SICOGG-9 Workshop on Syntax-Phonology Interface Kwangwoon University, Seoul Aug 8th-11th, 2007

Phonology in Syntax Tommi Leung University of Southern California Main claims: -

Some conceptual considerations of the theory of Merge (Chomsky 1995, Collins 2002) leads to an alternative conjecture of the syntax-phonology interface. The design features of the phonological component (PF) are highly relevant to the basic architecture of Narrow Syntax (NS). The syntax-phonology asymmetry does not exist as far as computation is concerned. The algorithm of NS derives both LF- and PF- representations simultaneously. SYNTAX is an algorithm that conceptualizes a generalized combinatorial system, whether it generates phrase markers (LF) or phonetic strings (PF) at the interface levels.

1. Two Competing Theories of Merge 1.1. Bare Phrase Structure (Chomsky 1995 et seq.) (i) Merge α, β → {γ, {α, β}}, where γ is the label of constituent projected by either α or β as the head. (1) (a)

α ty α β

(b)

the ty the boy

(ii) The concept of labels and heads is adopted in various guises in almost all phrase structure grammars (Chomsky 1955/75, 1965, 1981, 1982, Jackendoff 1977). - rewrite rules (e.g. NP → Det N, VP → V NP, etc) - X’-theory (e.g. XP → Spec X’, X’ → X Comp) (iii) The notion of labels/heads generalizes across syntactic categories and economizes the theory of grammar in general (i.e. methodological minimalism). 1.2. Eliminating Labels 1.2.1. Problems of Merge (i) Labels go beyond the basic considerations of syntax as a combinatorial system (e.g. set theory, abstract algebra, etc). (ii) Labels are not terms for computation and they violate the Inclusiveness Condition. (2) Terms (Chomsky 1995:243, Chametzky 2000:122): For a syntactic object K, a. K is a term of K. b. If L is a term of K, then the members of the members of L are terms of K. c. Nothing else is a term.

1

(3) Inclusiveness Condition (Chomsky 1995:225): “[O]utputs consist of nothing beyond properties of items of the lexicon (lexical features)  in other words, that the interface levels consist of nothing more than arrangements of lexical features.” (iii)The particular notion of labels lacks a principled explanation  No compatible organic (e.g. visual system) or non-organic (e.g. mathematical) combinatorial systems incorporate the notion of labels or something similar. “An even more basic question from the biological point of view [concerning the language faculty] is how much of language can be given a principled explanation, whether or not homologous elements can be found in other domains or organisms.” (Chomsky 2005b: 2) (iv) The recursive system defined by Merge is conceptually independent of the postulation of labels.

(4) None of the following concerning the design features of Merge requires labels: “A preferable result is that the label is predictable by a general rule. A still more attractive outcome is that L requires no labels at all.” (Chomsky 2004:109, emphasis added) “[T]he most elementary property of language – and an unusual one in the biological world – is that it is a system of discrete infinity consisting of hierarchically organized objects.” (Chomsky 2005a:4, emphasis added) “A natural requirement for efficient computation is a “no-tampering condition” NTC: Merge of X and Y leaves the two SOs unchanged. If so, then Merge of X and Y can be taken to yield the set {X, Y}, the simplest possibility worth considering.” (Chomsky 2005a:5) (v) The function of labels that ensure successive derivation can be summarized by the following axiom that is well instantiated in other mathematical systems (to be entertained seriously): (5) Closure operator: -

A set is said to be closed under some (mathematical) operation if the operation on members of a set produces another member of that set.

-

E.g. For any natural number X and Y, and the operator +, X+Y is also a natural number. Thus the set of natural number is closed under addition.

-

Likewise for the closure of real numbers and complex numbers under addition. (n.b.: natural numbers are NOT closed under the subtraction, e.g. 2-3 = -1)

-

In syntax, Merge can be defined as a binary union operator { __ , __ } with the following axiom:

(6) The syntactic object X and Y are closed under the Merge operator { __ , __ }, i.e. {X, Y} is also a syntactic object. → the generative capacity is theoretically independent of labels. 1.2.2. Label-Free Merge (Collins 2002) (i) Given the ad-hoc properties of Merge, Collins (2002) postulated an alternative version of Merge that totally dispenses with labels. Call this thesis Eliminating Labels (EL). 2

(ii) For α, β as syntactic objects, a. Merge α, β → {α, β}

a’.

ty α β

b. Merge γ, {α, β} → {γ, {α, β}}

b’.

ty γ ty α β

(iii) In the absence of labels, syntactic derivation cannot make reference to the X’- and XP level: (7) “No operation or condition may be defined that makes reference to VP versus NP” (iv) All operations and conditions are stated at the level of lexical items only. 1.2.3. Derivation of Syntactic Relations (i) Syntactic relations include the following four major types: (8)

Theta (X, Y): X assigns a theta-role to Y. EPP (X, Y): Y satisfies the EPP feature of X. Agree (X, Y): X matches Y, and Y values X. Subcat (X, Y): X subcategorizes for Y. Call them ‘selector features’. (cf. Probe-Goal approach in Chomsky 1999, 2001)

(ii) Merge → the ‘selector’ feature of one element is satisfied. (9) a. Merge (there, T) = {there, T} b. Merge (drink, {the, wine}) = {drink, {the, wine}}

Selector feature: (EPP (X, Y)) Selector feature: (Subcat (X, Y))

(iii)Derivation is ongoing depending on the ‘saturation’ of lexical items, i.e. an item that bears one/more selector feature(s) need to be saturated before new lexical items are introduced: (10) An example of good derivation i. Select the and man ii. Merge (the, man) = {the, man} iii. Select see iv. Merge (see, {the, man}) = {see, {the, man}} (11) An example of bad derivation i. Select the and see ii. Merge (see, the) = {see, the} iii. Select man iv. Merge ({see, the}, man)={{see, the}, man}

(Subcat (the, man)) (Theta (see, {the, man}))

(*Subcat (see, the)) (*Theta ({see, the}, man))

(iv) Minimality Condition (Rizzi 1990, Chomsky 1995, 2001): (12) “Let X (a lexical item) contain a probe P. Let G be a matching goal. Then Agree (P, G) is established only if there is no feature F (a probe or a goal) matching P, such that P asymmetrically c-commands F and F asymmetrically c-commands G.”

3

(13)

ty tell ty on Z

(Probe: tell; Goal: on (not PP))

(14) a. It is important that someone says that John is nice. (Probe: that; Goal: finite T) b. *It is important that to say John is nice. (v) Generating the X-bar schema: i. Certain configurations are banned: [X’ X YP], [XP ZP X’], *[X’ X Y’], *[XP Z’ X’] ii. A saturated selector (i.e. YP) merges with X that bears a selector feature and satisfies Subcat (X, Y), i.e. [X’ X YP] iii. The selection of ZP satisfies another selector feature of X, EPP (X, Y), i.e. [XP ZP X’] iv. Y’ and Z’ are unsaturated constituents that cannot merge with another selector. Derivation cannot proceed further, i.e. *[X’ X Y’], *[XP Z’ X’]. 2. Problems of Eliminating Labels 2.1. First Problem: Minimality is Undefined (i) Minimality is defined by antisymmetric c-command that is stated in terms of categories, domination and exclusion (e.g. Kayne 1994). The notion of phrasal categories (i.e. labels) are tacitly involved: (15) X asymmetrically c-commands Y iff X c-commands Y and Y does not c-command X. (16) X c-commands Y iff X and Y are categories and X excludes Y and every category that dominates X dominates Y. (17)

ty A ty B C

(ii) Therefore, in a label-free theory of Merge, Minimality is ill-defined. (iii) What about a derivational approach to syntactic relation that defines c-command (e.g. Kitahara (1997), Epstein et al. (1998) and Epstein (1999))? (18) “X c-commands all and only the terms of the category Y with which X was paired by Merge or by Move in the course of the derivation.” (19) For α and β as syntactic objects, Merge α and β yields: i. the composite set {α1, {α2, β}}, whereas α1 functions as the label; ii. the sisterhood relation between α2 and β, and; iii. the motherhood relation between α1, and α2 and β. (iv) However, a derivational theory of syntactic relation needs to resort to the notion of ‘motherhood’ that is considered as insignificant in EL. 2.2. Second Problem: Chains are Undefined (i) Chains (CHs) as a list of occurrences (OCCs) (Chomsky 1955/75, 1981, 1982, 1995) 4

(ii) An occurrence of α in context K is the full context of α in K. (iii) OCCs are defined by sisterhood. (20) Johni seems ti to ti thrive. CH (John) = (Tseems, to, thrive)

(A-movement)

(21) Whoi do you think ti that Mary likes ti? (A’-movement) CH (who) = (Cdo, that, likes) (iv) In EL, OCCs cannot be defined in all positions since the sister of the head of a CH is always an X’level: ty β ….. ty α tβ (v) Syntactic chains as defined by the list of occurrences in phrase structure grammars are undefined under EL. 3. The Associative Nature of Syntactic Computation Puzzle 1: How to maintain the concept of EL while defining (i) constituent asymmetry; (ii) chains? Puzzle 2: According to EL in which the composite set {γ, {α, β}} is formed without labels, how to define a formal relation between γ and α/β? Immediate suggested solutions: (i) Some common properties of combinatorial systems can be incorporated into NS, thus yielding a principled explanation. (ii) Typical examples of combinatorial systems: Sets, abstract algebras, etc. (iii) Law of Associativity of set union and mathematical addition: - For any set A, B and C, A ∪ (B ∪ C) = (A ∪ B) ∪ C. (set union) - For all a, b ∈ S, a + (b + c) = (a + b) + c. (algebraic operation) (iv) However: Law of associativity is NOT a design feature of LF!! The design features of LF (Prinzhorn et al. 2004, Leung 2007, Vergnaud 2007): (22) a. The algebraic operation defined at LF is commutative and non-associative. b. For any two phrase markers α, β, the concatenation operation that applies to α and β is commutative if and only if the constituent [α β] = [β α]. c. For phrase markers α, β, γ, the concatenation operation that applies to α, β and γ is non-associative if and only if [γ [α β]] ≠ [[γ α] β]. (v) If LF does not allow the associative law, what else in the language faculty allows this?? (vi) The design features of PF: linear ordering!!! 5

(23) a. The algebraic operation defined at PF is non-commutative and associative. b. For any two phonetic strings α and β, the concatenation operation that applies to α and β is non-commutative if and only if (α β) ≠ (β α). c. For phonetic strings α, β and γ, the concatenation operation that applies to α, β and γ is associative if and only if (γ (α β)) = ((γ α) β). (24) A linear order on a set X is any binary relation on X that is antisymmetric, transitive, and total (also Kayne 1994): i. if a ≤ b and b ≤ a then a = b ii. if a ≤ b and b ≤ c then a ≤ c iii. a ≤ b or b ≤ a

(antisymmetry) (transitivity) (total)

(vii) Among all linear relations we define the generalized notion of ‘string adjacency’ (SA): (25) a. The SA between two strings a and b is notated by ‘a/b’. b. SA is symmetric, i.e. a/b if and only if b/a. c. At PF (i.e. along the timing axis), any string is adjacent to at least one string (e.g. peripheral strings) or at most two strings (e.g. non-peripheral strings) (Vergnaud 2007). # ••••••••• # d. SA is not transitive, i.e. if a/b and b/c then it is not the case that a/c. (viii) Under the development of EL, SA as a design feature of PF forms part of syntactic representation. 4. Redefining Chains by String Adjacency (26) ty β γ ….. ty α tβ (27) a. The syntactic relation is established between α and tβ via SA, i.e. α/tβ. b. The syntactic relation is established between β and γ via SA, i.e. β/γ. (28) a. SA dispenses with labels; b. SA adequately defines a chain. Assume that SA can define syntactic relations and form part of syntactic representation, i.e. (viii): (29) SA can define a syntactic constituent.

5. Evidence for (29) I: Incremental Derivation (i) String-adjacent elements can behave as conjuncts and pass the coordination test. (30)a. [John likes] but [Mary hates] Peter. b. I like [John’s] but not [Peter’s] work. c. Wallace gave [Gromit a biscuit] and [Shawn some cheese] for breakfast. (ii) Incremental derivation (Phillips 2002): Constituenthood is dynamic depending on the derivational stage. 6

(31)

X ty A B

+C →

X ty A Y ty B C

(32)a. b. c. d.

[Wallace] and [Wendolene] will give Gromit crackers before breakfast. [Wallace will] and [Wendolene probably won’t] give Gromit crackers before breakfast. [Wallace will give] and [Wendolene will send] some crackers to Gromit for his birthday. [Wallace will give Gromit] and [Wendolene will give Preston] a shining new collar for walking about town. e. [Wallace will give Gromit crackers] and [Wendolene will give Preston dog food] before breakfast. f. Wallace will [give Gromit crackers before] and [throw him crackers during] the final day of the thrilling cricket match.

6. Evidence for (29) II: Contraction (i) SA can feed phonological contraction. (33) a. They’ll (< will) leave soon. (O’Grady 2005) b. He’s (< is) there. c. They hafta (< have to) go. d. Who do they wanna (< want to) see? e. Who do you think’s (< is) the winner? (ii) Contraction via SA is pervasive depending on the derivational stage (34) a. They’ll (
* no

* mi

* * nee

Line 1 Line 0

(ii) Analogy between syntactic structure and metrical structure: (Chomsky and Halle 1968, Halle and Vergnaud 1987, Vergnaud 2003, 2007) -

The metrical constituent defined on line 0 is ‘labeled’ at the third syllable on line 1. This asterisk corresponds to the ‘projection’ of a ‘head’ represented by nee.

(iii) The expression of occurrences and CHs in metrical constituents: - The two asterisks above the third syllable nee in (35) share all attributes (e.g. feature specification) but their height (Halle 1982 MIT lecture, Vergnaud 2003, p.618) - The asterisks in the same column within the grid are de facto identical objects that are distinguished by different contexts, in this case the metrical tiers 7

(iv) The instantiation of locality in metrical constituents: - an asterisk on line n (e.g. in (35))→ an asterisk on line n-1 within the same column. (minimal link) (36) Halle’s Conjecture: “Given a metrical grid M and some position i in M, the asterisks in the column above i form a chain.” 8. Phonology in Syntax (i) The computational algorithms of NS are intrusive to the derivation of the LF- and PF-representation simultaneously. (ii) Some abstract algorithms of NS are equally instantiated at LF and PF, in different guises. (iii) Call these common algorithms ‘SYNTAX’ (read as ‘big syntax’) that serve as the cover term for such a combinatorial system. (iv) In a narrow sense, one can define a ‘SYNTAX of syntax’ (SYNS) and a ‘SYNTAX of phonology’ (SYNP). (37) La Spéculation de Vergnaud (2007): “There must be a component of phonology which is more abstract than the mere processing of speech gestures would require, since language can also be manifested as sign language. In that sense, there must exist a [SYNTAX] of phonology.” (v) SYNTAX is an algorithm that conceptualizes a generalized combinatorial system of grammar, whether it derives phrase markers (LF) or phonetic strings (PF) observable at the interface levels. (vi) SYNTAX derives syntactic relations between elements; syntactic relations can be restated by the ‘inthe-context-of’ relation (Prinzhorn et al. 2004, Leung 2007, Vergnaud 2007): (38) Defining syntactic and contextual relations: (X, [  Y]) ≈ X ↔ [  Y] ≈ [X Y] (vii) The ‘in-the-context-of’ relation can be defined syntactically (via sisterhood) and phonologically (via SA). (viii) The syntax-phonology asymmetry (Chomsky 1995) does not exist as far as computation is concerned. 9. Conclusions (i) The algorithm of Narrow Syntax derives LF and PF representation simultaneously (cf. Bobaljik 1995, 2002, Brody 1995). (ii) PF is not necessarily a mapping of syntactic representation or a result of some ad-hoc mechanism (e.g. Spell-out). Therefore the syntax-phonology asymmetry does not exist with respect to computation. (iii) String adjacency as a design feature of PF can define the basic architecture of Narrow Syntax. (iv) Some syntactic notions (e.g. chains) are readily found at the phonological component. (v) Bare Output Condition is upheld – The design features of Narrow Syntax ensure the derivation of both LF- and PF-interpretable objects at the interface (Epstein and Seely 2006, Leung 2007). 8

(vi) Narrow Syntax is merely a combinatorial system that should receive a principled explanation, whereas both the conceptual-intentional and sensorimotor interfaces impose external conditions.

Acknowledgements This paper serves as a sequel to Jean-Roger Vergnaud’s The Syntax of Phonology (2007). I would like to thank Chris Collins, Roumi Pancheva and Jean-Roger Vergnaud for their general comments to the early version of this paper. All errors remain my own.

References Bobaljik, Jonathan David. 1995. Morphosyntax: The Syntax of Verbal Inflection. Doctoral dissertation, MIT, Cambridge, Mass. Bobaljik, Jonathan David. 2002. A-chains at the PF-interface: Copies and ‘covert’ movement. Natural Language and Linguistic Theory 20: 197-267. Brody, Michael. 1995. Lexico-Logical Form: A Radically Minimalist Theory. Cambridge, Mass: MIT Press. Chametzky, Robert A. 2000. Phrase Structure. Oxford: Blackwell. Chomsky, Noam. 1955/75. The Logical Structure of Linguistic Theory. New York, Plenum. Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris. Chomsky, Noam. 1982. Some Concepts and Consequences of the Theory of Government and Binding. Camb, Mass: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Camb, Mass: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: a Life in Language, ed., Michael Kenstowicz, MIT Press. Chomsky, Noam. 2004. Beyond explanatory adequacy. In Structures and Beyond, ed. Adriana Belletti, 104131. Oxford: Oxford University Press. Chomsky, Noam. 2005a. On Phrases. Ms. MIT. Chomsky, Noam. 2005b. Three factors in language design. Linguistic Inquiry 36. 1: 1-22. Chomsky, Noam and Morris Halle. 1968. The Sound Pattern of English. Cambridge, Mass: MIT Press. Collins, Chris. 2002. Eliminating labels. In Sam D. Epstein and T. Daniel.Seely (eds.), Derivation and Explanation in the Minimalist Program. Blackwell. Epstein, Sam. D., Erich. M. Groat., Ruriko Kawashima, and Hisatsugu Kitahara. 1998. A Derivational Approach to Syntactic Relations. New York, Oxford: Oxford University Press. Epstein, Sam. D 1999. Un-principled syntax: the derivation of syntactic relations. In Working Minimalism, ed. Sam. D. Epstein and Norbert Hornstein, 317-345. Cambridge, Mass: MIT Press. Epstein, Sam D and T. Daniel Seely. 2006. Derivations in Minimalism. Cambridge: Cambridge University Press. Halle, Morris and Jean-Roger Vergnaud. 1987. An Essay on Stress. Cambridge, Mass: MIT Press. Jackendoff, Ray S. 1977. X-bar Syntax. Cambridge, Mass: MIT Press. Kayne, Richard. 1994. The Antisymmetry of Syntax. Camb, Mass: MIT Press. Kitahara, Hisatsugu. 1997. Elementary Operations and Optimal Derivations. Cambridge, Mass: MIT Press. Leung, Tsz-Cheung. 2007. Syntactic Derivation and the Theory of Matching Contextual Features. Doctoral dissertation, University of Southern California. O’Grady, William. 2005. Syntactic Carpentry. Mahwah, New Jersey: Lawrence Erlbaum Associates. Phillips, Collin. 2003. Linear Order and Constituency. Linguistic Inquiry 34.1:37–90 Prinzhorn, Martin, Jean-Roger Vergnaud, and Maria Luisa Zubizarreta (2004). Some explanatory avatars of conceptual necessity: elements of UG. Ms., University of Southern California, Los Angeles. Rizzi, Luigi. 1990. Relativized Minimality. Cambridge, Mass: MIT Press. Vergnaud, Jean-Roger. 2003. On a certain notion of “occurrence”: the source of metrical structure, and of 9

much more. In Living on the Edge, ed. Stefan Ploch, 599-632. Berlin: Mouton de Gruyter. Vergnaud, Jean-Roger. 2007. Syntax of phonology, CUNY Conference on Precedence Relations.

Email: [email protected]

10