«Title of Document: A MOVEMENT ACCOUNT OF LONGDISTANCE REFLEXIVES Rebecca Katherine McKeown, Doctor of Philosophy, 2013 Directed By: Professor Norbert ...»
Title of Document: A MOVEMENT ACCOUNT OF LONGDISTANCE REFLEXIVES
Rebecca Katherine McKeown, Doctor of
Directed By: Professor Norbert Hornstein
Department of Linguistics
This thesis examines reflexive pronouns, such as Icelandic sig (Cf. Thráinsson
2007), which may be bound from outside of an infinitive clause (which I call MD
“medium distance” binding) in addition to being bound locally. I propose that such reflexives are linked to their antecedents via sisterhood followed by movement: the reflexive and antecedent are first merged together as sisters, and the antecedent
subsequently moves to receive its first theta-role, as schematized below:
1. He ordered Harold to shave he+sig This links the properties of bound simplex reflexives to the properties of movement. I argue that reflexives such as sig must be bound within the first finite clause because finite CP is a spell-out domain and its escape hatch is inaccessible to A-movement.
Furthermore, I derive the subject-orientation of sig and other simplex reflexives from merge-over-move, combined with a numeration divided into phases including vP. Since the antecedent is moving into its first theta-role, and merge is preferable to move, the antecedent will end up in the highest position in the phase: that is, the subject.
I then examine long-distance (LD) uses of sig as well as Chinese ziji, Japanese zibun, and Kannada tannu. I propose that in such cases the reflexive still has a double, which is not the antecedent but a null element, possibly an operator. It undergoes A’ movement to a position in the left periphery of a finite clause, associated with point-ofview (with a divided left periphery as in Speas 2004)—and this operator is in turn associated with an antecedent either outside the finite clause, or outside the sentence entirely. This accounts for the observation that LD reflexives often must refer to POV holders (Sells 1987). Evidence for LD reflexives being mediated by an A’ position comes from the interaction of binding with wh-movement in Kannada (Lidz 2008), and is one way of describing where blocking effects do and do not occur in Chinese (Anand 2006). Furthermore, in Japanese there are sometimes overt morphemes, potentially leftperiphery heads, that indicate POV and can co-occur with the use of LD reflexives (Nishigauchi 2005, 2010).
A MOVEMENT ACCOUNT OF LONG-DISTANCE REFLEXIVESBy Rebecca Katherine McKeown Dissertation submitted to the Faculty of the Graduate School of the
Professor Norbert Hornstein, Chair Professor Howard Lasnik Professor Jeffrey Lidz Dr. Tonia Bleam Professor Susan Dwyer Acknowledgements I’d like to thank my advisor, Professor Norbert Hornstein, for all of his helpful suggestions and particularly for his patience and moral support. Thank you also to all of my committee members for your proofreading, editing, and advice. I’d also like to thank Professor Taisuke Nishigauchi for the useful discussions we had, and to thank Professor Kyle Johnson for his helpful comments on an early presentation of my ideas at the ECO5 conference in 2007.
I am indebted to the many people who gave me data, particularly Kjartan Ottóson, RIP. I also got data from Maki Kishida, Eri Takahashi, Shin Tanigawa, Chizuru Nakao Nakano, Wing Yee Chow, Terje Lohndal, Minna Lehtonnen, Halldór Sigurðsson, and others. Thank you all very much for your help and for your assistance in making finicky grammatical judgments about complicated sentences. Also thank you very much to Professor Jeffrey Lidz for your translations of the Kannada scenarios in the appendix.
I would also like to thank all of the graduate students and faculty of the University of Maryland who attended my defense, my 895 defense, the binding theory reading group, and various other earlier presentations of my work. Finally, I’d like to thank friends and family for their support and encouragement, and my parents for giving me a celebratory sword.
Table of Contents
Chapter 1: Introduction
1.1 The pronouns
1.2 My analysis: an overview
Chapter 2: SE reflexives and movement
2.1 Movement and Binding
2.2 SE reflexives
2.3 Icelandic sig
2.3.1 Properties of local and MD sig
2.3.2 Sig as a sub-part of a complex reflexive
2.3.3 Logophoric uses of sig
2.4 Doubling and antecedent movement
2.5 An overview of A-movement
2.5.1 Minimality in A-movement to a case position
2.5.2 Movement from one theta position to another
2.5.3 Minimality in A-movement
2.5.4 SE reflexives involve a type of A-movement
2.5.5 Sig and minimality
2.6 Deriving the properties of sig through movement constraints
2.6.1 Deriving the locality of sig: spell-out domains
2.6.2 Deriving the subject-orientation of sig
2.6.3 The Numeration
2.7 Sample Derivations
2.7.1 Locally-bound sig
2.7.2 MD-bound sig
2.7.3 Binding out of multiply embedded infinitive clauses
2.7.4 Prohibiting ‘vacuous merges’
2.8 MD reflexives: a summary
Chapter 3: Long-Distance Reflexives, Movement, and Point-of-View
3.1 Long-distance reflexives: an introduction
3.2 Extending a doubling-and-movement account to LD reflexives
3.2.1 Why do I group local and MD uses together, but not LD
3.3 Long-distance reflexives refer to point-of-view holders
3.3.1 Long-distance reflexives as sources
3.3.2 Long distance reflexives as selves
3.3.3 LD reflexives used as pivots
3.3.4 Empathic vs. logophoric LD reflexives: Oshima’s terminology................ 138 3.3.5 LD reflexives and subject orientation
3.3.7 MD and LD reflexives in the same language
3.4 Coding Discourse Roles in the Syntax
3.4.1 Subjunctive mood in Icelandic
3.4.2 Island effects caused by Kannada tannu
3.4.3 Abe n-pronouns require left-periphery binders
3.4.4 Multiple ziji and what it indicates about the left periphery
3.4.5 Japanese modal heads and zibun
3.4.6 Summing up
3.5 Doubling and movement
3.5.1 Some evidence in favor of a movement account: locality
3.5.2 More evidence in favor of a movement account: successive-cyclicity....... 244 3.5.3 An attempted set of further tests
3.5.4 A concern for my account: LD reflexives do not show adjunct island effects 3.5.5 Summary
3.6 Can antecedents of LD reflexives move directly?
Chapter 4: Conclusion, with notes on typology
This thesis examines reflexive pronouns that can take antecedents outside of their immediate clause. I argue that such pronouns are related to their antecedents by sisterhood, followed by movement of the antecedent. In this introduction I first discuss the class of pronouns that I will be considering for the rest of the thesis (section 1.1), and then present a brief overview of my analysis (section 1.2).
1.1 The pronouns This thesis addresses a class of pronouns that have been termed long-distance reflexives. These are pronouns that are like local reflexives (e.g. English himself) in some ways: they require antecedents (usually), and allow local binding—like local reflexives, and unlike non-reflexive pronouns. For instance, Icelandic sig patterns with English himself in (1) and (2) below.
However, unlike himself, and other reflexives that (mostly) obey “Condition A” (Chomsky 1981), these reflexives may take an antecedent outside of their immediate governing category.
I will distinguish between two different types of long-distance use of reflexives.
Some of the reflexives I consider, particularly Icelandic sig, have what I’ll call “mediumdistance” (MD) readings in which they allow binding from outside on an infinitive clause, such as a control or ECM clause.
Other reflexives (including sig, but also Chinese ziji, Japanese zibun, and Kannada tannu) have uses in which they may take an antecedent from outside of one or more finite clauses.
‘Ramai says that selfi,*j is very clever.’ [Amritavalli 2000 ex. 9] Even though some of the same reflexives may be used with either medium-distance or long-distance antecedents, I will argue that the properties of these uses are different. I will argue that local and medium-distance binding of a reflexive such as sig are established in the same way: the antecedent starts out as a sister to the reflexive and moves to its surface position. In contrast, I will argue that long-distance uses of reflexives including sig are mediated by a left-periphery position associated with point-of-view.
However, my accounts of long-distance and medium distance reflexives will not be entirely disjoint.
Some of the reflexives I examine, including Icelandic sig, have been described in the literature as “simplex expression” (SE) reflexives (Reinhart and Reuland 1991). I will sometimes use this terminology as well. As Reinhart and Reuland note, these SE reflexives are often morphologically simpler than other reflexive pronouns in the same language. Many times, these SE reflexives can be used along with another morpheme, analogous to English “self”, as in Icelandic sjálfan sig (self SE). In this respect, they resemble ordinary pronouns, which may also be used along with self (English himself, Icelandic hann sjálfan—literally him+self).
For some linguists, the morphologically simple nature of these reflexives is important to explaining their movement properties. For example, Pica (1985, 1987), writes that reflexives such as sig may take antecedents outside their immediate clause because they undergo head movement that brings them near to their antecedents: this head movement requires that they be heads, which sig is but sjálfan sig is not. Reuland (2001a,b), meanwhile, argues that SE reflexives are noteworthy for having fewer syntactic features than ordinary pronouns in the same language. For instance, Icelandic sig is mandatorily third-person, but does not have number or gender. Reuland’s account relies the fact that sig lacks inflection for number to explain how it differs in use from other Icelandic pronouns, such as hann ‘him’ (3msg). For me, although I will use the term SE reflexive to refer to this class of reflexives, I want to note that my analysis of them differs from Reinhart and Reuland’s. In particular, I remain agnostic on how important the morphological simplicity of these pronouns is to their ability to be used long-distance. Furthermore, at least one of the reflexives that I want to account for, (Kannada tannu, has a plural form, taavu, so my analysis cannot rely on underspecification for number features (Amritavalli 2000).
Additionally, in classifying reflexives that have long-distance uses, many linguists have noted subcategorizational and semantic constraints on the local uses of the same words. For example, some reflexives with LD uses can only be coargument-bound when the predicate in question is particularly well-suited to reflexive meanings. In Kannada, for instance, verbs may be made lexically reflexive by the addition of an extra morpheme.
The reflexive tannu, which also has LD readings, can be bound locally if used with a reflexive-marked verb, but not with its non-reflexive-marked counterpart. In contrast, the morphologically complex form of tannu, which does not allow LD readings, may be used locally without the reflexive morpheme on the verb.
Lidz analyzes the restrictions on these uses of tannu as going along with semantic differences between tannu and the complex form of tannu. There are interesting interpretive differences between (7b) and (7c). Notably, (7c) allows ‘near-reflexive’ readings in which the object and subject of ‘see’ are not quite the same person: for instance, (7c) could mean that Hari saw literally saw himself (as in a reflection), or more loosely that Hari saw a representation of himself, such as a statue of himself. In contrast, (7b) requires a “true-reflexive” interpretation: it works if Hari saw his own reflection, but not if he saw a statue. Similar verb restrictions and interpretive differences have been found for the simplex vs. complex forms of reflexives in other languages, such as for instance Dutch zich compared to Dutch zichzelf (see for instance Reinhart and Reuland 1993).
Although it is true that a variety of simplex LD reflexives show near-reflexivity, I will not be using it as one of the diagnostics connecting all of the reflexives I am considering. As Lidz notes, not all LD reflexives require true-reflexive interpretations.
Chinese ziji, for example, allows both local and LD readings—but in its local readings it may be used with any verb and it allows for near-reflexive interpretations.
Because I am pursuing a unified account of how reflexives like tannu and ziji come by their long-distance interpretations, I will not be directly linking true-reflexivity and LD binding.
For the purposes of this dissertation, then, I will assume that when a relation is established between a reflexive and its antecedent, that the relation may not be one of complete identity. However, I will further assume that in languages where the complex reflexive allows near-reflexive interpretations whereas the simplex reflexive does not, that this is because the meaning of the SE reflexive (established in a sisterhood relation with the antecedent) can itself be further modified by another morpheme such as –self.