A u t h o r ' s p e r s o n a l c o p y
The simpler syntax hypothesis
Peter W. Culicover
and Ray Jackendoff
Department of Linguistics, The Ohio State University, Columbus, OH 43210, USA
Center for Cognitive Studies, Tufts University, Medford, MA 02155, USA
What roles do syntax and semantics have in the gram-mar of a language? What are the consequences of theserolesforsyntacticstructure,andwhydoesitmatter?Wesketch the Simpler Syntax Hypothesis, which holds thatmuch of the explanatory role attributed to syntax incontemporary linguistics is properly the responsibilityofsemantics.Thisrebalancingpermitsbroadercoverageof empirical linguistic phenomena and promises a tigh-ter integration of linguistic theory into the cognitivescientiﬁc enterprise. We suggest that the general per-spective of the Simpler Syntax Hypothesis is well suitedto approaching language processing and language evo-lution, and to computational applications that drawupon linguistic insights.Introduction
What roles do syntax and semantics have in the grammarofalanguage,andwhataretheconsequencesoftheserolesfor syntactic structure? These questions have been centralto the theory of grammar for close to 50 years. We believethat inquiry has been dominated by one particular answertothesequestions,andthattheimplicationshavebeenlessthan salutary both for linguistics and for the relationbetween linguistics and the rest of cognitive science. Wesketch here an alternative approach, Simpler Syntax (SS),which offers improvements on both fronts and contrast itwith the approach of mainstream generative grammar[1–3]. Our approach, developed in three much more extensiveworks[4–6], draws on insights from various alternativetheories of generative syntax[7–11].
Two views on the relation between syntax andsemantics
Acentralidealizationbehindmainstreamgenerativegram-mar,sharedbymuchofformallogicandotherapproachestolanguage, is classical Fregean compositionality (FC):
: ‘The meaning of a compound expression is afunction of the meaning of its parts and of the syntacticrules by which they are combined’. Although many linguistic phenomena are known to beproblematic forthis view,itisfair tosaythatastrongformof FC is generally taken to be a desideratum of syntactictheory construction.FC appears to beviolated, for example, in circumstanceswherecertainaspectsofsentencemeaningdonotseemtoberepresented in the words or syntactic structure of thesentence. In sentence (1), one understands Ozzie to be notonly the ‘tryer’ but also the ‘drinker’, even though the nounphrase
is not overtly an argument of the verb
.(1) Ozzie tried not to drink.Themasterstroke behind mainstream generative gram-mar was to propose that the missing piece of meaning issupplied by an element in a covert level of syntactic struc-ture (‘Deep Structure’ in early work, later ‘Logical Form’).Sentence (1) has the covert form (2), in which the verb
actuallydoeshaveasubject–PRO,anunpronouncedpronoun whose antecedent is
(2) Ozzie tried [PRO not to drink].Such an approach is effective – and appealing – forrelatively straightforward situations such as (1). However,weshowthatcarryingthisstrategythroughsystematically leads to unwelcome consequences. Alternatives to FC are:
: Phrase and sentencemeanings are composed from the meanings of the wordsplus independent principles for constructing meanings,only some of which correlate with syntactic structure.
: Syntactic structure isonly as complex as it needs to be to establishinterpretation.Under SSH, sentence (1) needs no hidden syntacticstructure. The fact that Ozzie is understood as the ‘drinker’results from a principle of semantic interpretation thatassigns
this extra role. Thus, semantics can havemore-elaborate structure than the syntax that expressesit.Box 1makes more precise our notion of syntacticcomplexity.
Mainstream syntactic structures compared withSimpler Syntax
Thechoicebetweenthetwoapproachesto(1)doesnotseemespecially consequential. However, following FC to itslogical end turns out to have radical consequences forthe syntactic analysis of even the simplest sentences.Forexample,Figure1ashowsthestructureofthesentence
Joe has put those raw potatoes in the pot
, based on thetreatment in a contemporary mainstream textbook forbeginning graduate students. The literature offersmany other variants of comparable complexity.Figure 1a is representative of the most recent version of mainstreamtheory,theMinimalistProgram[3,14].Suchastructuretypicallyincorporatesmanyelementsthatdonot
TRENDS in Cognitive Sciences
Jackendoff, R. (firstname.lastname@example.org). Available online 8 August 2006.
www.sciencedirect.com 1364-6613/$ – see front matter
2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2006.07.007