Copyright © 1999 Massachusetts Institute of Technology
A child's linguistic system is shaped to a significant degree by the utterances to which that child has been exposed. That is why a child speaks the language and dialect of his family and community. Nonetheless, there are aspects of the linguistic system acquired by the child that do not depend on input data in this way. Some cases of this type, it has been argued, reflect the influence of a genetically prespecified body of knowledge about human language. In the literature on GENERATIVE GRAMMAR, the term Universal Grammar -- commonly abbreviated UG -- refers to this body of "hard-wired" knowledge.
Questions concerning the existence and nature of UG arise in all areas of linguistics (see discussion in SYNTAX, PHONOLOGY, and SEMANTICS). Research on these questions constitutes a principal point of contact between linguistics and the other cognitive sciences.
Three streams of evidence teach us about the existence and nature of UG. One stream of evidence comes from crosslinguistic investigation of linguistic universals, discussed in the article on TYPOLOGY. Crosslinguistic investigations help us learn whether a property found in one language is also found in other unrelated languages, and, if so, why. Another stream of evidence concerning UG comes from investigation of LANGUAGE ACQUISITIONand learnability, especially as these investigations touch on issues of POVERTY OF THE STIMULUS ARGUMENTS. Work on acquisition and learnability helps us understand whether a property found in the grammar of an individual speaker is acquired by imitation of input data or whether some other reason for the existence of this property must be sought. Finally, evidence bearing on the specially linguistic nature of UG comes from research on MODULARITY AND LANGUAGE. Features of language whose typological and acquisitional footprint suggests an origin in UG may be confirmed as reflections of UG if they reflect aspects of cognition that are to some degree language-specific and "informationally encapsulated." If a fact about an individual speaker's grammar turns out to be a fact about grammars of all the world's languages, if it is demonstrably not a fact acquired in imitation of input data, and if it appears to be specific to language, then we are warranted to suspect that the fact arose from a specific feature of UG.
The questions one asks in the process of building the theory of UG are varied and complex. Suppose the linguist discovers that a property P of one language is present in a variety of other languages. It is often possible that P arises from some more general property of cognition. For example, the repertoire of linguistically relevant THEMATIC ROLES such as "experiencer" or "agent" may reflect language-independent facts about the categorization of events -- and therefore fall outside of UG. On the other hand, while the repertoire of thematic roles may be language-independent, the opposite is true of the apparently universal mapping of specific thematic roles onto specific designated syntactic positions -- for example, the fact that agents are mapped universally onto a structurally more prominent position than patients (e.g., subject position). This specifically linguistic mapping thus constitutes one of the properties attributed to UG.
It is also important to try to distinguish UG-based universals from apparent universals that merely reflect usability conditions on languages that must serve a communicative function (functional universals). For example, there is probably a lower bound to the size of a language's phonemic inventory. Does this restriction form part of UG? Not necessarily. It is equally possible that the limitation merely reflects a consequence of usability conditions for linguistic systems. The words of a language whose only phonemes are /m/ and /a/ would be extraordinarily long and hard to distinguish. Such a language might fall within the range permitted by UG, yet never occur in nature because of its dysfunctionality.
Because of the ever-present possibility that a universal may have a functional explanation, researchers interested in discovering properties of language that derive from UG often focus on those universals for which functional explanations are the least likely. For example, syntactic research has paid particular attention to a number of limitations on form-meaning pairs that have just this property of "dysfunctionality." One example is a set of restrictions specific to how and why questions. How did you think Mary solved the problem? can be a question about Mary's problem-solving methods, but the sentence How did you ask if Mary solved the problem? cannot. (It can only be a question about methods of asking.) The restriction concerns the domains from which WH-MOVEMENT may apply, which in turn correlates with sentence meaning. The restriction appears to be a genuine universal, already detected in a wide variety of languages whose grammars otherwise diverge in a number of ways. Crucially, the restriction makes no evident contribution to usability. Just the opposite: it prevents speakers from posing perfectly sensible questions, except through circumlocution -- for example, I know that you asked if Mary had solved the problem in some particular way. What was that way? The study of such seemingly dysfunctional aspects of language has provided an especially clear path to a preliminary understanding of UG. This fact also explains why the data of generative grammar stray so often from "everyday" linguistic facts -- a central difference between the concerns of generative grammarians and those researchers more concerned with "everyday" language use (a group that includes some sociolinguists as well as computational linguists interested in practical language technologies).
The existence of "Universal Grammar" (uppercase) does not necessarily entail the existence of a "universal grammar" (lowercase) -- in the sense of a usable linguistic system wholly determined by genetic factors. UG must allow for language variation, though by its very nature it restricts the range of variation. This is why certain nonuniversal properties of language nonetheless recur in widely scattered, unrelated languages, while other equally imaginable properties are never found. For example, the placement of the finite verb in "second" position characteristic of the Germanic languages (see HEAD MOVEMENT) is also found in Vata (Ivory Coast; Koopman 1983), Kashmiri (Bhatt 1995), and Karitiana (Brazil; Storto 1996). By contrast, in no known language are verbs obligatorily placed in third position. In other words, UG allows languages to vary -- but only up to a point.
There are several theories of how variation is built into UG. One proposal holds that the principles of UG define the parameters of possible variation. Language acquisition involves "setting" these parameters (see PARAMETER-SETTING APPROACHES TO ACQUISITION, CREOLIZATION, AND DIACHRONY). Another suggestion, advanced by Borer (1981) and Borer and Wexler (1987), holds that true variation is limited to the LEXICON (one aspect of language that we know must vary from language to language). Apparent syntactic variation on this view arises from the differing syntactic requirements of lexical items (see also SYNTAX, ACQUISITION OF). Another proposal, developed within OPTIMALITY THEORY, attributes variation to differences in the ability of particular grammatical principles to nullify the action of other principles with which they conflict (i.e., differences in "constraint ranking").
It is not entirely clear what aspects of UG are subject to variation. In particular, though no one doubts that syntactic and phonological systems vary across languages, the question of variation in semantics is more contested. The details of semantic interpretation are probably less obvious to young children acquiring language than are the details of word positioning and word pronunciation that provide evidence about syntax and phonology. Consequently, it is conceivable (though not inevitable) that the laws governing compositional semantic interpretation of syntactic structures are wholly determined by UG -- hence invariant across languages. In fact, the notion and the term was borrowed by Chomsky (1965, 1966) from an earlier grammatical tradition that explicitly sought universal semantic roots of syntax (for example, the 1660 Port-Royal Grammaire génerale et raisonée). Semantic universals do exist, of course. Not only basic laws of semantic composition (Heim and Kratzer 1998), but also many details recur in language after language. For example, the classification of predicates into something like "states" versus "events," and the interaction of this classification with such properties as quantifier interpretation, seems to be invariant (or nearly so) across the languages that have been studied. On the other hand, other facts might cause one to doubt whether all languages command exactly the same semantic resources. For example, although "multiple questions" such as Who bought what? receive similar interpretations in many languages (as questions whose answer provides a list of pairs; e.g., John bought the wine and Mary bought the dessert), this semantic possibility is entirely absent in some languages, including Italian and Irish. To native speakers of these languages, the counterparts to Who bought what? (e.g., Italian *Chi a comprato che cosa?) seem quite uninterpretable. Whether such facts indicate the existence of semantic variation, or merely reveal lexical or syntactic differences with a predictable impact on semantics, remains an open question.
The fact that variation is "built into" some aspects of UG does not preclude the possibility that UG might characterize a usable "default" grammar on its own. This is also a matter of considerable controversy. Bickerton (1981), for example, has suggested that CREOLES represent the spontaneous flowering of a purely UG-based grammar, but this view is controversial (Mufwene 1996, 1999). Furthermore, a precedent for "usable UG" is provided elsewhere in the animal kingdom by songbird species whose song is partly learned rather than totally innate. Researchers have identified a "UG" for the song of several such species. When birds of these species are reared in isolation, they spontaneously develop a song that falls recognizably within the parameters of their UG, though rudimentary in many ways (Marler 1991, 1996; see also ANIMAL COMMUNICATION). The UG of songbirds is of importance for another reason. Among those who have not made a study of the relevant evidence, theories of UG are often thought to require special pleading, as if the hypothesis of species-specific innate knowledge constituted a violation of Occam's razor. The evidence from songbirds makes it clear that a priori objections to UG are nothing more than a prejudice. The nature of human UG remains, however, a topic of lively debate and continued research.
Bhatt, R. (1995). Verb movement in Kashmiri. In R. Izvorski and V. Tredinnick, Eds., U. Penn Working Papers in Linguistics, vol. 2. Philadelphia: University of Pennsylvania Department of Linguistics.
Bickerton, D. (1981). Roots of Language. Ann Arbor, MI: Karoma.
Borer, H. (1983). Parametric Syntax. Dordrecht: Foris.
Borer, H., and K. Wexler. (1987). The maturation of syntax. In T. Roeper and E. Williams, Eds., Parameter Setting. Dordrecht: Reidel.
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press.
Chomsky, N. (1966). Cartesian Linguistics. New York: Harper and Row.
Heim, I., and A. Kratzer. (1997). Semantics in Generative Grammar. Oxford: Blackwell.
Koopman, H. (1983). The Syntax of Verbs: from Verb Movement Rules in the Kru Language to Universal Grammar. Dordrecht: Foris.
Marler, P. (1991). Differences in behavioural development in closely related species: birdsong. In P. Bateson, Ed., The Development and Integration of Behaviour. Cambridge: Cambridge University Press, pp. 41-70.
Marler, P. (1996). Song Learning. http://www.hip.atr.co.jp/~bateson/hawaii/abstracts/marler_ms.html.
Mufwene, S. S. (1996). The Founder Principle in creole genesis. Diachronica 13:83-134.
Mufwene, S. (1999). On the language bioprogram hypothesis: Hints from Tazie. In M. DeGraff, Ed., Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge, MA: MIT Press.
Storto, L. (1996). Verb Raising and Word Order Variation in Karitiana. Unpublished manuscript, Massachusetts Institute of Technology .
Copyright © 1999 Massachusetts Institute of Technology