WHAT IS A FUNCTIONAL PERSPECTIVE?

THERE ARE at least five senses in which the word functional is commonly used by linguists. Our concern will only be with the last of these. However, so that we have some idea of what is not meant by this term in the present discussion, a quick review of all five meanings is in order.

FUNCTIONAL WITHIN A LINGUISTIC SYSTEM

Chomsky has remarked upon the multivocal uses of the term functional as it applies to linguistic analysis. Although noting that some linguists have looked upon functionalism as a doctrine which holds that “use of language influences its form” (1979:85), Chomsky himself uses the term rather differently in an early paper with George Miller (Miller and Chomsky, 1963). In a later summary he argues that

Thus, for Chomsky, the term functional is roughly equivalent to “well-designed,” and can be applied to the way in which elements of the grammar are put together.

This application of the notion of function to a whole linguistic system is reminiscent of one of the ways in which members of the Prague School understood the idea of function. An example is Martinet’s work on historical phonology (e.g., 1952, 1955). Using such notions as the principle of least effort, functional load, and push chains and drag chains, Martinet has argued that once phonological systems fall out of “balance” (e.g., equal number of voiced and voiceless stops, even distribution of stops and fricatives), they “strive” to regain a state of equilibrium. The subsequent discussion of causes of language change (“The Problem of Diversity,” below) will return to the Prague School’s systemic approach to language function.

THE SENTENTIAL CONTEXT

Another use of the term functional— and again one which interested the Prague School—derives from an analysis of individual sentences. Within the Prague School, this approach was initially developed by Vilém Mathesius (e.g., 1964) under the name Functional Sentence Perspective, or FSP, which considers how different languages (or synonymous sentences within a single language) express the same real-world experience through differential distribution of old and new information within the sentence. An example of FSP is the use of optional word order changes in an inflected language such as German or Czech to indicate what constitutes old and what constitutes new information in a sentence. Renewed interest in FSP is reflected in the work of Henry Kučera (e.g., Kucera and Cowper, 1976) and Susumo Kuno (e.g., 1972).

The term functional can be applied in other ways at the sentential level of analysis. Fillmore (1968b) speaks of his cases (Agent, Object, Instrument, Locative, etc.) as playing “roles” in the sentences in which they appear, and Dik (1978) intends the term functional to label semantic, syntactic, and pragmatic functions as they relate to the formation and utterance of a sentence.

DISCOURSE

It is in the domain of discourse that the idea of function is probably most familiar to linguists. In the discourse interpretation of function, we identify factors within the production and reception of a linguistic message and indicate which role those factors might serve in the act of communication.

By now it is customary to begin any discussion of language function qua discourse with the name Karl Bühler. This Austrian psychologist identified three distinct functions a language might serve, associating each with a factor in the discourse (1934):

Jakobson (1960) expanded this list to six pairs of factors and functions (“factors” are indicated in capital letters, “functions” in lower case):

Variations on the discursive theme appear in the work of Halliday (1970, 1973) and Hymes (1974).

The domain of these discussions of language function relating to discourse is the uttering of a message by an addresser to a receiver within some context. Of prime interest, especially in the work of Hymes, is how one determines (or predicts) which of these functions will dominate in a given conversation. Note, however, that these are not questions about how the language system got to be the way it is, but rather, given the language system, how it is used on a given occasion.

LANGUAGE ACQUISITION

A very different use of the term functional has emerged in the literature on language acquisition. The relevant domain here is that of semantic development: Which factors influence the order in which young children learn new words? Eve Clark (1973) has argued that semantic complexity of words, along with perceptual salience of the objects or properties to which these words refer, are the major factors influencing semantic acquisition. Katherine Nelson (1973) has taken a different stance, suggesting that the use of the objects being referred to is the all-important factor in early semantic development. Children first learn names for things that change (location, directionality), not for static things which are merely easy to perceive (color, shape). Other linguists (e.g., Greenfield and Smith, 1976; Bloom and Lahey, 1977) have assumed positions somewhere in between.

In discussions of function within the contemporary language acquisition literature, the major variables are the child and the object whose name must be learned. What is often missing is any notion of social function (e.g., the child early on learning the word more to get a bigger amount of something). The social and pragmatic origins of such high frequency words as more in the young child’s lexicon are sometimes acknowledged (e.g., Brown, 1973). However, until fairly recently, work in language acquisition has tended to retain a structuralist bent.¹

SOCIAL LINGUISTICS

Finally, there is the notion of function with which we will be concerned in this book: that of language being shaped by the social tasks for which speakers use it. The idea of form following function will need to be qualified with many caveats. Chief among them is the realization that a social explanation of linguistic shape can only be applied to a rather circumscribed set of linguistic phenomena. Subsequent sections of this chapter will investigate the kinds of linguistic phenomena which might belong to this set. First, though, the investigation of a functional perspective must be prefaced by considering linguistic diversity. Unless there are alternative shapes languages can choose from, there is no “stuff” to which one or another functional perspective can be applied.

THE PROBLEM OF DIVERSITY

Why are there so many different languages in the world, and why do they keep changing their shapes? These two simple questions lead us rather quickly into territory which is at once provocative and intellectually treacherous. Asking which shapes language assumes leads us to construct linguistic models whose appropriateness we can at least test with linguistic data. Asking why those shapes arise requires us to introduce a largely new set of variables, many of which go well beyond the linguistic data themselves. As David Hume established, causation of any sort is difficult (he would have said impossible) to prove; and causation in language is every bit as difficult to establish as causation in, say, history or economics.

Why, then, should we venture into a domain fraught with such difficulty? The answer lies in our reasons for thinking about language in the first place. I have argued that human language can be both a source of and a solution for human interpersonal problems. This does not necessarily imply that language users are aware of what they are doing with their language (or what their language is doing to them). Yet it does suggest, at least in the therapeutic aspects of language, that some linguistic processes serve identifiable ends. To recognize such purposiveness is not to imply that the linguistic process was necessary or that an alternative process could not have done the job as well. Languages often hobble along with inadequacies, such as the lack of a common gender pronoun in English to be used in place of he or she. Languages are also renowned for arriving at alternative solutions to the same problem. The word television is conveniently abbreviated to TV in the United States, but to telly in Britain. What is relevant is that a generally accepted common gender pronoun would find a ready niche in contemporary American English and that both the English and the Americans have worked out solutions for abbreviating the name of an extremely common item in their experience.

We shall be looking at linguistic diversity as the result of language changes; some of the purposes of these changes can be identified (or at least conjectured), others cannot. The discussions of spoken (chapter 5), written (chapter 6), and signed (chapter 7) systems examine in some detail the hypothesis that linguistic properties (relationships between signifié and signifiant, choice of modality, range of vocabulary or grammar) may be intimately associated with concomitant non-linguistic events. However, to understand why such arguments have received little attention in contemporary linguistic discussions, we need to look more closely at the notion of language as purposeful activity. Let us begin by way of an analogy.

BOAS AND THE ARMENIANS

In his intellectual biography of Franz Boas, Melville Herskovits describes a lawsuit in which Boas became involved:

Characteristically, Boas approached the problem with the same premise on which he based his work with American Indian societies and languages: that genetic differences are not to be confused with cultural or culturally induced differences. In the Armenian case, he distinguished between two groups of Armenians: those born in the United States and those who had emigrated from Europe. He measured the head size of members of each group and found a marked discrepancy, the explanation of which caused the case to be dismissed:

That is, immigrants had more rounded heads than American-born Armenians. This occurred because

The issue at the heart of this case has plagued us for a long time: How many of our human characteristics—both physical and mental-are biologically inherent, and how many are the result of interaction with the environment? The problem appears in many guises, the most common of which are “nature vs. nurture” or “heredity vs. environment.” Which side of the argument one advocates may have serious implications for theory and practice within the social sciences. Arguments for and against integration, justification or rejection of social welfare programs, as well as theories of education and of human language—all have been influenced by presuppositions about human malleability after birth.²

In linguistics, the question of malleability proves crucial to the study of language function. If we believe, as does Chomsky,³ that a significant portion of human language is innate, then we immediately restrict the range of variation which is possible within human language. By restricting the range of innateness to such metalinguistic considerations as the ability to learn language or the ability to determine the grammaticality of a simple utterance, we have not lost any of the flexibility relevant to the discussion of language varying according to its social purpose. Once we enter the realm of what Chomsky called “substantive universals” (1965), we have, in principle, added constraints to the actual language that speakers might produce.

There is a second way of looking at the question of what is inherent in human language qua human language and what is malleable. If there are parts of human language which are not innate, how do they come into being—in the individual, or in the linguistic community? This question leads to an even more difficult one: What kinds of change in language are possible?

CAUSES OF LANGUAGE CHANGE

Changes in language can happen for a host of reasons, not all of which are identifiable, and not all of which are interesting. No one really knows, for instance, why the First Germanic Consonant Shift came about, leaving us with English father in place of Latin pater, or Germanic bropar instead of Latin frater. (A number of hypotheses have been proposed over the years, but they have been rejected for lack of evidence. One such proposal suggested that the cold climate of northern Europe led to a greater huffing and puffing when speaking, causing stops to become fricatives. Another version claimed that the switch from stops to fricatives reflected the aggressive spirit of the Germanic people.) On the other hand, there are many changes whose origins we can easily document. We do know why English added the word sputnik to its general lexicon. However, the conditions are sufficiently straightforward as to arouse but little linguistic interest. And, of course, there are those changes which fall somewhere between these two extremes.

What types of changes can languages undergo? The literature on the question is rich and varied,⁴ but several concerns seem to stand out. Does the change originate within the language itself, or is its source external, coming either from another language, or from social factors affecting the members of the speech community in question? A second query deals with linguistic effects rather than causes: Does the change make for a more or less “unified” system than the prior state of affairs yielded? (Obviously, our notion of “unified” depends very much on the particular model of grammar we subscribe to.) Another sort of “effect” which change can bring about concerns the speech community: its power of social identification and its power of expression.

Figure 4.1 summarizes the sources—and outcomes—of change which human language may undergo. Obviously, many combinations of variables are possible. Changes initiating within a given language may yield increased balance or generality within some portion of the grammar (for example, English noun plurals). Alternatively, as Jakobson observed (1962), increased simplification in one portion of a grammar may yield imbalance elsewhere, as in the loss of nasals in some Salishan languages (see Ferguson, 1963). Change originating outside a language typically adds complexity to a grammar (often enforced by prescrip-tivism), though increasing generality may also result. Simplification occurs in the addition to English of a phonemic distinction between /š/ and /ž/, when /ž/ was borrowed from French. The addition of phonemic /ž/ made the voiced/voiceless distinction a general property of English obstruents. A case of external influence causing increased complexity is seen in the borrowing of certain noun + adjective phrases from French (e.g., notary public). These phrases complicate English syntax, since the normal ordering of substantives and modifiers in English is adjective + noun.

In considering outcomes of linguistic change vis-à-vis the speech community, we must introduce a distinction which will prove important in our attempts to understand what it means for a linguistic structure (or a linguistic change) to be functional (see “Complications,” below). At the least, we need to ask whether the linguistic construction itself is significant in interpersonal linguistic interaction, or whether social value is being attached to a fortuitously selected linguistic feature. The latter category has attracted much attention in the recent sociolinguistic literature by Labov and his colleagues (see Labov, 1972b). Several of the examples of both categories of change which are presented in figure 4.1 will be discussed in subsequent chapters.

Without underestimating the importance of language as a marker of social or economic identification (or aspiration), we must not forget that the particular linguistic construct which bears social prestige is fortuitous⁵ and that in most (if not all) instances the use of language as a social marker is itself fortuitous. Patterns of dress, or, as a colleague once quipped, the sporting of handlebar mustaches, might equally well serve to alert the general public to the individual’s social intent.

Fig. 4.1 Typology of Linguistic Change

The other type of communal variable, power of expression, is most interesting here because it involves a “necessary” relationship between the linguistic and the social variables initially identified in chapter 1 (figure l.1). Such a relationship is described as “necessary” not because every language used in a specified social context will invariably undergo change of the sort noted, but rather because the change noted is a logical outcome of the social conditions specified. If a change does occur, this is the one we predict. Thus, increased use of sign language logically leads to a reduction in iconicity (see chapter 7). This does not mean that all sign languages whose use increases become less iconic (prescriptive forces might prevent this). But it does mean that, empirically, we do find a correlation between use and arbitrariness—and that we would be surprised to find either iconicity increasing with use or decreasing with lack of use.

COMPLICATIONS: CLASSIFICATION AND METHODOLOGY

After wending our way through an assortment of types of language change, let us take a serious look at the hypothesis that the shape language assumes may result from nonlinguistic needs of language users. We have reached this point by eliminating whole classes of language change for which we make no functionalist claims, leaving us a more tractable (and plausible) set of linguistic phenomena in terms of which to examine the hypothesis.

WHEN IS A CHANGE SOCIALLY MOTIVATED?

How do we decide which specific linguistic changes might be functionally motivated? We might consider language change according to different linguistic levels (phonology, morphology and syntax, and semantics). While some levels of grammar never admit of functional analysis of the sort we are interested in, others might be very likely domains for functional explanation.

Consider structures (and the changes which created them) on the phonological level. The likelihood of finding functional explanations here is slight—there is nothing inherently “aggressive” about Germanic fricatives, any more than there is anything inherently “indecisive” about languages which distribute equal stress on all syllables. Such associations are after-the-fact notions which, under culturally appropriate circumstances, one might try “reading into” the respective phonologies, but the objective status of such “explanations” is indeed questionable.⁶

On the other hand, the domain of semantics seems particularly ripe for functional explanations. We add new words—or new meanings to old words—as the experiences we need to talk about expand. Twenty years ago, tripping was something one could do to a mechanical device (“Please stop tripping all those switches”), while today it is something one might do while with friends at a party. Meltdown is no longer only a verb plus a locative, describing what we do with a stick of butter or a ton of iron ore. Now it can also be a noun signifying the process whereby uranium pellets in a nuclear reactor liquefy, causing nuclear havoc (“There was a serious chance of meltdown at the Three Mile Island Nuclear Reactor”). Because language is conventional, it is still up to a linguistic community to determine the meaning of a particular word or phrase. In that sense, meaning is not inherent in a word or phrase. There are, however, precedents within the language which severely restrict the range of likely candidates. Meltdown is a logical name for what could potentially happen in a nuclear reactor, while the terms apple pie or Old Black Joe would not be.

The fact that convention enters into the change process in semantics should not lead us to question the functional base of many semantic constructs. Contrast this circumscribed conventionality in semantics with the open-ended choice speakers have in selecting phonological factors with which to imbue social import. The aspiration to middle-class status might be indicated just as simply by using (or not using) postvocalic /r/, by devoicing final consonants, or by raising one’s left eyebrow before initiating a conversation.

The remaining domain to examine is that of morphology and syntax. Like the search for functional explanation in semantics, functional inquiry in syntax has appeared promising to many, and yet, like arguments in phonology, is has borne little fruit. The writings of Karl Vossler and later of André Martinet exemplify this tradition—and its failures.

Vossler had hoped to demonstrate that grammatical changes between Classical Latin and Vulgar (particularly later Vulgar) Latin reflected changes in the cognitive and cultural style of the population. The Classical Latin passive construction, so he argues, vanished from Vulgar Latin because of a change in the conceptions of personal feeling and personal agency. Speakers began

Vossler does concede that the link between cognitive, and grammatical shift is not strictly causal:

nonetheless,

Vossler draws upon similar armchair social anthropology to explain the loss of the overtly marked future tense from Classical Latin:

In the case of future tense markers, Vossler does hint that cognitive changes hastened the decline of the grammatical forms:

The Romance languages later replaced the old future with “several other fresher and stronger means of expressing the new meaning” (ibid.).

A more recent proponent of a similar causal line of thinking is Martinet, whose general thesis is that

Martinet attempts to illustrate his bold hypothesis with two examples. First, he argues that the evolution of Indo-European from an ergative to a nominative-accusative system⁷ resulted from advances in the cultural and economic level of the Indo-European people (1962:72; 1958). Similarly, Martinet claims that the Latin case system was superseded in modern Romance languages by prepositional constructions because the restricted case system was “no longer capable of taking care of the expression of all the relations needed in Roman society” (1962:138).

Martinet’s theses, like Vosslers, are intriguing. There is even the possibility that they are correct. Unfortunately, there seems to be no way of knowing. Without evidence (and it is unclear what such evidence might look like), they have no more credibility than a thesis that English makes heavy use of truncated passives (“The window was broken”) because Anglo-Saxon peoples avoid taking responsibility —another logically possible thesis, but not one with much tangible support.

This survey of functional arguments on the phonological, semantic, and grammatical levels leaves us having learned very little. On the face of things, we would need to conclude that our commonsensical intuitions about inappropriateness of functional explanations in phonology (and appropriateness in semantics) are correct, and that the domain of grammar is not much different from that of phonology. But is the analysis accurate? Or perhaps, asking the question differently, will we find that the functionalist hypothesis is indeed supported?

An alternative approach is to ask whether some principle (or principles) of organization other than linguistic level might be associated with functional factors. Consider a principle such as redundancy. When applied to the use of sounds, redundancy might take the form of reduplication. Phonological reduplication appears in a variety of contexts; in many languages, it is a way of deriving new grammatical forms. Indo-European languages sometimes formed perfective verbs through reduplication, as in the Latin

tango = “I touch”

tetegi = “I touched”

In Tagalog, reduplication is used to indicate an agent:

['su:lat]           = “a writing”

[su:-'su:lat]     = “one who will write”

['ga:mit]         = “thing of use”

[ga:-'ga:mit]   = “one who will use”

[see Bloomfield, 1933:218]

Reduplication is also commonly used in talking to children (the reduction of stomach to tum-tum), and children themselves frequently reduplicate syllables they are learning to pronounce. Hence, daddy becomes dada, and water might become wawa.

In the case of the Indo-European perfective or the Tagalog agentive, there is no obvious reason why reduplication (rather than some other grammatical device) bears the syntactic burden of forming a new grammatical category. However, in language used with and by children, a functional explanation is plausible. We know that adults speaking to children simplify their speech in a number of predictable ways: speaking slowly, reducing the range of normal vocabulary, exaggerating intonation contours.⁸ We also have good reason to believe that for children just developing a phonological repertoire, reduplicated sequences (e.g., papa) are simpler to produce than sequences with dissimilar consonants and vowels (e.g., pagu).⁹ Lexical redundancy at the grammatical level is another example of redundancy in language. Its use in pidgin languages will be studied in chapter 5 in an attempt to demonstrate a functional basis for this linguistic principle of organization.

In addition to looking at organizational principles (rather than at linguistic levels) as a way of probing the functionalist hypothesis, we might also ask which domains of language would be most likely to yield evidence for functionalist arguments. Vossler and Martinet ventured functionalist hypotheses on the basis of the language of adult speech communities. In each instance, it is difficult even to determine what might constitute supporting evidence for a functional hypothesis. On the other hand, when we look at examples of children learning to talk, adults talking to children, or pidgin languages, prospects for productive exploration are far more promising. It seems likely, therefore, that while linguistic constructions which are part of everyday speech may indeed have functional origins, it may be easier to pinpoint such origins by looking at language use which does not involve the entire speech community.

We must consider the possibility that a feature of language which was first introduced under functional circumstances may no longer be functionally motivated. Napoleon is purported to have introduced buttons on the ends of coat sleeves to stop his soldiers from wiping their noses on their sleeves. Today’s haberdashers have no such motives as they continue the tradition. Within the linguistic domain, the word penknife was a logical addition to our vocabulary when writing was still done with quills which needed to be sharpened, and we began speaking of the sun rising and setting at a time when, empirically, we believed such to be the case. Today, quills are no longer used as writing instruments, and our views on astronomy have changed; our locutions have not.

Another reason for placing much of our emphasis on language use which either is not fully developed or is restricted to small segments of a language community is strictly methodological: it is often easier to understand a phenomenon piecemeal than in its entirety. By learning how to make functional arguments for limited uses of language, we discover how to apply our findings to the speech community at large.

ARE ALL INFORMANTS EQUALLY INFORMATIVE?

Once we have identified the domain of language we wish to subject to a functional analysis, how do we collect data? More specifically, What sorts of informants do we need? How many? With what specific qualifications? Examined in which particular circumstances?

The study of language requires the use of a selective sampling of the data. Only a fraction of the potential utterances in a language have actually been uttered, so it is fruitless to attempt to amass a “complete” corpus. Moreover, by the time one managed to poll roughly 200 million speakers of English in the United States alone, one could not be said to be studying the “same” language in the first and the last interviews.

But how do we select which informants to study? An answer to this question is apt to be laden with presuppositions about how much of language is “communal” in a speech community, and how much of human linguistic abilities are biologically “built in” (like bookshelves or a fireplace?) at birth. Such presuppositions may themselves be products of social circumstances which appear, at first glance, completely unrelated to theories of linguistic abilities.

In chapter 1, the notions of langue (Saussure) and competence (Chomsky) were introduced. While Saussure’s langue refers to the linguistic knowledge of an entire speech community (which is manifested, through parole, to greater or lesser degrees in different individuals), Chomsky’s competence resides in each member of the linguistic group. In both instances, it is assumed that there is a body of linguistic conventions that characterize the language community. In Chomsky’s case, this body of information is taken to reside equally in all language users (Saussure is ambiguous on this point). Thus is born Chomsky’s “ideal speaker-listener”:. “Linguistic theory is concerned primarily with an ideal speaker-listener, in a completely homogeneous speech-community, who knows its language perfectly” (1965:3). Chomsky is well aware that in reality, speech communities are not homogeneous and that no single individual has perfect linguistic knowledge (1979:55). In practice, though, linguistic inquiry, especially in the United States, has increasingly proceeded as if his idealization held true. There has been a further tendency to assume that if such “complete” linguistic data can be found in a single individual, linguistic data need not be elicited in the context in which that language would spontaneously be used. Perfect linguistic knowledge evidenced while participating in a syntax seminar should therefore be no different from that same perfect knowledge displayed in talking to one’s in-laws or in shooting pool.

These assumptions of decontextualization and equality among language users have dominated much of the structuralist and generativist traditions. How did these assumptions arise? Have presuppositions about data collection influenced linguists’ attitudes toward the efficacy of searching for functional explanations of linguistic patterns?

The Triumph of Structuralism

In light of the predominant structural biases of American linguists for over forty years (roughly, from 1930 to 1970), the early twentieth-century emphasis on studying language within the context of its use seems particularly refreshing. Franz Boas, pioneering the study of both anthropology and linguistics in America, repeatedly argued for the necessity of studying language naturalistically in order to get the most accurate measure possible of the range of constructions used and the meanings attributed to them. In the words of Robert Lowie, “Boas must be understood, first of all, as a fieldworker” (1937:131). Not surprisingly, Boas was incensed at Bronislaw Malinowski’s implication that the doctrine of functionalism constituted a British invention. In a review of Malinowski’s Crime and Custom in Savage Society, Boas writes: “Dr. Malinowski has a strange impression of what modern anthropology is. He accuses modern anthropologists of a complete disregard of the actualities of life.” To the contrary, Boas boasts, Malinowski “will find that the general approach of the modern American anthropologists is quite similar to his own” (cited by Herskovitz, 1953: 67). Were Malinowski’s impressions wholly unfounded? Possibly not. The Boasian mystique of the importance of fieldwork seems based, at least partly, in myth. Leslie White (1963) has calculated that of the forty-odd years Boas worked as an ethnographer, only thirty-three and a half months were spent in the field. Where did Boas gather his materials? Largely in major cities. For instance, he writes:

Had Boas (or subsequently, his students) even wished to gather information on indigenous languages from the proverbial Indian seated before his tepee, opportunities diminished as the century wore on. Surviving tribes were relocated on reservations, where traditional activities—and linguistic contexts—no longer were found. Moreover, a number of Indians gravitated to the cities, where job prospects were brighter. Many of these strangers to modernization made their way to the YMCAs. In fact, for several decades, especially in New York, linguistics graduate students searching for an informant frequented YMCAs to make their initial contacts. This phenomenon is recent enough that just ten years ago, a student from Columbia offered to introduce me at the next YMCA open house if I would only agree to work on the language of one of the YMCA regulars.

The results of this decontextualization were predictable. Sociolin-guists have shown that individuals’ perceptions of what they might say in a given circumstance often bear little resemblance to the language they would actually use spontaneously. A Yana Indian sitting in Horn and Hardart could not be expected to tell the linguist how he would speak at a hunting ritual, or what differences there are between, say, male and female language. He could, however, give enough samples of language to enable the linguist to deduce the phonemic inventory, the basic morphology, some grammatical rules, and some of the lexicon of the language. In the study of indigenous languages, form took precedence over function as a matter of necessity. In ensuing years, however, it might be argued that necessity gave way to virtue; linguists came to view structure as the only aspect of language worth examining. At the turn of the century, Saussure had dismissed the analysis of language use (parole) as uninteresting. In striking parallel, decades later, Chomsky argued for the study of linguistic competence. Linguistic performance—that is, language as actually used in real social contexts—was rejected as lying beyond the bounds of proper linguistic theory.

Equality among Language Users

The removal of the native informant from his social milieu contributed to a dramatic shift from the study of structure within the context of situation (Malinowski, 1923) to the study of structure alone. In addition, changes in the demographic status of American Indians resulted in the linguist’s increasing reliance upon a single informant for data on a particular language.

In the early years, the use of a single informant was judged to be an unfortunate methodological necessity. The classical case in the literature is Mary Haas’s study of Tunica, a language once spoken in the area of what is now Louisiana (1940). By the early 1930s, when Haas began her study, there was only one individual, Sesostrie Youchigant, who spoke Tunica “with any degree of fluency” (ibid., p.9). (Youchigant spoke mainly Louisiana French and English.) No other speakers had sufficient knowledge of the language to converse with him. In fact, Youchigant had had “no occasion to converse in Tunica since the death of his mother in 1915, and even before her death, he preferred to speak French to her” (ibid., p. 10). Using data from Youchigant, Haas produced a grammar, a dictionary, and a set of texts in Tunica, despite the fact that at the time the data were collected Tunica “serv[ed] no sociological function whatsoever” (ibid.). Even she admits that

This tradition of relying on one informant—with all its inherent dangers—has persisted in contemporary grammatical analysis. Underlying Chomsky’s hypostatized ideal speaker-listener is the belief in the inherent linguistic equality of all language users. Consequently, any speaker of language X should be as good an informant as any other, whether or not additional informants are available.¹⁰

The issue of linguistic equality in America is complex because it tends to be confused with the idea of sameness. The original Boasian tradition maintained the equality of all linguistic variants, while the exigencies of the ethnographic situation made it convenient to equate “equality” with “homogeneity.”

The same confusion arising from the study of American Indian languages has its parallel among Indo-European populations (and their languages). On the one hand, the United States has insisted that diversity is to be tolerated, while in actuality, conformity is typically rewarded. The preponderance of ethnic and linguistic diversity in this country of immigrants caused some Americans to question whether all immigrants (and all of their languages) were to be considered equal in sophistication and worthy of recognition. The tacit but powerful assumption by the majority of the country that equality lies in sameness has been evident in the strong governmental and social pressure to abandon native languages in favor of English. Government-sponsored bilingual education programs are a very recent phenomenon (it was generally assumed that bilingualism lowered a child’s intellectual abilities). Even dialectal variation broadcast on radio and television was long seen as detrimental to economic and social progress.

This governmental and attitudinal stance against linguistic diversity was directly at odds with the Boasian insistence on equality of speakers and languages. Despite the efforts of the government and the education system to eliminate linguistic diversity, the linguistics profession actively recorded dialect diversity in America. The best known of the American dialect studies, the Linguistic Atlas of New England (1939-1943), was directed by Hans Kurath. On a more popular level, Henry Lee Smith used his expertise in dialectology as host of a radio program entitled “Where Are You From?” His job was to identify the place of residence of people whom he interviewed. The Providence Evening Bulletin writes:

Not until the 1960s, however, did linguists’ voices begin to be heard by school committees, funding agencies, and the courts. Echoing Boas decades earlier, Labov (1972a) argued that Black English displays the same level of sophistication and regularity as standard (white) American English. Through the efforts of Labov and his colleagues, Black English has become the subject of serious linguistic study. In some instances it has been adopted as a teaching vernacular in schools in which the children’s native dialect is Black English. Similarly, bilingual programs have been federally mandated, continuing the prototypically American belief in the equality of peoples, and (linguistically) in the equality of languages and language users.

Implications for Functional Investigations

The first lesson to be learned from looking at contemporary presuppositions about data collection is simple enough: idealization may be useful for armchair linguistics but is not as well suited to an analysis of language as people actually use it. What is more, a linguistic model based on idealizations has less need for functional explorations than does one which is designed to study language as a response to social needs. Therefore, we should not be surprised at the lack of functional investigations undertaken in the American linguistic tradition, for Chomskians, like their immediate predecessors, were not concerned with the sorts of questions to which functional inquiries might yield answers.

The preamble to this book is now complete. I began by looking (chapter 1) at language as a source—and a solution—to problems of human social interaction. Stepping back from the problem in order to sharpen our perspective, I explored what it means to use language to represent our experiences to others (chapter 2), and what we mean by “language” in the first place (chapter 3). Since I argued up to this point that linguistic representation is strongly governed by human purpose, I then looked more closely at the functionalist framework in terms of which language use, acquisition, and change might be studied and explained (chapter 4).

It is the thesis of this study that the same types of factors that influence the shape spoken language assumes also operate in written and gestural linguistic representation. Let us now examine this proposal in detail.