapter 4). (iii) A certain property of ECM constructions with infinitival complements in Japanese, which cannot be predicted by RM-type locality conditions, can be correctly predicted by a Barriers-type locality condition (chapter 5).

Naoto Tomizawa (2003) Some Concepts and Consequences of the the Derivational Computational System


This thesis investigates the nature of the strictly derivational model of computational system, with special reference to the procedures of semantic feature licensing and formal feature checking: the stage at which semantic feature licensing is achieved, the formal mechanism required for such licensing and checking, and the consequences of such procedures. I propose that Lexical Array is an ordered set of lexical items, where semantic relations among the items are specified. Derivation, then, is (partly) a process to license such semantic relations as early as possible under certain structural configurations. The derivational licensing and checking, coupled with a freezing principle which we call the Complementary Fixation, derive such island effects as the subject condition, adjunct condition, and "shifted object" condition (Chapter Two). The formal mechanisms required for the derivational licensing and checking are uniformly Merge (the Generalized Copy Theory). The traditional Move is now external Merge, which explains the obviation phenomena of the BC (C) effects in certain 'wh'-movement configurations and the antecedent-contained deletion constructions (Chapter Three). It follows from our derivational analysis that the notions of minimal domain, accessible subject, BT-compatibility, and c-command are unnecessary in the account of the anaphor-antecedent relations (Chapter Four), and that the distributional properties of 'there' and its associate DP in the 'there'-constructions are derived from the defective D-feature of 'there,' with no bearing on the operational "cost" of Move/Merge or Case (Chapter Five). The derivational approach to the relation of 'wh'-elements and their Q markers, coupled with parametric specifications of Q markers, accounts for the differences in availability of 'wh'-movement among English, Chinese, and Japanese (Chapter Six).

Etsuro Shima (1998) Syntactic Operations in the Minimalist Program


This dissertation, conducted within the Minimalist Program outlined and further elaborated in Chomsky (1993, 1994,1995), seeks a maximally simple design for the computational systems, with special focus on the relation between Merge and Move, triggers of movement operations, and characterization of strong-weak distinction of formal features. Contrary to Chomsky's claim, I propose that Move is selected over Merge. This proposal provides a unified analysis of the following three seemingly distinct syntactic phenomena: strict cyclicity, superraising, and overt object shift. Furthermore, as to movement operations, I argue that NP-movement is triggered by a Case-feature instead of a categorial-feature of Tense, and that wh-movement involves two types of features: one is involved in interpretation at LF and the other relevant to movement. My proposal about the trigger of movements gives a natural account of raising, expletive, partial wh-movement, and multiple-question constructions. Finally, I claim that both a strong and weak formal feature must be checked off before Spell-Out, and that the only difference between these two formal features is whether or not phonological features are attracted: a strong formal feature attracts phonological features, whereas a weak formal feature leaves phonological features in the original position. My characterization of strong and weak formal features eliminates A/A'-distinction about trace-deletion and a stipulation for checking.

Yoshiki Ogawa (1997) A Unified Syntax of Verbal and Nominal Projections


This dissertation provides a unified theory of verbal and nominal projections, in which "elements" in clauses are parallel to those in noun phrases in all respects, and "syntactic operations" apply similarly in both clauses and noun phrases. By "elements," I mean lexical and functional categories, null affixes, Case-features, phi-features, definiteness features, event arguments, thematic arguments and adjunct elemnts. By "syntactic operations," I means Merge, Move, feature checking, argument suppression, event-identification, and so on. In this unified theory of verbal and nominal projections, I argue for the following analyses or proposals, among others: a syntactic verb raising analysis of nominalization and an ECP-free account of the distribution of null complementizers (chapter 2), checking of definiteness features by Spec-Head agreement (chapter 3), the relation between the definiteness of functional categories and the possibility of multiple Specifier licensing (chapter 3), deriving of the Case-adjacency effects from a general condition on null affixes (chapter 4), the stage-level/individual level distinction for non-derived nominals (chapter5), a syntactic verb raising analysis of middle formation (chapter 6), and ellimination of levels of lexical representation (chapter 6). Interaction of these proposals leads us to simplification of the model of grammar, and hence is in accordance with the spirit of the minimalist program of linguistic theory (Chomsky 1993, 1994, 1995).