400 likes | 417 Views
Explore Chomsky's conditions of descriptive and explanatory adequacy in syntactic variation within constraint-based theories. Discover the tension between complexity and invariance in language structure, the distinction between E-language and I-language, and how communities impact linguistic norms. Gain insight into Chomsky's view on probability in grammar and why variation is considered peripheral in his theories. Delve into alternative theories of grammar such as LFG, Categorial Grammar, Construction Grammar, and HPSG, focusing on their key properties and the central construct of HPSG - the Sign. Explore how HPSG allows uniform representational format intertwining phonology, syntax, semantics, pragmatics, and social factors.
E N D
Syntactic Variation in Constraint-based Theories Tom Wasow Stanford University
Chomsky’s Conditions of Adequacy “The grammar of a particular language satisfies the condition of descriptive adequacy insofar as it gives a full and accurate account of the properties of the language, of what the speaker of the language knows. To satisfy the condition of explanatory adequacy, a theory of language must show how each particular language can be derived from a uniform initial state under the ‘boundary conditions’ set by experience. In this way it provides an explanation of the properties of languages at a deeper level.” - Noam Chomsky, New Horizons in the Study of Language (2000), p. 7. (emphasis added)
Chomsky’s “serious tension” “The search for descriptive adequacy seems to lead to ever greater complexity and variety of rule systems, while the search for explanatory adequacy requires that language structure must be invariant, except at the margins.” - Noam Chomsky, New Horizons in the Study of Language (2000), p. 7 (emphasis added)
E-language vs. I-language • The distinction • E-language: any conception of language as something external to the speaker. • I-language: language identified as the internalized knowledge of an individual speaker. • Chomsky on E-language: “The concept of E-language has no place…in the theory of language” -Knowledge of Language (1986), p. 26
Chomsky on Communities “Communities are formed in all sorts of overlapping ways, and the study of communities and their norms quickly degenerates into the study of everything….I can understand Jones, within limits, because my I-language is not too different from his, and because he and I share other unknown properties…; this is not a topic of empirical inquiry as it stands.” - Noam Chomsky, New Horizons in the Study of Language (2000), p. 72-73
Chomsky on probability “It seems clear that probabilistic considerations have nothing to do with grammar, e.g., surely it is not a matter of concern for the grammar of English that ‘New York’ is more probable than ‘Nevada’ in the context ‘I come from ---.’ In general, the importance of probabilistic considerations seems to me to have been highly overrated in recent discussions of linguistic theory.” - A transformational approach to syntax (1962), p. 128
Why variation is peripheral in Chomsky’s theories Chomsky claims: • The big questions are about universals of language. • Language is taken to be a property of an individual’s psychology. • Communities are ill-defined and can’t be dealt with scientifically. • Interesting generalizations are categorical.
Alternative theories of grammar • Lexical Functional Grammar (LFG) • various forms of Categorial Grammar • various forms of Construction Grammar • Head-driven Phrase Structure Grammar (HPSG)
Key properties of these theories • Only one phrase structure associated with each (unambiguous) sentence: no transformations • Lexicalist: information-rich lexicons • Constraint-based: grammars consist of sets of constraints that must be simultaneously satisfied by well-formed structures. • Explicitly formulated: descriptive adequacy a realistic goal.
Additional useful property of HPSG Uniform representational format allows any types of information relevant to linguistic form to be related to one another: • phonology • syntax • semantics • pragmatics • context of use • social status of interlocutors • other social factors (gender, ethnicity, age, etc.)
Central construct of HPSG: the Sign • Signs are complexes of phonological, syntactic, semantic, and contextual information. • Information is represented by feature-value pairs. • Values of features may themselves be feature structures or lists of feature structures. • Feature structures are assigned to types, and the types form hierarches, in which lower types inherit features and values from higher types.
Simplified Examples • Taken from Sag, Wasow, & Bender, Syntactic Theory: A Formal Introduction, 2nd edition (CSLI Publications, 2003) • For more thorough treatment, see Emily Bender’s (2000) Stanford dissertation, Syntactic Variation and Linguistic Competence: The Case of AAVE Copula Absence.
NICE Properties of the English Auxiliary System • Negation: Sentential not follows a finite auxiliary verb. • Inversion: The finite auxiliary precedes the subject in yes-no questions (inter alia) • Contraction: Many finite auxiliaries have negative forms ending in -n’t • Ellipsis: The complement of an auxiliary verb may be absent, if context makes the interpretation clear.
How Ellipsis is Different • The auxiliary verb need not be finite. • No idiosyncratic exceptions • Negation: refusal sense of will not; different scope patterns for not with different modals • Inversion: 1st singular aren’t; some speakers only use shall in inverted constructions • Contraction: won’t, mustn’t, *amn’t, etc.
Plug for a recent HPSG analysis… Ivan Sag’s ‘Rules & Exceptions in the English Auxiliary System’ (to appear in Journal of Linguistics) posits lexical rules for negation, inversion, and contraction, but a constructional analysis of ellipsis.
Variation in the English Auxiliary, Example 1 • Main verb have can show auxiliary properties in some cases: Have you any idea what time it is? I haven’t a clue what you’re talking about. • In other cases, it’s impossible: *Had Mary her baby yet? *I haven’t allergies. • The alternative with do is always possible: Do you have any idea what time it is? I don’t have a clue what you’re talking about. Did Mary have her baby yet? I don’t have allergies.
Judgments Vary About Main Verb have in Auxiliary Contexts • Americans seem to allow auxiliary uses of mv have less than speakers of other varieties. • There is considerable individual variation, as well, both across speakers and across uses. • Emily Bender polled 10 speakers (5 UK, 2 Australians, 1 Indian, and 1 New Zealander) about auxiliary behavior of mv have in 4 auxiliary contexts. • Generalization that emerged: If a speaker accepted a use of mv have in any auxiliary context, s/he allowed it in ellipsis.
An HPSG Analysis • Many idiosyncratic meanings of have will require separate lexical entries, to get the semantics right. • Some speakers will list the entry with the core possession meaning as [AUX +] or ([AUX +]).
Sag’s Analysis of the Auxiliary System Helps Explain This • Lexical rules have lexical exceptions. • Constructions (phrase structure rules) don’t. • So when a speaker treats a reading of mv have as [AUX +], it will license ellipsis, but could still be an exception to one of the lexical auxiliary rules.
Variation in the English Auxiliary, Example 2 • Certain finite forms of be can be omitted in AAVE, e.g., You lookin good. • Factors conditioning omission are well studied. • The standard analysis (Labov) is phonological: contracted forms of is and are can be deleted
Problems with the Phonological Analysis • Doesn’t generalize to other languages (e.g., Russian, Hungarian) with zero copulas, but no analogous copula contraction. • Not all zero copulas are present tense: “I dug being militant cause I was good. It was something I could do. Rap to people. Whip a righteous messsage on em. People knew my name. They’d listen. And I’d steady take care of business. This was when Rap Brown and Stokeley and Bobby Seale and them on TV. I identified with them cats… [emphasis added] - John Edgar Widemann, Brothers & Keepers (1984), p. 114.
More Problems with the Phonological Analysis(due to Emily Bender) • It entails (in Labov’s words): Where other English dialects do not permit contraction of the auxiliary, AAVE does not permit deletion. • Counterexamples: • How old you think his baby? (AAVE) • *How old (do) you think his baby’s? • Tha’s the man they say in love. (AAVE). • *That’s the man they say’s in love. (with full contraction)
Two Possible Syntactic Analyses: • Constructional: posit a construction licensing a finite S consisting of an NP followed by a predicative phrase of any type. • Silent verb: posit a finite variant of be that has a null phonology.
Evaluation of the Constructional Analysis • Strengths • Its WYSIWYG character is appealing • If [FORM fin] is dropped, it would license invariant be and several other attested AAVE forms. • Weaknesses • The generalization to the other forms would incorrectly predict that they are incompatible with 1st singular subjects. • The killer: it doesn’t cover • How old you think his baby? • Tha’s the man they say in love.
But this isn’t general enough… • Existential be takes two complements, and it can be silent, too: There a car blockin my way. There a book gone from my desk. • Arguably, the equative be takes a [PRED ] complement, and it can be silent, too: Ain’t Lew Alcindor Mohammed Ali? No, Cassius Clay Mohammed Ali.
Aside: Paolillo’s Analysis of the Sinhala Zero Copula • Note the BACKGROUND feature • Source: Paolillo, John C. (2000) “Formalizing Formality: An Analysis of Register Variation in Sinhala” Journal of Linguistics 36: 215-259.
Summary: What HPSG Can Offers Variationists • Precise, well-defined descriptive tools • Detailed lexical representations • Uniform representational framework • Facilitates interactions among information types • Social information straightforwardly assimilated • Probabilities could easily be added (not yet done) • Constraint-based architecture: order-independence • Minimal use of abstract structures • No transformations • Fairly direct relationship between observable forms and linguistic analysis