Encyclopedia of Language & Linguistics || Formalism/Formalist Linguistics

580 Formalism/Formalist Linguistics

Formalism/Formalist Linguistics
P ten Hacken, University of Wales, Swansea, UK
� 2006 Elsevier Ltd. All rights reserved.

In formalist approaches to linguistics, the form oflanguage is described independently of other aspects,such as its function. Mentalist approaches, includingChomskyan linguistics, Lexical-Functional Gram-mar, and Head-Driven Phrase Structure Grammar,aim to describe and explain the linguistic competenceof a speaker. Purely formalist approaches, includingGeneralized Phrase Structure Grammar and Monta-gue Grammar, study language as an abstract object.In Chomskyan linguistics, a grammar describes theindividual speaker’s competence, and universal gram-mar describes the genetically determined languagefaculty. The language faculty is essential for (first)language acquisition. This model is also used as abasis to explain aspects of language use.

General Characterization

Language can be studied from a variety of perspec-tives. Formalist approaches to linguistics have incommon that they focus on the form of language.This does not mean that other aspects of language –for example, its use – are necessarily ignored. It doesmean, however, that the aim is to describe the formof language independently of these other aspects.

An alternative interpretation of the label formalistfor an approach to linguistics is that the aim of suchan approach is to express generalizations about lan-guage in terms of a formalism. Although this propertyis logically independent of the focus on form, formal-ist approaches generally satisfy both interpretations.

The assumption that the form of language is deter-mined by an autonomous system is shared by allformalist approaches. Autonomy implies that the sys-tem can be described independently of other systems.It does not imply that there is no interaction withother systems. It is this autonomy that makes itworthwhile to study language through its form.

Two main positions can be identified on the ques-tion as to how this autonomous system is realized.One is the mentalist view, which assumes that lan-guage is a knowledge component in the mind/brain ofthe speaker. This is the position adopted, for instance,by Noam Chomsky. The other position is formalistin the narrow sense. It assumes that a language isan abstract object. This view is found especially inthe logical tradition of formal semantics and impliesthat what individual speakers know is an imperfectreflection of the actual language.

Formalist approaches to linguistics are opposed toapproaches that focus on the function or use of lan-guage. The latter include functionalist linguistics andsociolinguistics. The aim of functionalist linguistics isto explain the form of language through its function.An example of a phenomenon often studied in thisway is grammaticalization. Grammaticalization is theprocess by which content words become functionwords or morphemes. Sociolinguistics studies theuse of language without drawing conclusions aboutthe nature of the underlying formal system. An exam-ple of a phenomenon studied in sociolinguistics ispoliteness. The relationship of these two approachesto formalist ones is not the same. Functionalist lin-guistics denies the autonomy of the language system,because it tries to explain it on the basis of externalfactors. Therefore it is not compatible with formalistapproaches. Most work in sociolinguistics, however,can be interpreted as complementary to formalistapproaches. Often it is neutral as to the choice of aformalist or a functionalist framework.

Among formalist approaches, Chomskyan linguis-tics occupies a prominent position. It is a long-estab-lished research program that attracts significantnumbers of scholars. In the nearly 50 years of itsexistence it has provoked extensive discussion, sodetailed documentation exists about its assumptionsand their implications. Botha (1989) gave a system-atic overview of the issues and summarized the dis-cussion of the 1970s and 1980s. Most alternativeformalist approaches to linguistics make a point ofspecifying how they differ from Chomskyan linguis-tics. For this reason, this article will first describe theresearch program of Chomskyan linguistics beforedescribing the main alternative approaches withinformalist linguistics. Finally, the position of languageuse in these approaches will be addressed.

The Research Program of ChomskyanLinguistics

Most formalist approaches to linguistics have theambition to develop linguistics as an empirical sci-ence. In an empirical science, a theory explainsthe observations in the real world by describing theunderlying system. Work in an empirical science isoften conceived of as consisting of a cyclical succes-sion of four stages, known as the empirical cycle.According to the empirical cycle, scientific workstarts with the collection of data. The next stageis to find generalizations about these data. Then atheory is hypothesized to explain the data. Finally,this theory is tested, which supplies further data,

Formalism/Formalist Linguistics 581

often contradicting the theory under consideration,so that the cycle can start again. Progress in science isthe incorporation of more and more data in the ex-planatory scope of the theory. This model is stilladopted, usually implicitly, by many researchersworking in empirical science.

Logically, the empirical cycle is not sufficientlyexplicit to guide scientific research. At every stage,too many alternative possibilities would have to beconsidered. In the transition from generalizations totheories, for instance, a sufficiently imaginative re-searcher could come up with more theories thanthey could test in their lifetime. In scientific practice,however, such problems do not arise. In the percep-tion of most researchers, it is difficult enough to comeup with a single theory that fits the data. The discrep-ancy between the logical plethora of theories andtheir scarcity in scientific practice can be explainedby assuming a research program. A research programis the set of assumptions guiding the researcher to-ward sensible decisions at any step in the empiricalcycle. These assumptions need not be conscious tothe researcher and are often not formulated explicitly.In the same way as a good chess player does not seeall logical possible moves but only the sensible ones, ascientist working in a research program automaticallyconstrains the possible theories to be considered.

Chomskyan linguistics is a research program inlinguistics. As such, it should be distinguished fromChomsky’s linguistic theory. While both were con-ceived by Noam Chomsky in the late 1950s, theiraims and later development are strikingly different.Chomsky’s linguistic theory went through a numberof stages in its development, including standardtheory (ST) in the 1960s, government and bindingtheory in the 1980s, and the minimalist programin the 1990s. In each of these stages, tree-structurerepresentations of sentences played a role, althoughthe constraints on the production and manipulation ofthese tree structures varied. Chomskyan linguistics,by contrast, remained stable during this period.It does not refer to tree structures but specifies whata linguistic theory should explain and how such atheory should be evaluated.

Chomskyan linguistics defines the object of studyas the knowledge of language a speaker has. Thisknowledge is called the linguistic competence orinternalized language (I-language). It is not open toconscious, direct introspection, but a wide range of itsmanifestations can be observed and used as data forthe study of language.

Competence is opposed to performance. Perfor-mance includes various types of language use; forinstance, a collection of utterances, a corpus oftexts, a set of grammaticality judgments, and a set

of results of psycholinguistic experiments. They canbe used as data in linguistic research, but in every casethe relationship they have to competence should bekept in mind in their use. Naturalistic data (utter-ances, texts) reflect the interaction of competencewith a number of other factors. Examples of suchfactors are knowledge of the situation in whichthe utterance was made, intention on behalf of thespeaker, memory limitations, and degree of fatigue.Because the role of many of these factors is poorlyunderstood, it is often difficult to use naturalistic dataas evidence for the nature of competence. Experimen-tal data also reflect competence in interaction withother factors, but in setting up the experiment, anattempt is made to control the non-competence fac-tors. The most frequently used type of experiment isthe grammaticality judgment. There is no principledreason for treating it in a special way, but becausegrammaticality judgments are both highly informa-tive and easy to obtain, they have always been widelyused in Chomskyan linguistics.

A grammar is a theory about the competenceof an individual speaker. It describes the competenceas the system underlying the observed data. As such,the grammar explains the data. The term grammar issometimes also used to refer to competence, butthis use is potentially confusing. In early stages ofChomskyan linguistics, including Standard Theoryas outlined by Chomsky (1965), grammars were de-scribed in terms of rewrite rules and transforma-tions. At this stage of the theory, rewrite rules ofthe type S ! NP VP generated tree structures andtransformations that operated on these tree struc-tures to account for phenomena such as subject-verb agreement, passive constructions, and questionformation.

The research program as described so far encoun-ters a serious epistemological problem. If we assumea set of grammaticality judgments as our data thereare infinitely many grammars (sets of rewrite rulesand transformations) consistent with our data. This isa result obtained in mathematical linguistics, thetheory of formal grammars. Because the aim of lin-guistics is to identify the grammar describing theactual competence, we need a method to distinguishthis grammar from the other ones compatible withour data. Adding more data of the same type willexclude some grammars but not reduce the problemin a principled way. At any point in time, our set ofdata will be finite, and for any finite set of data thereare indefinitely many different grammars.

As a solution to this problem, Chomskyan lin-guistics adopts a cross-linguistic perspective, focusingon language acquisition. In other theoretical frame-works, different languages have been compared to

Figure 1 The research program of Chomskyan linguistics.


reconstruct a common parent language (historical-comparative linguistics) or to find universals(typology). In Chomskyan linguistics it is assumedthat certain genetically determined properties of thehuman mind are necessary for language acquisition.These properties are therefore reflected in one way oranother in any language. By comparing different lan-guages, linguists working in the Chomskyan frame-work try to find evidence of properties of thelanguage faculty that make the acquisition of theselanguages possible. These properties have little incommon with superficial inductive generalizations.Biologically, they are supposed to be encoded geneti-cally in the same way as, for instance, the propertythat the human hand has four fingers and an opposedthumb, but they are more difficult to detect becauseindividual languages may reflect them in differentways.

Various indications make the choice of languageacquisition as the source of universals plausible. Ac-quisition of a language is necessary for its existence.Data from child language acquisition show that formany features of syntax, the child does not learn byrepetition and resists correction. Moreover, the dis-crepancy between competence and performance leadsto many ungrammatical and incomplete sentencesin the input that children receive. The resultingI-language acquired by the child, however, is muchmore similar to the competence of the people in his orher environment than is a grammar that allows thesentences in the performance the child receives asinput. While such considerations make a languagefaculty as a genetic component of the human mindplausible, conclusive empirical evidence for or againstsuch an assumption is hard to imagine and probablyimpossible in principle.

In the research program of Chomskyan linguistics,the language faculty is described by a universal gram-mar (UG). Sometimes the language faculty is alsocalled the language acquisition device or UG, butthese names are less appropriate, because the formerconflates the knowledge incorporated in the languagefaculty with its use and the latter conflates it with itsdescription. The interaction of UG and grammars ofindividual I-languages is an essential factor in theepistemological validity of the Chomskyan researchprogram. Since the purpose is to describe on the onehand all I-languages and on the other the languagefaculty that makes their acquisition possible, a properbalance has to be struck between what is geneticallydetermined and what is acquired. If UG is too restric-tive, it will not be possible to come up with grammarsfor all I-languages. If UG is not restrictive enough, itwill not be possible to explain the learnability of theI-languages.

The interaction of the main components ofChomskyan linguistics can be represented as shownin Figure 1. Language is studied at the universallevel (1), at the individual level (2), and at the levelof data (3). These levels are represented as the threenumbered boxes. On the left we find the real-world entities at each level: the language faculty, theindividual competence, and the performance data.On the right we find the constructs of linguistics:UG, individual grammars, and observations aboutperformance. At each level, the linguist’s constructdescribes the real-world entity. In interpreting therelationships between the different levels, it shouldbe kept in mind that there are many instances of thelower level corresponding to a single instance ofthe higher level. For example, by describing anI-language, a grammar can explain (aspects of) alarge set of performance data. Conversely, thesedata can be used as a test for the grammar. At thehigher level, UG describes the language faculty andexplains thereby (aspects of) a large set of I-lan-guages. Again, conversely, I-languages can be usedas a test for UG, because UG has to allow for adescriptively adequate grammar that is in additionlearnable.

While Chomsky (1965) recognized the epistemo-logical need for UG, he also observed that ST didnot incorporate the mechanisms for formulating it.By progressive generalization and cross-linguistic


comparison, a model of the language facultyemerged, which was formulated as the Governmentand Binding Theory (GB theory) by Chomsky (1981).In this model, UG consisted of a number of principles,which were universal. Differences between I-lan-guages were accounted for by including parametersin the principles. A parameter specified a number ofoptions, typically two. In language acquisition, para-meters were set by selecting one of these options. Thegrammar of an I-language was thus determined by theparameter settings selected. The Minimalist Program(MP) outlined by Chomsky (1995) changed manyof the theoretical assumptions of GB theory butcontinued to adhere to the principles and parametersmodel. An example of a difference between GBtheory and the MP is the way movement was con-strained. In GB theory, the individual transformationrules of ST were replaced by a single, general rulemove a. This meant that anything could move any-where unless a constraint prohibited it. In the MP, theperspective was reversed: every movement had to bemotivated.

A noteworthy consequence of Chomskyan linguis-tics is that language can be studied at the level ofindividual I-languages or of the universal languagefaculty, but not at the level of major languages ordialects. There is no sense of language in Chomskyanlinguistics, such that, for instance, David Beckhamand Paul McCartney share the same language. Sincethey have different minds, they cannot share a mentalcomponent. The similarity of their I-languages, whichby no means amounts to identity, can be explained asa consequence of the limited choice permitted byparameter settings, combined with a large overlap invocabulary.

Other Mentalist Approaches

Adoption of a formalist approach to linguistics andacceptance that language is a knowledge componentin the mind of the speaker by no means implies ad-herence to the research program of Chomskyan lin-guistics. There are many different ways of elaboratingthese two basic assumptions into a research program.Although the research program of Chomskyan lin-guistics has been debated more intensively and devel-oped in more detail, this does not mean that theothers are less valid in any sense. In this section, twoapproaches will be presented that have been devel-oped over the past decades and continue to attract asignificant group of researchers: Lexical-FunctionalGrammar (LFG) and Head-driven Phrase StructureGrammar (HPSG).

Lexical-functional grammar emerged in the late1970s as a reaction against certain aspects of

Chomskyan linguistics. However, the main focus ofdebate at the time was not the research programof Chomskyan linguistics, as described in the previ-ous section, but the interpretation of a number ofindividual theoretical results. An example of a prom-inent issue was the so-called psychological reality ofgrammars for individual languages. In any stage ofChomskyan linguistics, a sentence is represented as atree structure generated by rewrite rules and sub-sequently affected by transformations that move ele-ments of the tree to other positions. The principlesand parameters of GB theory are constraints on re-write rules and movement. Some researchers arguedthat if such a grammar is a psychologically real de-scription of the speaker’s competence, a sentencethat requires more movement operations or morecomplex ones should take longer to process thanone with fewer or simpler movement operations.They devised psycholinguistic tests to evaluate thishypothesis and found that processing times were notaffected in the expected way. Cognitive complexitycould not be related to the complexity of syntacticderivations in a transparent way. This stimulatedthe development of grammar formalisms in whichthe role of transformations was reduced or elimi-nated. Lexical-functional grammar is one of themost successful of them.

As the contributions to Bresnan (1982) indicated,LFG was devised mainly by Joan Bresnan. Instead oftree structures and transformations, LFG has a treestructure and a functional structure for each sentence.The functional structure represents grammaticalfunctions, such as subject and predicate, in a featurestructure. The tree structure and the functional struc-ture are linked by a unification-based procedure. Inthis way, transformations are no longer required, sothe problem they pose for the psychological reality ofgrammars no longer arises.

At first sight, the research program of LFG was notfundamentally different from that of Chomskyan lin-guistics. Superficially, the main difference seems to bethe nature of the grammar formalisms. A grammar inLFG also intends to describe the competence of aspeaker. When we consider the universal level, how-ever, a more principled difference can be observed.Lexical-functional grammar assumes that the formalmapping procedure between the tree structure andthe functional structure is universal. This assumptionconstrains the formulation of grammars in a wayparallel to the function of UG in Chomskyan linguis-tics. This means that the language faculty in LFG isinterpreted not as the knowledge needed for languageacquisition but as a mechanism for language proces-sing. Therefore, the research program of LFG can alsobe represented as in Figure 1, but rather than UG, a


universal mechanism for processing language, i.e.,mapping between form and meaning, describes thelanguage faculty.

Another influential framework is HPSG, whichemerged in the 1980s. Originally, as represented inPollard and Sag (1987), it was strongly influenced bylogical approaches to syntax (generalized phrasestructure grammar; see the following section) andsemantics (situation semantics). In more recentrepresentations of the framework, Pollard and Sag(1994) and Sag and Wasow (1999) adopted a men-talist position. However, work in HPSG is moreconcerned with developing grammars than with dis-cussing their status. In HPSG formalism, treestructures have been replaced entirely by featurestructures. Rather than different levels of representa-tion that are related to each other by transformationsor other operations, HPSG adopts a single level. Allinformation about what is called a sign (lexical item,phrase, sentence, etc.) is combined into a single struc-ture. This includes phonology, syntax, semantics, andpragmatics. Relationships between components ofthe structure are indicated by co-indexation. Thisapplies to both, for instance, a pronoun and its syn-tactic antecedent and to syntactic, semantic, andpragmatic representations of the same item. Thegrammar formalism in HPSG seems to be understoodas a universal mechanism with language-specificparameters for building up feature structures, whichsuggests a research program similar to the one forLFG. The HPSG literature is in general more interest-ed in describing individual languages than in thenature and status of universals.

Purely Formalist Approaches

Formalist approaches to grammar are often labeled asgenerative grammar. The term ‘generative’ stemsfrom the theory of formal languages. A formal gram-mar is a set of rewrite rules that generates a set ofsentences. The set of sentences is called a language.

In early Chomskyan linguistics, the mechanism ofrewrite rules played an important role as a theoreticaldevice. The role of the formulation of individual re-write rules has declined in the transition to the prin-ciples and parameters model in GB theory. Theconception of a language as a set of sentences goescounter to the definition of the research topic in anystage of Chomskyan linguistics. Chomsky (1986)called it an E-language (E for externalized), asopposed to the I-language he wanted to study.

Other linguists and philosophers pursued the ideathat natural language should be studied as a set ofgrammatical sentences. Quine (1972) summarizedthe epistemological implications of this assumption.

In his view, a language like English is ultimately aset of grammatical sentences. Although this set isinfinite, it can still be determined for each sentencewhether it is grammatical or not. For any set of sen-tences, there are infinitely many possible grammars.Rather than finding the correct one among them, thisapproach considers all of them as equally valid. Itis not the knowledge of individual people that isinteresting but only the language as an abstractobject. English is an abstract object, of which individ-ual speakers have imperfect knowledge, encoded inpotentially quite different, more or less imperfectgrammars. This position was elaborated by Katzand Postal (1991).

In the domain of semantics, this view of languagehas a long tradition in the philosophical study of truthand reference. In formal logic, models were developedfor calculating whether a particular logical formulais true or false, given a specific assignment of truthvalues to primitive statements. It was RichardMontague who extended the use of these mechanismsto complex phenomena in natural language, such asquantification. Compared to Chomskyan linguistics,Montague grammar reversed the relative importanceof syntax and semantics. Whereas in Chomskyanlinguistics, syntax is the central area of linguistics,Montague found syntax not interesting by itself butonly as a contribution to calculating the semantics ofsentences. He used categorial grammar instead ofrewrite rules to represent syntax. In this formalism,constituents are characterized in terms of their rela-tionship to referring expressions and truth values. Hismodel-theoretic semantics was meant to characterizeformally the set of possible worlds correspondingto the truth of a sentence. After Montague’s death,a large number of followers continued to pursuethis idea.

In the domain of syntax, the progressive formula-tion of principles generalizing over individual rules inChomskyan linguistics gave rise to the objection thatthe grammars were no longer generative because theywere not formalized in enough detail to determinewhether a particular sentence would be generated ornot. Many researchers who had been attracted byChomsky’s (1965) ST because of its formality didnot accept this development. The introduction ofthese principles was motivated at least in part by theproblems caused by the excessive mathematicalpower of transformations. Therefore, Gerald Gazdardeveloped a formalism that approached this problemin an alternative way and dispensed with transforma-tions altogether. In Generalized Phrase StructureGrammar (GPSG), even long-distance dependenciesare expressed in terms of a purely ‘context-free gram-mar’ (CFG). A CFG consists only of rewrite rules of


the type A ! a, where A stands for a single symboland a for a string of symbols. An example of a long-distance dependency is found in the following: Whodid John think Mary loves? Here, who is interpretedas the object of loves. In Chomskyan linguistics,the usual way of expressing this is to say thatwho originates to the right of loves but is moved tothe start of the sentence by a sequence of transforma-tions. Gazdar demonstrated that a CFG can accountfor this as well as for many other relationships tradi-tionally described with transformations. To achievethis, he decomposed syntactic categories into featurestructures and formulated so-called meta-rules. Ameta-rule is a rule that takes a CFG rule as inputand produces another CFG rule. Meta-rules haveaccess to individual features inside the syntacticcategories.

One of the main advantages of CFGs is their easyimplementation on a computer. However, Bartonet al. (1987) demonstrated that the system of meta-rules adopted in GPSG made grammars computa-tionally intractable. Although GPSG has lost muchof its attraction and is no longer pursued as a majorgrammar formalism, it contributed significantly tothe theory of feature structures and was at the basisof the HPSG formalism.

Formalist approaches in the narrow sense do nothave a research program parallel to the ones discussedin earlier sections of this article. They do not investi-gate language as an empirical entity, represented inthe speaker’s mind but only as an abstract object. Asa consequence, their research is not guided by theempirical cycle. It is more similar to formal logic ormathematics, in which theorems are derived fromaxioms. The axioms correspond to the grammarand the theorems to the sentences. Contrary to com-mon practice in logic and mathematics, formal lin-guistics starts with a set of theorems (grammaticalsentences) and non-theorems (ungrammatical sen-tences) and searches for a set of axioms that generatesall theorems and none of the non-theorems.

Aspects of Language Use

In all formalist approaches to linguistics, the theoryof the language system is intended to be part of ageneral, empirical account of language-related phe-nomena. The hypothesis that language can be de-scribed as an autonomous system, whether realizedas a component of the mind/brain or as an abstractsystem, implies that the interaction between thissystem and other factors can be invoked to accountfor phenomena related to language use.

The use of language in communication is one of themost prominent phenomena that have to be coveredin this context. Yet it should be emphasized thatlanguage and communication are independent. Thiscontrasts with the view adopted in functionalistapproaches, where communication is invoked to ex-plain language. In formalist approaches, language isnot considered as a tool for communication (althoughit can be used for communication) and communica-tion is not seen as dependent on language (although itoften involves language). By recognizing this mutualindependence, relevance theory, as developed bySperber and Wilson (1986), is compatible with for-malist approaches to the description of language. InHPSG, the information pertaining to language usecan be encoded directly in the feature structuresrepresenting signs. In Chomskyan linguistics, it hasbeen suggested that there is a mental component ofpragmatic competence interacting with grammaticalcompetence. Kasher (1991) elaborated this idea.

Since the 1980s, a number of other phenomenabased on language use have gained a prominent posi-tion on the research agenda of Chomskyan lin-guistics. They include first and second languageacquisition and language change. The reason fortheir prominence is that they can be used as a sourceof data about the principles and parameters involvedin language.

Whereas the logical problem of language acquisi-tion is at the basis of the language faculty hypothesisdiscussed in the section on Chomskyan linguistics, thepractical problem of first language acquisition is ana-lyzed in the principles and parameters model as theprocess of parameter setting by the child on the basisof performance data provided by people in the child’senvironment. The main question here is how the lan-guage faculty interacts with general cognitive devel-opment. Wexler (1999) advocated the hypothesis thatthe language faculty matures in the process of lan-guage acquisition. Lust (1999) proposed instead thatthe language faculty remains stable and that the firstlanguage competence is a distinct component fromthe language faculty.

Second language acquisition leads to a state ofcompetence that is usually less than the one achievedfor the first language. It is often called interlanguage.In Chomskyan linguistics, an interlanguage is anI-language. The question is what roles are played bysuch factors as the learner’s cognitive development,the type of access to the language faculty, and thefirst language competence in the emergence ofinterlanguage competence. Gregg (1996) gave anoverview of some of the main positions.


Language change is the historical development oflanguages such as English. In Chomskyan linguistics,English is considered an epiphenomenon. Its changemeans that different generations of speakers havesystematic differences in language competence. Light-foot (1999) presented a model in which it is assumedthat parameter settings determining a speaker’s gram-mar do not change in their lifetime but their use of thegrammar does. The use of certain constructions maydrop below a threshold necessary for a new genera-tion of speakers to set the parameters in the same wayas their parents. When this happens, the language isperceived as having changed.

In this section, various fields have been touchedupon that crucially involve language use. The rolethat these fields play in approaches to linguisticsthat assume an autonomous system of languagedemonstrates that this assumption does not preventthe study of language use. Instead, it guides this studyin a fruitful and constructive direction. The principlesand parameters model adopted in Chomskyan lin-guistics stimulates this type of research in particular,because it provides external evidence about thenature of principles and parameters.

See also: Autonomy; Chomsky, Noam (b. 1928); E-Lan-

guage versus I-Language; Functionalist Theories of Lan-

guage; Grammar; Head-Driven Phrase Structure

Grammar; Lexical Functional Grammar; Linguistic Para-

digms; Linguistic Universals, Chomskyan; Minimalism;

Montague, Richard (1931–1971); Politeness; Pragmatics:

Overview; Principles and Parameters Framework of Gen-

erative Grammar; Second Language Acquisition: Phonol-

ogy, Morphology, Syntax.

References

Barton G E Jr, Berwick R C & Ristad E S (1987). Compu-tational complexity and natural language. Cambridge,MA: MIT Press.

Botha R P (1989). Challenging Chomsky: the generativegarden game. Oxford: Blackwell.

Bresnan J (ed.) (1982). The mental representation ofgrammatical relations. Cambridge, MA: MIT Press.

Bresnan J (2001). Lexical-functional syntax. Oxford:Blackwell.

Chomsky N (1965). Aspects of the theory of syntax.Cambridge, MA: MIT Press.

Chomsky N (1981). Lectures on government and binding.Dordrecht: Foris.

Chomsky N (1986). Knowledge of language: its nature,origin, and use. Westport, CT: Praeger.

Chomsky N (1995). The minimalist program. Cambridge,MA: MIT Press.

Gamut L T F (1991). Logic, language, and meaning.Chicago: University of Chicago Press.

Gazdar G, Klein E, Pullum G & Sag I (1985). Generalizedphrase structure grammar. Oxford: Blackwell.

Gregg K R (1996). ‘The logical and developmental prob-lems of second language acquisition.’ In Ritchie W C &Bhatia T K (eds.) Handbook of second languageacquisition. San Diego: Academic Press. 49–81.

Jackendoff R S (1993). Patterns in the mind: languageand human nature. New York: Harvester/Wheatsheaf.

Kasher A (1991). ‘Pragmatics and Chomsky’s research pro-gram.’ In Kasher A (ed.) The Chomskyan turn. Oxford:Blackwell. 122–149.

Katz J J & Postal P M (1991). ‘Realism vs. conceptu-alism in linguistics.’ Linguistics and Philosophy 14,515–554.

Lightfoot D (1999). The development of language: acquisi-tion, change, and evolution. Oxford: Blackwell.

Lust B (1999). ‘Universal grammar: the strong continuityhypothesis in first language acquisition.’ In Ritchie W C& Bhatia T K (eds.). 111–155.

Newmeyer F J (1998). Language form and language func-tion. Cambridge, MA: MIT Press.

Pollard C & Sag I A (1987). Information-based syntax andsemantics 1: Fundamentals. Stanford, CA: Center for theStudy of Language and Information.

Pollard C & Sag I A (1994). Head-driven phrase structuregrammar. Stanford, CA: University of Chicago Press,Chicago: Center for the Study of Language and Informa-tion.

Quine W V (1972). ‘Methodological Reflections on CurrentLinguistic Theory.’ In Davidson D & Harman G (eds.)Semantics of natural language. Dordrecht: Reidel.442–454.

Ritchie W C & Bhatia T K (eds.) (1999). Handbook of childlanguage acquisition. San Diego: Academic Press.

Sag I A & Wasow T (1999). Syntactic theory: a formalintroduction. Stanford, CA: Center for the Study ofLanguage and Information.

Sperber D & Wilson D (1986). Relevance: communicationand cognition. Oxford: Blackwell. [2nd edn., 1995.]

Thomason R H (ed.) (1974). Formal philosophy: selectedpapers of Richard Montague. New Haven: Yale Universi-ty Press.

Uriagereka J (1998). Rhyme and reason: an intro-duction to minimalist syntax. Cambridge, MA: MITPress.

Wexler K (1999). ‘Maturation and growth of grammar.’ InRitchie W C & Bhatia T K (eds.). 55–109.

Documents

Encyclopedia of Language & Linguistics || Formalism/Formalist Linguistics