Encyclopedia of Linguistics
Words can be divided into two basic classes: (1) lexical or open class words and (2) function or closed class words.
Nouns (e.g., dog, house), verbs (e.g., to go, to see), and adjectives (e.g., green, nice) supply the bulk of the meaning in a sentence and belong to the lexical class. This class is called “open” because languages can freely add new words to the set. English function words, on the other hand, include determiners, such as the and a(n); auxiliaries, such as might, have, and be; conjunctions, such as and, that, and whether; and degree adverbs, such as very and too. These words are referred to as “functional” or “grammatical” words because they carry little meaning (have no synonyms) and typically “help” another word. Determiners, for instance, add grammatical information about specificity and definiteness (the dog vs. a dog), but do not essentially alter the meaning. They are also called “closed class words,” since languages do not easily add new words to the set.
Lexical words typically carry intonational emphasis or stress, while function words are generally unstressed. Therefore function words are prone to contraction--for example, the auxiliary have in I've seen it.
The distinction between function and lexical words has been very fruitful for linguistic description. So-called analytical languages, such as Chinese, are characterized by an abundance of function words. In contrast, function words are typically lacking in the speech of young children, certain kinds of aphasia, and telegraphic speech. It is also well known that languages rarely borrow function words from other languages or make up new ones (hence their status as closed class). Most recent innovations in the English vocabulary, such as pizza, angst, fax, e-mail, phat, AIDS, website, browser, screenager, to surf, Nethead, and techno-babble are lexical rather than functional in nature (see, for example, the journal American Speech for lists of new words).
Function words add mainly grammatical information, which means that they are defined above all by their syntactic behavior. Most traditional grammars assumed (and their descendents continue to assume) that the structure of sentences and phrases is determined mainly by lexical words. Function words were regarded as mere additions to lexical phrases. Thus, the sentence The rabbit will see the fox was analysed as a noun phrase the rabbit, followed by a verb phrase will see the fox. The determiner the was thus an addition to the noun phrase, and the auxiliary will was added to the verb phrase.
A shift in this thinking came in the 1980s within the framework of generative grammar. From then on, auxiliaries were attributed with an independent contribution to the sentence structure. However, function words still did not determine the categorial status of a phrase--for example, a phrase such as the rabbit continued to be regarded as a noun phrase containing a determiner.
This view changed radically by the mid-1980s, as function words were increasingly interpreted as the determinants of the categorial status of sentence elements. To use the technical terminology, function words were “projecting to a phrase” or “heading a phrase.” Determiners, for example, came to be regarded as the head of determiner phrases--that is, the rabbit was now interpreted as a determiner phrase the . . . containing the noun phrase rabbit. “Functional projections” were thus assigned a structure similar to “lexical phrases.”
Research in the late 1980s and 1990s revolved around the question of exactly which functional projections a sentence may contain. Each function word expressing a grammatical function was soon regarded as a main structural building block of the sentence. The increasing importance of function words in linguistic theories went hand-in-hand with an increasingly abstract description of sentence structure. This shift provided many empirical and theoretical advantages.
First, sentence structure could now be divided into three functional domains: (1) a lexical domain around the verb, which establishes semantic relations between the main sentence elements; (2) a grammatical domain around the auxiliary, which establishes grammatical relations such as agreement (the auxiliary agrees in number and person with the subject: I am/She is/They are leaving.); (3) a discourse domain around the complementizer that, which links an embedded clause to a main clause (I know that this is true or I wonder whether this is true).
Second, differences between languages could be explained by how the function words, and the domains they define, were used. For instance, the so-called verb-second languages such as German, Middle English, Dutch, and Swedish move the verb to the complementizer domain, whereas languages such as English refrain from doing so. The word order of the equivalent German sentence Yesterday the rabbit saw the fox would thus be Yesterday saw the rabbit the fox. Differences between even unrelated languages were reduced to very basic principles.
Function words and lexical words are not sharply distinct categories but rather form a continuum. Certain classes of words can thus share features with both prototypical lexical words and prototypical function words. The English preposition is a case in point: some prepositions have lexical meaning, such as location (behind) and direction (toward); others have little meaning (of or to). Many are used to introduce sentences (after, for, like) and are therefore similar to prototypical function words, namely complementizers.
Grammatical meaning can be expressed in different ways. English uses independent auxiliaries to express present or past tense (I am leaving vs. I was leaving) but also inflects the verb for the same purpose (I think vs. I thought). Languages exhibit great variation along these lines: some languages express all grammatical meaning via independent function words and are called “analytic.” So-called synthetic languages, on the other hand, employ inflection and other markings on lexical words throughout.
This distinction between analytic and synthetic languages also represents a continuum, and languages can change in this respect over time. Old English made extensive use of grammatical markings on lexical words. English has lost much of this capability since then and introduced auxiliaries to fill the gap. In fact, the auxiliary will used to be a lexical verb in Old English, but it lost its meaning (“to want”) when it was recruited for expressing future tense. In modern theoretical approaches, which tend to focus more on underlying differences rather than surface variation, the distinction between analytic and synthetic languages becomes negligible.
In short, function words have little lexical meaning and no stress. In traditional grammars, they do not have their own projection or phrase, whereas in some modern approaches they do. They are very similar (and are historically related) to grammatical markers on lexical words.
ELLY VAN GELDEREN
Abney, Steven, “The English Noun Phrase in Its Sentential Aspect,” Ph.D. diss., Massachusetts Institute of Technology, 1987
Chomsky, Noam, The Minimalist Program, Cambridge, Massachusetts: MIT Press, 1995
Fukui, Naoki, “A Theory of Category Projection and Its Applications,” Ph.D. diss., Massachusetts Institute of Technology, 1986
Gelderen, Elly van, The Rise of Functional Categories, Amsterdam and Philadelphia: Benjamins, 1993
Lightfoot, David, Principles of Diachronic Syntax, Cambridge and New York: Cambridge University Press, 1979
Radford, Andrew, Syntactic Theory and the Structure of English, Cambridge and New York: Cambridge University Press, 1997
Radford, Andrew, Syntactic Theory and the Acquisition of English Syntax, Oxford and Cambridge, Massachusetts: Blackwell, 1990
Traugott, Elizabeth, and Berndt Heine, editors, Approaches to Grammaticalization, 2 vols., Amsterdam and Philadelphia: Benjamin, 1991