SYMMETRY IN SYNTAX
While much has been written on asymmetric aspects of sentence structure, symmetric aspects have been largely ignored, or claimed to be non-existent. Does symmetry in syntax exist, and if it does, how do we account for it? Barbara Citko sets out to tackle these questions and offers a unified approach to a number of phenomena that have so far been studied only in isolation. Focusing on three core minimalist mechanisms, Merge, Move – and Labeling – she advances a new theory of these mechanisms, by showing that, under certain well-defined circumstances, Merge can create symmetric structures, Move can target either of two potentially moveable objects, and labels can be constructed symmetrically from the features of two objects. This book is aimed at researchers and graduate students interested in minimalist syntax, the structure of questions, relative clauses, coordination, double object constructions, and copular sentences.
barbara citko is Assistant Professor of Linguistics at the University of Washington in Seattle. Her research includes work on phrase structure, coordination, relative clauses, wh-questions and the syntax of Slavic languages.
CAMBRIDGE STUDIES IN LINGUISTICS
General Editors: p. austin, j. bresnan, b. comrie, s. crain, w. dressler, c. j. ewen, r. lass, d. lightfoot, k. rice, i. roberts, s. romaine, n. v. smith
Symmetry in Syntax
Merge, Move, and
Labels
In this series
84 nomi erteschik-shir: The dynamics of focus structure
85 john coleman: Phonological representations: their names, forms and powers
86 christina y. bethin: Slavic prosody: language change and phonological theory
87 barbara dancygier: Conditionals and prediction
88 claire lefebvre: Creole genesis and the acquisition of grammar: the case of Haitian creole
89 heinz giegerich: Lexical strata in English
90 keren rice: Morpheme order and semantic scope
91 april mcmahon: Lexical phonology and the history of English
92 matthew y. chen: Tone Sandhi: patterns across Chinese dialects
93 gregory t. stump: Inflectional morphology: a theory of paradigm structure
94 joan bybee: Phonology and language use
95 laurie bauer: Morphological productivity
96 thomas ernst: The syntax of adjuncts
97 elizabeth closs traugott and richard b. dasher: Regularity in semantic change
98 maya hickmann: Children’s discourse: person, space and time across languages
99 diane blakemore: Relevance and linguistic meaning: the semantics and pragmatics of discourse markers
100 ian roberts and anna roussou: Syntactic change: a minimalist approach to grammaticalization
101 donka minkova: Alliteration and sound change in early English
102 mark c. baker: Lexical categories: verbs, nouns and adjectives
103 carlota s. smith: Modes of discourse: the local structure of texts
104 rochelle lieber: Morphology and lexical semantics
105 holger diessel: The acquisition of complex sentences
106 sharon inkelas and cheryl zoll: Reduplication: doubling in morphology
107 susan edwards: Fluent aphasia
108 barbara dancygier and eve sweetser: Mental spaces in grammar: conditional constructions
109 hew baerman, dunstan brown and greville g. corbett: The syntax–morphology interface: a study of syncretism
110 marcus tomalin: Linguistics and the formal sciences: the origins of generative grammar
111 samuel d. epstein and t. daniel seely: Derivations in minimalism
112 paul de lacy: Markedness: reduction and preservation in phonology
113 yehuda n. falk: Subjects and their properties
114 p. h. matthews: Syntactic relations: a critical survey
115 mark c. baker: The syntax of agreement and concord
116 gillian catriona ramchand: Verb meaning and the lexicon: a first phase syntax
117 pieter muysken: Functional categories
118 juan uriagereka: Syntactic anchors: on semantic structuring
119 d . robert ladd: Intonational phonology second edition
120 leonard h. babby: The syntax of argument structure
121 b. elan dresher: The contrastive hierarchy in phonology
122 david adger, daniel harbour and laurel j. watkins: Mirrors and microparameters: phrase structure beyond free word order
123 niina ning zhang: Coordination in syntax
124 neil smith: Acquiring phonology
125 nina topintzi: Onsets: suprasegmental and prosodic behaviour
126 cedric boeckx, norbert hornstein and jairo nuňes: Control as movement
127 michael israel: The grammar of polarity: pragmatics, sensitivity, and the logic of scales
128 m. rita manzini and leonardo m. savoia: Grammatical categories: variation in romance languages
129 barbara citko: Symmetry in syntax: Merge, move, and labels
Earlier issues not listed are also available
1 Rationale
1.1 Introduction
While there has been a lot of research on asymmetry and antisymmetry in syntax, symmetry has been mostly ignored or claimed to be outright impossible (Kayne 1994, Di Sciullo 2002, 2005). This is somewhat surprising from a biolinguistic perspective, which seeks to integrate linguistics with the natural sciences, where symmetry is the normal state of affairs and asymmetry requires an explanation (as pointed out by Boeckx and Piattelli-Palmarini 2005, Brody 2006, Chomsky 2005, Jenkins 2000, among others). My main goal in this book is to remedy this gap by examining symmetric aspects of three fundamental syntactic mechanisms: the mechanism responsible for recursion, the mechanism responsible for displacement, and the mechanism responsible for determining the categories of syntactic objects. I look at these three mechanisms through the lens of Chomsky’s minimalist program, which takes the mechanism responsible for recursion to be External Merge (often referred to simply as Merge), the mechanism responsible for displacement to be Internal Merge (often referred to simply as Move) and the mechanism responsible for determining categories of both Merge and Move structures to be Labeling. The standard minimalist assumption is that the structures created by Merge are asymmetric (because only such structures can be linearized), that Move is asymmetric (because it ‘privileges’ one of two potentially movable elements) and that labels are asymmetric (because they contain features of only one element). In the course of the book I will challenge these three assumptions and argue that Merge can also create symmetric structures, that Move can sometimes treat two elements in a symmetric fashion, and that labels can sometimes contain features of two objects undergoing Merge. The rest of this introductory chapter serves three goals. First, it provides a general introduction to the concepts of symmetry, asymmetry and antisymmetry. It outlines what these concepts mean in general, as well as in
more specific, linguistic terms. Second, it provides an overview of the theoretical framework assumed throughout the book, the minimalist program. The overview focuses on the workings of Merge, Move and Labeling, which are at the core of the claims I advance in the book. This chapter also explains why the empirical focus of the book is on symmetric aspects of these three mechanisms, as opposed to many other phenomena that the image of symmetry in syntax might conjure. And third, this introductory chapter provides an overview of the rest of the book.
1.2 Symmetry and asymmetry
The terms symmetry and asymmetry are used in two different ways in the literature. One is a fairly intuitive non-technical sense, and the other one is somewhat more technical and tends to vary from discipline to discipline.
In its non-technical sense, the term symmetry is used to refer to the similarities between two parts of an object (or two objects), and the term asymmetry to the differences between them. In a linguistic context, the objects in question could be syntactic features, categories or transformations. Let us first look at a couple of simple cases. For example, we know that arguments differ from adjuncts in that they are bearers of theta roles. Thus we might speak of the symmetric behavior of different types of arguments (i.e. subjects and objects) with respect to theta theory, and the asymmetric behavior of arguments and adjuncts in the same respect. Another well-studied example involves cross-categorial symmetry, such as the symmetry between noun phrases and clauses, which has been studied quite extensively at least since Chomsky’s (1970) “Remarks on nominalization” (see Abney 1987, Douglas-Brown 1996 and Hiraiwa 2005, among others, for more recent ways to capture this symmetry). The data in (1a–b) illustrate the symmetric behavior of noun phrases and clauses with respect to theta role assignment.
(1) a. The RomansAgent destroyed the cityTheme b. the RomanAgent destruction of the cityTheme
And the Hungarian data in (2a–b) illustrate the symmetric behavior of subjects and possessors with respect to case marking; both are marked with the same (nominative) case. Furthermore, the possessee in (2b) agrees with the possessor in a way that parallels subject–verb agreement.
(2) a. Te ve-tt-el egy kalap-ot.
2SG.NOM buy-past-sg.indef indef hat-acc ‘You bought a hat.’
b. a te kalap-ja-i-d D 2SG.NOM hat-poss.pl-2.sg
‘your hats’ (Hiraiwa 2005:19–20, citing Szabolcsi 1994:186)
In a more technical (not necessarily linguistic) sense, the terms symmetry and asymmetry are used to describe geometric patterns, or relationships between two elements in a set. In geometric terms, an object is symmetric if it can remain unchanged when a transformation applies to it. Geometric figures under rotation transformation provide a straightforward illustration. A circle, for example, is symmetric under any rotation; if we rotate it by any degree, the result is still going to be a circle, as shown in (3a–c). A diamond, on the other hand, is only sometimes symmetric, as shown in (4a–c). If we rotate it by 45 degrees, the result is a square. However, if we rotate it by 90 degrees, the result is a diamond again.
(3) a.
b. 45° rotation c. 90° rotation
(4) a.
b. 45° rotation c. 90° rotation
Mathematicians distinguish four types of symmetric transformations: reflection or mirror symmetry, rotation symmetry, translation symmetry and glide reflection symmetry (see Lee 2007 for an accessible overview). Rotation rotates an object (as we have just seen), translation shifts it (whilst preserving its orientation), reflection yields a mirror image of it, and glide reflection combines reflection and translation. As we will see shortly, the ones that apply most straightforwardly to linguistic patterns are translation and reflection symmetries, illustrated in (5a–b).
(5) a. translation symmetry
b. reflection or mirror symmetry
In set theory, the terms symmetry and asymmetry are used to refer to binary relationships between elements in a set.1 This is by far the most common usage of the two terms in linguistics. A relationship between two elements in a set is symmetric if for every ordered pair <x,y> in the set, the pair <y,x> is also in that set. A good illustration comes from the domain of kinship terms; the relationship ‘cousin of’ is an example of a symmetric relationship. If John is Bill’s cousin, Bill has to be John’s cousin as well. A relationship between two elements is asymmetric if it is never the case that for any pair <x,y> in the set, the pair <y,x> is in the same set. The relation ‘is older than’ is asymmetric; if John is older than Bill, Bill cannot be older than John. A related concept is that of antisymmetry. A relationship between two elements in a set is antisymmetric if whenever both <x,y> and <y,x> are members of the set, x must be the same as y.
With this general background on symmetry (and asymmetry), we are almost ready to begin our examination of symmetry in syntax. First, however, let me briefly introduce the theoretical framework assumed in this book, the minimalist program. This is the topic of the next section.
1.3 Theoretical framework
The general framework of this book is the minimalist program pioneered by Chomsky (1995), in particular the version of it laid out in Chomsky (2000, 2001) and subsequent works, often referred to as Phase Theory. 2 My goal in this section is not to provide a comprehensive overview of minimalism (or even a general introduction to it), but to give readers less familiar with it sufficient background to follow the rest of the book.3 The minimalist program is couched within the biolinguistic tradition, which takes the language faculty to be a biological organ, a product of evolutionary processes and pressures. The shape of the language faculty is determined by the following three factors, with the third factor gaining more prominence in recent years.
(6) (i) external data;
(ii) genetic endowment (for language, the topic of UG);
(iii) principles of structural architecture and developmental constraints that are not specific to the organ under investigation, and may be organism independent. (Chomsky 2008:133)
At the core of the minimalist program is the so-called Strong Minimalist Thesis (SMT), which states that “language is an optimal solution to interface conditions” (Chomsky 2008:135).4 The interface conditions are
those imposed by the sensorimotor (SM) and conceptual-intentional (C-I) systems. The SMT thus significantly changes the general architecture of the grammar. Readers well versed in Government and Binding theory (and its predecessors) will recognize the Y model of the grammar given in (7a) below, with four distinct levels of representation; D-structure, S-structure, Phonetic Form (PF) and Logical Form (LF). Operations could happen en route to any of these four levels. Likewise, conditions, principles and filters could apply at any level. The “new” minimalist architecture is given in (7b); there are only two relevant levels, the interface levels. Thus, all the syntactic conditions and principles have to be (re-)stated as interface conditions; there is no S-structure or D-structure levels to appeal to.
(7) a. D-structure (X-bar theory, lexical insertion, Theta Criterion)
overt transformations
S-structure (expletive insertion, EPP)
covert transformations
PF
b. Numeration
LF (quantifier scope)
External Merge, Agree, Internal Merge
External Merge, Agree, Internal Merge
External Merge, Agree, Internal Merge
Each derivation starts with a Numeration: a set of lexical items (or features, to be more accurate) to be manipulated in the course of the derivation. Once the Numeration is exhausted, the derivation is complete.
Another crucial innovation in current minimalism is the idea that derivations proceed in chunks called phases and that transfer to the two interfaces can happen more than once per derivation. The terms Phase Theory or Multiple Spell-Out Theory reflect this aspect of the theory.5 The points of transfer to the interfaces are determined by phase heads, which are taken
to be (transitive) v and C heads (perhaps also D heads).6 More specifically, every time one of these phase heads is merged, the complement of the lower phase head is transferred to the interfaces (and becomes inaccessible for further computation). For example, when C is merged, the complement of v is spelled out. This means that the v head itself, its specifiers and adjuncts (if any) remain accessible to the derivation. Otherwise, each derivation would stop with the first Transfer. The condition that ensures this is called the Phase Impenetrability Condition, one formulation of which is given in (8).
(8) The Phase Impenetrability Condition (PIC) (Chomsky 2004:108)
a. PH = [α [ H β]]
b. The domain of H is not accessible to operations, but only the edge of HP.
As mentioned above, the three mechanisms at the center of this book are External Merge, Internal Merge and Labeling. Let me thus conclude this overview with a brief discussion of how they work.
The issue of what kinds of structures Merge can generate is arguably one of the most fundamental issues in syntactic theory. It becomes particularly pressing in the context of recent work by Hauser, Chomsky and Fitch (2002) (see also Fitch, Hauser and Chomsky 2005), who propose that recursive Merge (or some mechanism akin to it) is the only uniquely human property of language, and, as such, it is what separates human language from the communication systems of other species.7
Merge comes in two guises, External Merge and Internal Merge. Simply put, External Merge takes two disjoint syntactic objects and combines them to form one larger syntactic object, as shown in (9a). One of these objects could itself be a result of a previous Merge operation (which is what captures recursion), as shown in (9b).
(9) a. External Merge of α and β
b. External Merge of γ and δ
Internal Merge, often referred to simply as Move, is an operation “responsible” for displacement in the grammar; it captures the intuition that syntactic objects can be pronounced and interpreted in different positions. Internal Merge is like External Merge in that it also takes two objects and combines them into one bigger object. What differentiates it from External Merge is that one of these two objects is a part of the other. (10a–b) below illustrate Internal Merge of α and β; (10a) represents it as a standard Copy and Merge (and Delete) operation, whereas (10b) represents it as literal Internal Merge; the moved element β, instead of being copied into a new position, is simply merged again in its new position.8 The issue of whether the choice between these two ways of conceptualizing movement is substantive or simply a matter of stylistic convention is not trivial but it is not an issue that is directly relevant for our purposes.
(10) a. Internal Merge of α and β
b. Internal Merge of α and β
The version of minimalism assumed here maintains the early minimalist assumption that uninterpretable features play a crucial role in syntactic computation. Uninterpretable features are the features that enter the derivation unvalued (marked in what follows as [uF]) and receive values in the course of the derivation via an operation called Agree. This is a major departure from early minimalism, where feature checking (now conceived of as feature valuation) required movement to a licensing position, typically a specifier of an appropriate functional projection. Now, they can get valued via Agree between a Probe (an item whose feature provides a value) and a Goal (an item whose feature is in need of a value), as shown in (11a–b).
(11) a. ProbeF:val > GoaluF: b. ProbeF:val > GoaluF:val
For Agree to take place, the following conditions have to be met. First, the Goal has to be active, where being active means having an unvalued feature. Second, the Goal has to be in the c-command domain of the Probe. And third, there can be no closer potential Goal. Crucially, in a system with Agree, there is no direct relationship between the need to value unvalued features and movement. Features can be valued in situ and what drives movement is the (generalized) EPP feature (or property) of the Probe, which requires it to have an overt specifier.
The last core concept that we will need is the concept of labels, whose existence has been implicit in the discussion of External Merge and Internal Merge, above. If External Merge were simply concatenation and Internal Merge displacement, we would expect their outputs to be (12a) and (12b), respectively.
(12) a. External Merge of α and β b. Internal Merge of α and β
This, however, is not sufficient; the output of Merge also needs a label. The view of labels that I assume throughout this book is essentially that of Chomsky (1994), where features of one of the two objects undergoing Merge determine the label of the new complex object. The issue of whether Labeling is a separate mechanism (or part of Merge itself) is an interesting one, but it does not bear directly on the issue at stake here, which is the symmetric (or asymmetric) nature of labels. I will thus treat Labeling as distinct from Merge and Move (as argued recently by Hornstein 2009, for example), although nothing substantive hinges on this choice, and my conclusions hold irrespective of whether it is an independent mechanism or not.
The idea that labels are necessary (or desirable) in a minimalist architecture is by no means uncontroversial. Chomsky (2004:109) states: “A still more attractive outcome is that L [language, B.C] requires no labels at all” (see also Collins 2002 and Seely 2006, among others, for arguments that labels might be dispensable in a minimalist system). In Chapter 5 of this book, however, I argue against this view, and provide concrete arguments
1.5 Disclaimer: other sources of symmetry? 9
in favor of the existence of labels. For now, let me proceed on what I consider the standard minimalist assumption, which is that the grammar needs labels, and that the result of External Merge is (13a), not (12a), and the result of Internal Merge is (13b), not (12b).
(13) a. External Merge of α and β b. Internal Merge of α and β
1.4
The proposal
Given that Merge, Move, and Labeling occupy a central spot in minimalist theorizing (for reasons outlined in the previous section), it seems natural to focus on them in our exploration into the locus and amount of symmetry in syntax. The three questions that I ask in the course of the book are:
(14) a. Does symmetric Merge exist?
b. Does symmetric Move exist?
c. Do symmetric labels exist?
The central claim of this book is that the answer to all three questions is “yes.” In the next five chapters, I will provide both theoretical and empirical arguments to support this claim, and, indirectly, against the claims that asymmetric relations are the core relations of the language faculty, articulated in various forms by various researchers (for example, as the Asymmetry Theory of Di Sciullo 2005 or the Antisymmetry Theory of Kayne 1994).
1.5 Disclaimer: other sources of symmetry?
My goal in this book is to argue for symmetry in Merge, Move, and Labeling. Arguably, these are not the first (or most obvious) mechanisms that come to mind when we think of symmetry in syntax. This raises a natural question of why focus on these three, as opposed to the perhaps more apparent symmetric patterns. Part of the reason is purely practical; the issue of symmetry (or the lack thereof) in syntax is vast, and I hope to make it more manageable by narrowing down the domain of inquiry to these three mechanisms. Furthermore, it is not clear that the symmetry we
see elsewhere is true (or only apparent). Let me nevertheless conclude this chapter with a brief discussion of other syntactic phenomena that could be (or have been) classified as symmetric and explain why this book is not about them.
Consider the schematic word order patterns given in (15a–b). If both exist, we have a case for the existence of mirror symmetry in the grammar; (15b) is a mirror image of (15a).9
(15) a. ABCD
b. DCBA
A fairly straightforward illustration of the two orders comes from the domain of adjective ordering, where both patterns are attested crosslinguistically (in addition to many others, not relevant here). For example, the ordering of French adjectives in (16a–b) is a mirror image of the ordering of their English counterparts in (17a–b). The factors that determine the ordering of the adjectives relative to each other are also not directly relevant here (for discussion and concrete proposals, see Bernstein 1993, Cinque 1994, Scott 2002, Sproat and Shih 1991, Svenonius 2008, among many others).
(16) a. une voiture italienne magnifique French a car Italian beautiful
b. une fusée américaine énorme
a rocket American huge (Laenzlinger 2005:658)
(17) a. a beautiful Italian car
b. a huge American rocket
This is not the only possibility; in (18a), the ordering of postnominal adjectives matches the ordering of prenominal adjectives in English, and in (19a), some adjectives precede the noun and others follow it.
(18) a. une voiture rouge française
a car red French (Laenzlinger 2005:658)
b. a red French car
(19) a. un joli gros ballon rouge
a pretty big ball red (Cinque 1994:101)
b. a pretty big red ball
The range of logical possibilities for a sequence consisting of two adjectives and a noun is given in (20a–f). Since there are three elements (two adjectives and a noun), there are 6 (3!) possible ways to order them. (20b) is a mirror image of (20a). The orderings in (20c) and (20d) have been
1.5 Disclaimer: other sources of symmetry? 11
shown by Bernstein (1993) and Cinque (1994), among many others, to be the result of head movement of the noun around Adj2 in (20c) and around both Adj1 and Adj2 in (20d), as shown in in (21b) and (21c) respectively. And the ordering in (20b) has been claimed to be the result of the so-called roll-up movement, schematized in (22a–c). First, the noun moves around Adj1 and, next, the constituent consisting of the noun and Adj1 moves around Adj2.
(20) a. Adj2 Adj1 N (=17a–b)
b. N Adj1 Adj2 (=16a–b)
c. Adj2 N Adj1 (=19a)
d. N Adj2 Adj1
e. Adj1 Adj2 N
f. Adj1 N Adj2
(21) a. Adj2 Adj1 N head movement
b. Adj2 Ni Adj1 ti
c. Ni Adj2 ti Adj1 ti
(22) a. Adj2 Adj1 N roll-up movement
b. Adj2 Ni Adj1 ti
c. [Ni Adj1]j Adj2 tj ti
Given all the movement possibilities, the existence of such symmetric patterns does not seem to be evidence for symmetry inherent to phrase structure rules (or Merge itself). Rather, the illusion of symmetry is created by a sequence of movement operations, involving either head movement or roll-up movement. This brings us to movement itself, more specifically, to the question of what kind of movement would “count” as symmetric movement. We can consider this question from different angles, focusing either on the relationship between the basic and the derived order or the directionality of movement itself. Let us consider them in turn. There are two ways to think of a symmetric relationship between the basic and the derived order; the movement can either preserve the basic order or yield a mirror image of it, as shown in (23a) and (23b) respectively. The two could be thought of as examples of translation and reflection symmetries discussed above.
(23) a. ABCD → ABCD translation symmetry
b. ABCD → DCBA reflection symmetry
If all elements move as individual units, the two illustrate crossing and nesting dependencies, schematized in (24a–b).
(24) a.
The contrasts between the a and b examples in (25) and (26) below, due to Pesetsky (1982), suggest that in a language like English only nesting dependencies are possible. These examples establish this point for whmovement and tough movement, but the preference for nesting has since been generalized to other movement types.
(25) a. What subjecti do you know whoj PRO to talk to tj about ti?
b. * Whoj do you know what subjecti PRO to talk to tj about ti?
(26) a. What balalaikaj are these partitas easy OPi PRO to play ti on tj?
b. * What partitasi is this balalaika easy OPj PRO to play ti on tj?
(Pesetsky 1982:268–9)
Richards (2001), on the other hand, argues that only crossing dependencies are possible if movements target specifier positions of the same functional head. Crucially, neither Pesetsky nor Richards argues that either crossing or nesting is possible, all things being equal.
Another possible way to conceive of symmetric movement involves directionality; example (27a) illustrates movement to the left and (27b) movement to the right.
(27) a.
Constructions that have been claimed to involve rightward movement include (but are not limited to) extraposition, Heavy NP Shift and right node raising, illustrated in (28a–c), respectively.
(28) a. We took a picture ti yesterday [PP of John]i
b. We gave ti to John [DP our most desirable king size suite]i
c. John wrote ti and Mary reviewed ti [DP a new article on right node raising]i
All of them, however, have since been reanalyzed in ways that either do not involve movement at all or involve leftward movement instead. Fox and Nissenbaum (1999), for example, analyze extraposition as countercyclic adjunction of the “extraposed element” to a covertly moved object, providing scope evidence in favor of such an analysis.10 Larson (1988; see also Kayne 1994) analyzes Heavy NP Shift as leftward movement of a larger constituent, consisting of the verb and the indirect object in (28b). In Chapter 3, we will see arguments against treating right node raising as rightward movement and in favor of analyzing it in a multidominant way (as also argued by Bachrach and Katzir 2009, Citko in press-b, Johnson 2007, McCawley 1982, De Vos and Vicente 2005, De Vries 2009, Wilder 1999b, 2008, among others). For now, it seems safe to conclude that the symmetry involving either leftward or rightward movement does not exist in the grammar.
Yet another way to think of movement in a symmetric way is illustrated in (29a–b). (29a) represents raising, and (29b) lowering.11 (29) a. b.
It is typically assumed that only (29a) is possible, as (29b) violates wellestablished conditions on movement, such as the Extension Condition of Chomsky (1995), which requires all Merge and Move operations to target the root of the tree, or the more general No Tampering Condition of Chomsky (2008), which prohibits any modifications to structures that have been already built. Raising transformations are by far the norm in syntactic theory, but there are some cases of lowering transformations that come to mind, such as affix hopping. However, since affix hopping is better analyzed as a PF-merger operation (as argued or implied by Bošković and Lasnik 2003, Bobaljik 1995, Embick and Noyer 2001, Halle and Marantz 1993, among many others), it seems reasonable to conclude that the type of symmetry schematized in (29a–b) does not exist either.12
We might also wonder about more general sources of symmetry in the grammar: parameters whose different settings give rise to symmetric patterns. An obvious candidate for such a parameter is the Head Directionality Parameter. In a very intuitive sense, the existence of consistently head-initial and consistently head-final languages could be viewed as evidence for symmetry in syntax. However, the existence of this type of symmetry has also been questioned in the relevant literature. As pointed
out by Kayne (1994), for example, this type of symmetry breaks down in many directions, and predicts the existence of many language types that we do not find. For example, we do not find mirror-image V2 effects (whose properties would follow from head-final and specifier-final setting of the Directionality Parameter), or wh-movement targeting the final position in the clause (which would follow from the same setting of the parameter).
One could also imagine a symmetry parameter of sorts stating that some languages are inherently (more) symmetric than others. A hypothetical formulation of such a parameter is given in (30). If true, it could be seen as a macro-parameter par excellence, akin to Baker’s (1996) Polysynthesis Parameter, Hale’s (1983) Nonconfigurationality Parameter or Huang’s (2005) Analyticity Parameter. The existence of such macro-parameters, however, is not an uncontroversial matter (see Kayne 2005 and Baker 2008 for two opposing views on the matter). (30) Symmetry Parameter
A language is either symmetric or asymmetric.
I am not defending in this book anything even close to such a parameter, nor am I aware of anyone else trying to. Its simplicity is appealing but its consequences would be rather dire. First and foremost, it is not clear what it would mean for a language to be either symmetric or asymmetric. Perhaps symmetric languages have overwhelmingly symmetric structures, as implied by Hale’s (1983) Nonconfigurationality Parameter, which posits a flat structure (in which arguments c-command each other) for nonconfigurational languages like Warlpiri. Again, it has since been shown that there is nothing unusual about such languages, and their nonconfigurational properties follow from the interaction of perfectly configurational UG principles and parameters (see Legate 2001 and 2002 for discussion). More generally, we might wonder about the status of macro-parameters in minimalist syntax. Since the Symmetry Parameter cannot be stated as a property of individual lexical items (or a class of lexical items), it is not compatible with the view going back to Borer (1984) that the locus of crosslinguistic variation is in the lexicon, more specifically, in the features of functional heads.
1.6 Organization of the book
In this final section, I offer a brief outline of the rest of the book. Most of the novel data in the book come from Polish, a West Slavic language that provides a nice illustration of many of the phenomena I discuss. By focus-
ing on Polish, I hope to make the book of interest not only to theoretically minded linguists interested in the workings of Merge, Move and Labeling, but also to more empirically minded linguists interested in the syntax of Slavic languages. I will, however, add a comparative perspective where it is useful and necessary.
Chapter 2 “Asymmetry in syntax” takes a step back, and reviews both empirical and theoretical reasons behind the claims that syntax only allows asymmetric relations. It examines the evidence that Merge can only create structures that involve asymmetric c-command between nodes (since only such structures are linearizable, according to the Linear Correspondence Axiom of Kayne 1994), that Move is asymmetric (in that it always targets the higher of the two potentially movable elements), and that labels are asymmetric (in that they contain features of only one element). This chapter serves mostly as background for what is to come in the following three chapters, which show that Merge can also create symmetric structures, that Move can sometimes treat two constituents in a symmetric fashion, and that labels can include features of two elements.
Chapter 3 “Symmetric Merge” provides both theoretical and empirical arguments in favor of a particular kind of symmetric structure, the multidominant structure of the kind given in (31), in which one node has two mothers.two mothers.
(31)
The existence of multidominant structures is by no means new in generative grammar (see Blevins 1990, Goodall 1987, McCawley 1988, Moltmann 1992a, Muadz 1991, for some pre-minimalist precursors and Citko 2000, 2003, 2005, Gračanin-Yüksek 2007, in press, Van Riemsdijk 2000, 2006a, 2006b and De Vries 2005a, 2005b, 2008, 2009 for more recent minimalist implementations). In my early work, I argued that adopting a symmetric structure of this kind can account for a number of otherwise mysterious properties of across-the-board wh-questions, questions such as the one given in (32), in which a single wh-element is extracted simultaneously from two (or more) conjuncts.
(32) What did John write and Bill review?
This chapter extends this analysis to a number of other constructions, thus providing new empirical evidence in favor of symmetric Merge structures. The constructions I discuss in this chapter are: questions with conjoined
wh-pronouns, gapping, right node raising, standard and transparent free relatives, serial verbs, and they are illustrated in (33a–f).
(33) a. What and why did John eat? wh&wh questions
b. John writes poems and Mary short stories. gapping
c. John wrote and Mary reviewed an article on multidominance. right node raising
d. John reads whatever Mary writes. standard free relatives
e. John wrote what some might call a poem. transparent free relatives
f. ò dà sε la nεnè ɔ ` ɔ ` .
3sg past roast fem meat eat
‘He roasted the meat and ate it.’
Dàgááré serial verbs
(Hiraiwa and Bodomo 2008:243)
Chapter 4 “Symmetric Move” turns to the evidence for symmetric movement, i.e. a type of movement that can target two elements in a given structure with equally grammatical results. Passive movement in double object constructions provides a good testing ground; since there are two objects, there are two candidates for movement. In some languages, passive movement is a symmetric operation in that it can target either the direct or the indirect object, whereas in others it is an asymmetric operation in that it can target only one of them. This variation has been studied quite a bit and attributed either to various ways of circumventing a locality violation (such as the availability of the word order in which the direct object precedes the indirect one, either via base-generation or movement) or to case properties of the two objects (and various patterns of case absorption in passives). What I argue for in this chapter is that we need both types of account to reach empirical adequacy.
Chapter 5 “Symmetric labels” gives arguments in favor of the existence of symmetric labels in the grammar. It also establishes a need for labels in the grammar. There are two ways a label can be symmetric – if both merged elements project as the label, or if neither of them does (as argued by Collins 2002, Seely 2006, among others). This chapter focuses on the former case – symmetric labels in which both constituents project – and argues that this is what is involved in the derivation of certain types of serial verb constructions, nominal copular constructions and comparative correlatives.
By the end of Chapter 5 I hope to convince the reader that symmetry does exist in the domain of the three basic syntactic mechanisms: Merge, Move, and Labeling. Chapter 6 “Conclusion” offers a brief summary and some general thoughts on the nature and origin of symmetry (and asymmetry) in the grammar.
Another random document with no related content on Scribd:
The
Project Gutenberg eBook of Jusqu'à la fin du monde
This ebook is for the use of anyone anywhere in the United States and most other parts of the world at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this ebook or online at www.gutenberg.org. If you are not located in the United States, you will have to check the laws of the country where you are located before using this eBook.
Title: Jusqu'à la fin du monde
Author: Adolphe Retté
Release date: March 13, 2024 [eBook #73164]
Language: French
Original publication: Paris: Albert Messein, 1926
Credits: Laurent Vogel (This file was produced from images generously made available by the Bibliothèque nationale de France (BnF/Gallica)) *** START OF THE PROJECT GUTENBERG EBOOK JUSQU'À LA FIN DU MONDE ***
ADOLPHE RETTÉ
Jusqu’à la fin du Monde
PARIS ALBERT MESSEIN, ÉDITEUR
19, QUAI SAINT-MICHEL, 19
1926
Librairie A. MESSEIN, 19, Quai Saint-Michel, Paris
DU MÊME AUTEUR
POÉSIES (1897-1906) : Campagnepremière, Lumières tranquilles, Poèmes de la forêt (Messein)
Une belle damepassa (Messein. 1 vol. in-12)
Le Symbolisme (anecdotes et souvenirs). 1903 (Messein. 1 vol. in-12)
ŒUVRES CATHOLIQUES
Du diable à Dieu, récit d’une conversion (Messein).
Le règne de la Bête, roman (Messein).
Un séjour à Lourdes, journal d’un pélerinage à pied, impressions d’un brancardier (Messein).
Sous l’étoile du Matin, la première étape après la conversion (Messein).
Dans la lumière d’Ars, récit d’un pèlerinage (Tolra).
Au pays des lys noirs, souvenirs de jeunesse et d’âge mûr (Téqui).
7 fr.
7 fr.
9 fr.
Quand l’esprit souffle, récits de conversions : Huysmans, Verlaine, Claudel, etc. (Messein).
Ceux qui saignent, notes de guerre (Bloud et Gay).
Sainte Marguerite-Marie, vie de la Révélatrice du Sacré-Cœur, d’après les documents originaux (Bloud et Gay).
Lettres à un indifférent, apologétique réaliste (Bloud et Gay).
Le Soleil intérieur, Saint Joseph de Cupertino, Catherine de Cardonne, une Carmélite sous la Terreur, la Charité du malade (Bloud et Gay).
Louise Ripas, une privilégiée de la Sainte-Vierge, préface de Mgr Landrieux, évêque de Dijon (Bloud et Gay).
Léon Bloy, essai de critique équitable (Bloud et Gay).
La Maison en ordre, autobiographie (Messein).
Les Rubis du Calice, méditations et oraisons sur des textes de la Messe (Messein).
La basse-cour d’Apollon, mœurs littéraires (Messein).
IL A ÉTÉ
TIRÉ DE
CE
LIVRE
: 5exemplairessurVergéd’Arches numérotésde1à5. No
In corde Christi
Benefactoribus
