FACTORING OUT SPEECH FROM OTHER AUDITORY REPRESENTATION

BECAUSE OF ITS familiarity, human speech might appear to be the easiest of the communicative representations to distinguish from its nonlinguistic counterpart. In practice, the line between language and nonlanguage turns out to be as difficult to draw for auditory as for visual representation. Let us see why this is so.

A first step in factoring out speech is to consider the means by which speech is normally produced: the human vocal apparatus. If we look at production rather than perception, we can rule out the clapping of hands, drum “languages,” and the like.¹ By adding a further restriction that the message be volitional, we eliminate sneezes, the chattering of teeth, and most coughs. The further criterion of “productivity” (however ill-defined the term may be—see chapter 3) lets us dismiss stylized coughs (e.g., a cough used to catch someone’s attention) and perhaps whistling.²

The most difficult part of the definitional process, though, is deciding whether the vocal features typically called suprasegmentals are part of language or not. Supersegmentals include accent, voice quality, pitch, tonality, duration. As a number of writers have noted (e.g., Abercrombie, 1972; Crystal, 1974), linguists are not consistent as to which suprasegmentals they include as a formal part of language. Part of the difficulty stems from the fact that a feature which regularly serves to distinguish meaning in one language (e.g., tone in Mandarin) may not in another; thus a unilateral judgment cannot be made as to whether a verbal production is “linguistic.”

This point is well illustrated by examining the way in which a single type of semantic information is expressed in different languages. David Abercrombie notes Jules Henry’s (1936) findings that

And Kenneth Pike (1967) observes that there are languages in which speech and gesture interact in what to us might seem peculiar ways. In Papago, for instance, to say “and they are just dropping off to sleep” requires a nodding gesture along with the spoken č?abhab aš ?i. In Mazatec, the object slot can be filled either by a verbal element or by a nonverbal event in immediate context.

Consider the use of the suprasegmental duration to make semantic distinctions in a language in which duration is not ordinarily distinctive. Must our definition of what is properly “linguistic” therefore vary even within a single linguistic community? A famous case in the language acquisition literature (Velten, 1943) is Joan Velten’s use of vowel length (which is nonphonemic in English) to distinguish between words (in the adult model) such as bad (Joan’s [ba:t]) and back (Joan’s [bat]). Another case involving duration occurs in Bengali, in which villagers use vowel duration to indicate distance. If a location is nearby, an appropriate response to the question “Where is X?” is [hu:-i], accompanied by a pointing finger, but if the place is far, the response becomes [hu:-i] (plus more emphatic gesticulation), the length of the vowel being a function of the distance to the place or object in question.³

Leaving aside the question of which verbal productions are to be considered “language,” we must come to terms with a problem that the model of representation excerpted at the head of this chapter wholly ignores: the so-called paralinguistic dimension of human communication. The first distinction in this representational model (between auditory and visual representation) implies that spoken language is produced and received exclusively through the auditory-vocal channel. By this token, telephone conversations should be indistinguishable from face-to-face conversations—and they are not. Rather, face-to-face conversation is an auditory message in the context of associated body postures, facial expressions, choices of distances between interlocutors, and the like. Linguists have traditionally dismissed kinesic and proxemic behavior as communicative but nonlinguistic. Such a restriction can be maintained if, a priori, we restrict face-to-face spoken language to aurally perceived messages alone. However, looking back at the general criteria established in chapter 3 for human language (having a finite lexicon, having a finite set of collocation rules, and combining lexical items grammatically to yield a potentially infinite set of wellformed constructions), it is not clear that kinesic and proxemic behavior should be eliminated from the domain of language. Continuing research on body movement and position (e.g., Birdwhistell, 1970; Scheflen, 1972; Watson, 1970) suggests that paralinguistic behavior satisfies at least the first two of these conditions. We still have comparatively little understanding of how this kinesic and proxemic communication interacts with spoken language. It is clear, though, that a full understanding of human speech as face-to-face auditory representation cannot be complete without the other information that is exchanged in a face-to-face encounter.

CLASSIFICATION OF SPOKEN LANGUAGES

Spoken languages have most commonly been grouped together on the basis of structural similarities—shared vocabulary, similarity of vowel repertoires, presence or absence of tones. It has generally been argued that likeness between languages may be viewed in one of three ways: genetic, areal, or typological.

STRUCTURAL GROUPINGS: GENETIC, AREAL, TYPOLOGICAL

Genetic Classification

Genetic classification groups languages on the basis of their historical connections with one another. Drawing on metaphors in both botany and human kinship, genetic classifications speak of family trees (what August Schleicher labeled the Stammbaum theory of languages) that branch off into divisions and subdivisions. Parent languages give birth to daughter languages, which are themselves sisters of one another. Figure 5.1 illustrates the best known of the genetic classifications, that of the Indo-European language family. There is still much debate among historical linguists as to the details of the branchings, but the idea that these languages are historically related is generally accepted.

Through the years, serious objections have been raised to the genetic model as a comprehensive way of describing relationships between languages. As with any metaphor, there are some properties which the literal and metaphoric terms hold in common, and some which they do not. (If all properties were analogous, the two objects being described would have to be identical.) The analogy between groups of languages and biological organisms assumes, among other things, homogeneity up to each new branching point on the tree. In biology, we know that conditions of propagation place severe restrictions on which organisms can be considered members of the same species. In the case of human languages, however, such homogeneity is neither necessary nor empirically documented. In nature, the proverbial lion can lie down with the lamb without one having a biological effect on the other because of their biological incompatibility. If genetically unrelated linguistic groups are in close contact, however, their languages tend to rub off on one another. This observation, at least in part, led linguists to supplement genetic classification with a second category.

Areal Classification

In an areal classification, we look for linguistic features held in common across languages that are not genetically related but whose speakers reside in geographically contiguous areas. (For languages which are genetically related, an areal classification might reveal linguistic similarities that have arisen since the languages diverged historically.) Language contact is assumed to be responsible for the spread of these features from one language to another.

Fig. 5.1 Partial Genetic Classification of Indo-European Languages

A major difficulty in using areal classification lies in determining whether the feature in question was actually borrowed from one language into another or whether it developed spontaneously in each. Take the case of the development of articles (the, a) or of compound tenses in western European languages: Lehmann states that because “we find these characteristics spread through mutually unintelligible languages, such as early French, pre-Old High German, and pre-Old English” we can conclude that “speakers of these languages were in close contact, that they belonged to a common ‘speech area’ “(1973: 233). There is, however, an alternative hypothesis: that these developments occurred independently. The common (genetic) western European syntax from which early French, pre-Old High German, and pre-Old English derived may have made these languages ripe for these syntactic developments, although the actual developments were initiated individually in the three separate languages. Or consider the case of vowel length in Czech. Genetically, it can be argued that the development of phonemically distinctive length is predictable from the rest of the Slavic sound system of which Czech is a part. However, it can also be maintained that the development would not have taken place if speakers of Czech were not geographic neighbors of speakers of Hungarian, a non-Indo-European language in which vowel length happens to be phonemic (see Greenberg, 1968:123).

Typological Classification

The indeterminacy problem in areal classification (as well as the problems of incompleteness in genetic classifications) might be solved by an altogether different approach. Rather than using history or geography as the relevant features with which to correlate linguistic similarities, we can look at the linguistic features themselves.⁴ In some instances, one ends up with a set of languages that also would have cohered under a genetic or areal qualification; usually, however, we must juxtapose languages that have neither temporal nor geographic ties. Consider, for example, a typology of languages that use phonemic tone distinctions to mark differences in meaning—included here would be Chinese, Mazatec (spoken in Mexico), and Ewe (an African language). Neither history nor geography can be invoked to explain the common phonological property of all three languages.

It is always possible that a typological analysis will reveal larger numbers of languages sharing a given feature than initially anticipated. In fact, his work on typological classification led Joseph Greenberg to study language universals. In his typological research, Greenberg often found that the feature in question was present in an overwhelming number of languages that he was examining; when it was absent, he could often predict its absence.⁵

CLASSIFICATION BY USER SUBGROUPS

Having looked at the three most common ways in which languages are classified, let us step back and ask what we can—or cannot—learn about language from these sorts of analysis. One obvious advantage is that linguistic classification helps us to get some sense of the range of constructions possible (or at least evidenced) in human languages. If we had had to look at every language individually, rather than as a member of a class, generalization would be difficult.

This sort of classification also gives us solid materials for considering how different languages can accomplish similar purposes by alternative structural means. We know, for instance, that agglutinating languages such as Turkish or Chinook string together large numbers of distinct morphemes to form a single word; isolating languages such as Chinese, and to some extent English, typically encode separate units of meaning in separate words. Thus, the Chinook word inialudam roughly translates into English as the seven word sentence “I came to give it to her.” Sapir identifies the following elements in the Chinook example:

  i- recently past time

-n- the pronominal subject“I”

-i- the pronominal object “it”

-a- the second pronominal object “her”

-1- a prepositional element indicating that the preceding pronominal prefix is to be understood as an indirect object

-u- movement away from the speaker

-d- radical “to give”

-am modifies the verbal content by adding to the notion conveyed by the radical element that of “arriving” or “going (or coming) for that particular purpose”

[1921:70]

In the English translation, the meaning elements and the relationships between them are largely expressed through syntactic linkage of separate words.

Is there any reason to believe that people who speak an agglutinating language are culturally different from those who speak isolating languages? Suppose that we hypothesized speakers of Chinook to be abrupt people, as evidenced by the fact that their language compresses many elements of meaning into a single word. To substantiate such a hypothesis, we would need to establish that all communities that speak agglutinating languages share the same cultural trait. (Such evidence is not likely to be forthcoming.) Thus, one of the major values of typological analysis is that it prevents us from pursuing irresponsible hypotheses by forcing us to examine such causal hypotheses in light of other languages belonging to the same typological group.

Are there explanations which cannot be drawn from data classified along structural lines? Can genealogical, areal, or typological classifications tell us, for instance, why languages in frequent use develop certain structural characteristics, while those used in more circumscribed contexts develop others? Can these classifications explain why adults who are ostensibly speakers of the same language often cannot understand one another? To find the answers, we need to think about classification of languages in different ways. Rather than taking constructional criteria alone as the basis of classification, we should consider social criteria as well: Is the language used as the basic means of communication between members of a social community, or is its use restricted to a subset of that community? If the use is limited, on what basis does an individual become a member of that subset of speakers? Is the language used in addressing members of other social communities? How much do members of one social community have to say to members of the other group? These sorts of questions are the stuff of which “social linguistics” is made; they also motivate change in “power of expression” (both of which were discussed in chapter 4).

These questions can be formulated with respect to a single speech community or used to contrast language types. The remainder of this chapter will examine the use of socially based classifications, both within and across languages. The first phenomenon is familiar to all of us—the issue of sublanguages. The second phenomenon, exemplifying a cross-societal perspective, concerns the development of “interface” languages, that is, languages that emerge when speakers of mutually unintelligible languages come in contact.

SUBLANGUAGES

As I have previously observed, language communities are heterogeneous; not all speakers of the “same” language understand each other. The reason for this has to do with the reasons we use language in the first place. If we assume that we use (and learn) language to talk about the things we experience, it should not be surprising that people who have different experiences will exhibit differences in their language.

One way of discussing experience is in terms of activity or work. What sort of things do we do? Do these differences in our activities show up as differences in our languages?

Ernst Fischer suggests that it is because humans do work that they have something to talk about. In human evolution,

In fact, it is work that distinguishes between human and animal communication:

Anthropological studies of so-called primitive societies suggest that our earliest ancestors shared most of their activities; idiosyncratic differences between individuals (shyness, self-assurance, bombast) may have yielded differences in speech style or rhetorical abilities, but underlying lexical and grammatical knowledge was probably similar across speakers. Hunting and gathering societies and agricultural societies illustrate this kind of productive homogeneity. While women may engage in different work than men, and a few individuals may be singled out as having distinctive functions (e.g., a shaman or a chief), the division of labor is not so specialized that a linguistic description made by a yam digger will not be understood by someone who mainly repairs canoes.

With the development of religion, basic technology, and social stratification, human society began to see a division of labor. As this division increased, the linguistic needs of society—to paraphrase Fischer, the words people used to talk about their work—were no longer the same for everyone. Instead, there began a development of specialized sublanguages.

Fig. 5.2 Drawing by S. Harris; (c) 1973 The New Yorker Magazine, Inc. 16 July 1973, p. 24.

Sublanguages are languages (or portions of languages) containing distinct lexicon or syntax and used by a subpopulation (of a larger community) that engages in specialized activities in which the majority of the population does not participate. A sailor, for example, needs to distinguish between spinnaker, mainsail, and jib; scientists must know how to interpret formulae (see figure 5.2); and a printer must know the difference between hot and cold type. As increased diversification and specialization of language use continues, however, language begins to fail in its more general function of creating and maintaining bonds between all members of the social system (see figure 5.3).

There are three specific functions that sublanguages serve. Perhaps the most important of these is the introduction of precision to the way one converses with ones fellow workers. If I am ere wing on your yacht, and you simply ask me to “let out the sail,” I have no idea whether to let out the mainsail or the jib.

Fig. 5.3 Drawing by Dana Fradon; © 1975 The New Yorker Magazine, Inc. 21 July 1975, p. 31.

Intimately connected with the function of precision is that of brevity. In the sailing context you might say, ‘Tm about to let out the larger of the two triangular sails, which is attached to the center mast, so watch out that you don’t get hit on the head when it swings over to the other side of the boat.” However, by the time you finish your instructions, the boom may have already knocked me over. An abbreviated name for an object or a process—like coming about— not only saves effort but reduces the confusion which periphrasis is likely to cause. This kind of abbreviation is no different in kind from the abbreviations entire communities adopt for objects or experiences to which they need to refer often. Most of the time we say TV, not television, and 1RS, not Internal Revenue Service, precisely because we frequently refer to televisions and taxes.

There is a third, derivative use of sublanguages as well: exclusion. Sublanguages can be used to exclude outsiders from participating in a conversation. Addeo and Berger cite an example of stockbrokers engaging in an exclusive form of “Jobspeak” for the benefit of a friend who has no idea what they are talking about:

Sometimes a dialect of ordinary language unknown to one’s interlocutors can serve the same function of exclusion typically filled by sublanguages. On a bus in southern France several years ago, a friend and I were being harassed by several young locals who obviously knew some English. Since there was no place we could move to, I decided to continue my discussion with my companion in a deep (though quite pseudo) Southern drawl. This tactic promptly ended the disturbance, since the youths trying to insinuate their way into our conversation clearly did not have knowledge of English which extended beyond standard English phonology.

Sublanguages, then, are systematic forms of lexicon and syntax that arise paradigmatically to enable a group of people who engage in the same activity to do so more productively than would be possible using only the language of the general speech community. The term paradigmatic distinguishes specialized languages that enable their users to increase productive efficiency from languages that, while distinctive, are partially or wholly arbitrary. In the latter, distinctive linguistic features may be created by a subgroup of the population that is socially or productively distinct, but for which much of the new language does not serve a function in the paradigmatic sense defined above. Rather, the primary function of the new language is to set its users apart from the rest of the population. In such cases, the sublanguage itself comes to function less as a tool of social production and more as a tool of social identification. A continuum of sublanguages thus emerges, ranging from those whose structures are determined by their functions, to those whose structures are fortuitously selected and then endowed with social significance.

ACTIVITY, SCIENCE, AND PSEUDOSCIENCE

I have stated that sublanguages are used, paradigmatically, by groups of people to assist them in carrying out a specialized activity. What kind of activity do I have in mind? Earlier, we mentioned activity and work in the same breath. And indeed jobs and professions of various sorts typically have specialized vocabularies or syntactic constructions associated with them.⁶Activity or work can, however, describe any undertaking which evokes or entails linguistic labels. Sports and hobbies are as full of specialized lexicons and syntax as are activities designed to produce profits. Baseball, football, or tennis are obvious examples (admittedly all of these sports have become “professionalized,” but the sublanguages associated with them are as necessary in amateur circles as among the pros). Tight end or fifteen-love are not part of the standard vocabulary of much of the English speaking world. To those who do not follow baseball, a phrase such as “Tidrow walked Yastrzemski” might lead one to surmise that Yastrzemski was a family pet, rather than a batting ace of the Boston Red Sox.

One of the domains in which sublanguages have been most productively developed is science and its concomitant technology. As in other types of activity, sublanguages in science may eventually change the very syntax of ordinary language in order to achieve the degree of precision necessary. This change may entail the creation of a new grammatical construction for a known activity or the transformation of an existing grammatical category to express a new conception of the constitution of the physical world. A famous case in the history of modern science is the development of the concept (and word) momentum.

The example is clearly set forth by Bhattacharya:

The development of the word momentum illustrates how a specialized activity (in this case, the need to talk about moving bodies) motivates a new use of language.. To the extent that the broader population finds cause to talk about the movement of bodies in space, the word may pass from the specialist’s sublanguage into everyday parlance. In this way, momentum has become part of the vocabulary of the general populace, though its meaning is not as precise as when used technically by physicists. Other words from scientific sublanguages such as quarks or pi mesons have remained in the domain of specialists.

The effect of scientific sublanguages on everyday language has been at least as important as a means of social identification as it has been for increasing power of expression. With the growing prestige of the scientific community, especially after the Second World War, language that sounds scientific has acquired high social value. A pseudoscientific use of language has resulted that, in terms of the activity of the speakers, serves no direct function. Its indirect function, though, is to confer a sense of authority upon what is being said. Often the effect is strained (though still intelligible):

At other times, though, the attempt to sound “scientific” results in semantic gibberish. This phenomenon is especially prevalent in college term papers:

From an undergraduate sociology paper:

They discovered that both partners are lower in marriage happiness when the woman is working out of economic necessity rather than choice and that both husbands and wives are lower in tensions and high in socioability [sic] if the wife chooses the labor market over the housework, [emphasis added]

Can “marriage happiness” be measured like the Dow-Jones average?

From a master’s thesis on language development:

Slobin proposes the observation that . . . [emphasis added]

Presumably one proposes theories, one makes observations.

From a doctoral dissertation on the history of English:

If . . . one finds zero instances of a form century after century, we may then say that such a form was for all practical purposes not used—or at least not used this side of the point of diminishing returns. [emphasis added]

Can grammatical constructions really be subjected to a cost-benefit analysis?

It might be argued that all these examples are from written, not spoken language and therefore inappropriate in the present discussion. However, to the degree that writing is a representation of spoken language (see chapter 6), we expect to find phenomena such as pseudo-scientific jargon in both written and spoken language roughly comparable, provided that the audience (e.g., a college professor) is held constant. Casual observation of classroom conversation supports this hypothesis. Admittedly, writing tends to be more formal than speech (again, see chapter 6), and it is commonly assumed that formality calls for weighty words. Nevertheless, in formal spoken contexts (e.g., court testimony), the degree of formality—and susceptibility to pseudoscientific jargon—typically approaches that characteristic of writing.

“WIPE THE BUGS OFF YOUR PLATES”: TRANSITIONAL CASES⁷

Pseudoscientific language indicates that the primary function of a sublanguage may derive more from its social impact than from its semantic import. There are, in fact, sublanguages expressly created to combine minimal semantic reference with maximal social identification, where such identification implies exclusion of outsiders. High school slangs typically function this way, as does CB radio language.

The character of CB language has changed rather dramatically since its inception in the late 1940s and ‘50s, when it was modeled after military codes developed for use on two-way radios in World War I. Codes must be precise, brief, and exclusive— the same three factors essential to paradigmatic sublanguages. In the early use of the citizens band by police, rescue personnel, or ham radio operators, the first two traits were especially important. CB language frequently needs to be more precise than face-to-face spoken communication because of problems inherent in radio transmission; messages can be lost with a weak or fuzzy signal. Consequently, a high degree of redundancy is used to increase the chances of the message being received correctly. Single letters are sometimes encoded as words (e.g., using Baker to indicate the letter B), and there is a high degree of syntactic redundancy, which strongly resembles the syntactic redundancy in pidgin languages (see “Trade Jargons and Pidgins,” below). In both instances, the sender has reason to question whether the interlocutor will comprehend the message transmitted. With CB, the problem may be in the physical transmission, while with pidgin languages, there may be structural uncertainty. (How much of the language does the interlocutor understand?)

Counteracting the high use of redundancy for precision is the feature of brevity, which, in its early days, was functional in transmitting emergency information. Today, the explosion in the number of CBers trying to use the same channel has made brevity a practical necessity. The resulting use of initials to abbreviate the names of familiar places (e.g., Bronx River Parkway = BRP) is the exact opposite of using full words to be sure a single letter is understood.

The most interesting sociolinguistic development is the use of CB language for social exclusion and social identification. There is a popular belief that its use “protects” the user from being understood by the police—a curious assumption, especially since it was in part modeled after codes used by police. Middle class users of CB language may believe that they are identifying with the world of truckers by learning their lingo. In fact, regular users of CB can easily detect the amateurs’ imperfect learning and may refuse conversation with the “nonlegitimate” users. Moreover, some of the truckers’ language has evolved away from the forms popularized by amateurs. Among truckers, the proverbial good buddy has taken on the meaning of “gay” and has been replaced by old buddy or driver.

While CB language bears many of the trappings of sublanguages par excellence, it functions, at least among many of its users, as much for social identification as to get work done. Much of the lexicon is self-evident: an expressway is a superslab, an armored truck is a branch bank, and a gasoline truck is a thermos bottle. These are not randomly assigned pairings. At the same time, use of the CB term (rather than the word in everyday language) serves no obvious function other than the identification of the speaker as a member of a group. The logical extreme of this is the use of language as a wholly fortuitous vehicle for social identification, to which we turn next.

DIPHTHONGS ON MARTHA’S VINEYARD: FORM OVER CONTENT

The case of diphthongization on Martha’s Vineyard is a classic example of the use of an arbitrary linguistic form for social purposes (Labov 1963; reprinted 1972b). Martha’s Vineyard is a small island off the coast of Massachusetts. Best-known as a summer vacation haven, it is also year-round home to many people, some of whom identify quite strongly with the island and take pride in distinguishing themselves from people living on the mainland. Labov’s study revealed that many islanders pronounced the diphthongs /ay/ and /aw/ in a more centralized position in the mouth than did their counterparts on the mainland. In other words, a word like five sounded more like [ƒəiv] than the standard English [faiv]; house sounded more like [həos] than the standard English [haos]—[a] being a lower sound in the vowel space than [ə].

Not all islanders evidenced the same amount of centralization; a study of social factors revealed that the more a speaker identified with the island, the greater the degree of centralization was likely to be. Factors such as age turned out not to be as important. There were cases of younger people who, upon returning to the island after receiving their education, adopted the increased degree of centralization. One woman remarked of her son, “You know, E. didn’t always speak that way . . . it’s only since he came back from college. Guess he wanted to be more like the men on the docks . . .” (Labov, 1972b:31).

Our discussion of sublanguages has illustrated how the choice of linguistic constructs can be socially motivated among groups of speakers from a single linguistic community. Let us now explore how similar functional explanations can be applied in cross-linguistic situations. The specific cases we shall examine are spoken language systems which are commonly known as trade jargons and as pidgins.

TRADE JARGONS AND PIDGINS INTRODUCTION

INTRODUCTION

The term pidgin is commonly used as a catchall to refer to a vast range of languages linked by two criteria: they contain elements from more than one language, and they are judged not to be as “complete” or “sophisticated” as any of the donors. In the previous attempts to define pidgin, we find great variety in the kinds of criteria chosen.⁸ Many descriptions have focused on the lexical and syntactic composition of pidgins (it is typically claimed that the vocabulary is predominantly from a European language and the syntax from a non-Indo-European language).⁹ Another common characterization is in terms of native speakers: pidgins are said to have none.¹⁰ A few linguists have even commented that pidgins seem to arise in coastal areas.¹¹ All of these criteria are useful, and, in general, accurate. But, as with the proverbial blind men and the elephant, these isolated characterizations do not leave us with a clear understanding of what a pidgin really is.

Why are pidgins so interesting in the first place? The simplest answer is that they are peculiar kinds of languages—and languages are what linguists are supposed to study—that arise when people who do not share a common language meet under special sets of circumstances. The next logical questions to ask are:

In exploring the nature of pidgin, we shall be comparing another peculiar language type—trade jargon. The deductive method can also be applied to Creoles, lingua francas, and koinés (although these languages go beyond the scope of the present discussion). All of these are linguistic “interfaces” between speakers not sharing a common language and may therefore be referred to as “interface languages.”

This section will examine the hypothesis that the structure of trade jargons and pidgins can be deduced once we know the objects and activities in each original speech community’s experience and the relative social and political footing on which the two groups confront each other.¹² Using these two variables, we shall also be able to ask diachronically why these languages arose at the times and places they did and between the groups they did, why many of these languages have died out, and whether such languages will continue to emerge in the future. However, before addressing these specific questions, we need a way of talking about relations between people, their language, and their modes of living.

PEOPLE, LANGUAGE, AND EXPERIENCE

In talking about human language, we can separate the speaker (S), the hearer (H), the language system (L), and experience (E). Three primary relations and one composite relation between these elements can be defined with respect to L. The primary relations are:

between the speaker and the language: SR₁L

between the hearer and the language: HR₂L

between the thing talked about (experience) and the language: ER₃L

The composite relation connecting speaker and hearer may be defined as SR₁LR₂H (this avoids the assumption that both speaker and hearer have the same relationship with the language). A person is said to be linguistically saturated when his command of the first two primary relations (SR₁L, HR₂L) and the composite relation (SR₁LR₂H) is such that the community does not correct his linguistic behavior.

Experience

Adult languages may differ from each other with respect to the third primary relation, ER₃L. This relation belongs to the realm of applicability—the way in which a linguistic community uses its language to encode experience as well as the adequacy of this encoding. The language of a (single) speech community may change when the language is not felt to encode experience adequately.

This suggests that the daily activity of members of a linguistic community is reflected in their vocabulary and syntax. On one level, this observation is trivial and self-evident: if I dig ditches, I need names for my tools and if I play chess, I need a set of syntactic constructions that lets me describe the movements of the pieces. On another level, the problem is more significant: Are all languages able to talk about all activities? Does the speaker of Navaho have any way of talking about notions like “momentum” or “impressed force,” or is the Neo-Melanesian Pidgin word pairapim (“explode”) adequate for talking about nuclear explosions?¹³

The “experience” of members of a speech community refers both to the material objects in the culture (shovels and chess sets), which I shall represent by the set {M}, and the activities which the people engage in (digging ditches and playing chess), which I shall represent by the set {A}. In order to distinguish between these two aspects of experience, the two relations which make up ER3L will be identified as MR4L and AR₅L.

Speakers of a language confront the adequacy problem when they encounter new objects or engage in new activities. As I have just said, one of the ways languages change is by devising ways to talk about new objects and new activities. The adequacy problem also applies to speakers of language L₁ and language L₂ when two speech communities come in contact with each other. If the groups encounter similar objects and engage in the same kinds of activity (if the relationships R in E₁RL₁ and E₂RL₂ are similar), translation or borrowing between L₁ and L₂ will not be particularly difficult. If differences between E₁RL₁ and E₂RL₂ reflect material differences between the two groups (i.e., M₁RL₁ ≠ M₂RL₂), contact between the groups may result in a language interface that centers upon the new material objects encountered. This situation typically lies behind the development of trade jargons. However, if both groups engage in radically different activities (if differences between E₁RL₁ and E₂RL₂ reflect incommensurate life styles—i.e., A₁RL₁ ≠ A₂RL₂),¹⁴ then adequate translation will be difficult, if not impossible. I hypothesize that, under these conditions, a radically different kind of language, a pidgin, arises.

Social and Political “Footing”

If we consider two communities of speakers who come in contact with each other, the first group (Gi) may be either (to oversimplify for a moment) on an equal or an unequal social or political footing with the second group (G₂). By social or political footing I simply mean, Is one group in a position to exert power over the other? In examining interface languages, we shall see various ways in which the sociopolitical relationships between groups in contact determine the kind of language that emerges from the contact situation. When the relative positions of the groups change, we predict concomitant changes in the interface languages linking the two communities.

I have isolated six types of relations—

and two kinds of variables—sameness or difference of material objects or activities, and relative social and political position. These relations and variables enable us to talk precisely about language used between speakers and hearers from different speech communities that come in contact with one another. It is possible to construct a matrix of relations containing all possible permutations of material objects, activity, and sociopolitical position, and to ask what kind of interface language might arise under each condition. The present investigation is, however, restricted to two cases: difference in material objects and equality of social and political footing; and differences in activity and inequality of footing. Let us begin by looking at the kind of situations that give rise to these collocations of variables, and then we shall examine the sorts of languages that derive from these situations. Finally, we shall come back to the questions posed about the temporal and geographic past and future of interface languages.

EXCHANGE AND EMPIRE: THE HUMAN CONTEXT

Since the dawn of human history, groups of people have been exchanging goods with other groups of people.¹⁵ Humans have exchanged everything from yams for shells, to cows for wives, to missiles for oil, to five days’ labor for one week’s pay. Exchange may take place under a seemingly limitless variety of circumstances. However, we can identify two ends of an exchange spectrum.

At one end—which includes the vast majority of human trade—two parties confront each other on a relatively equal social footing. Both groups stand to benefit equally from the exchange (although, like the unsuspecting Indians who sold Manhattan, they may be misled), and both groups can choose whether or not to engage in the transaction.

At the other end of the spectrum, which historically includes a much more restricted set of encounters, groups of people come together on an unequal basis. This may take the extreme form of slavery, a more intermediate form like contract labor, or the apparently voluntary form of freely hired labor. However, in each case the group on the short end of the stick has 110 real choice about participating in the exchange.¹⁶ Moreover, instead of exchanging goods, the subordinate group exchanges labor for subsistence income (or the equivalent) from the dominant group. This labor may entail not only performing new kinds of tasks (e.g., loading large ocean-going vessels) but also living according to a radically different kind of social organization (e.g., a hierarchical society which imposes a regular daily routine on the worker).¹⁷ Frequently, exchange under coercion was preceded by an earlier period of exchange of goods between equals (as in the cases of Hawaii, China, and New Guinea). This change of circumstances is known to historians as the development of colonial empires by western Europe.

What does commerce have to do with language? When two people (or groups of people) who do not share a common language meet for the purpose of exchange, they nearly always need to find some way of talking to one another. It is under these situations that interface languages develop.

We can further specify two types of encounters from which we predict that two very different types of linguistic interfaces will arise. In considering trade between equals who do not share a common language, I shall restrict myself to encounters between representatives of both groups (typically adult males) in which the amount of time spent together is relatively short, and the majority of conversation centers upon the material objects being traded.¹⁸ The opportunity for one group to fully learn the language of the other is limited by amount of exposure and by motivation. This kind of contact gives rise to a linguistic system known as a trade jargon, which is not actually a language but an intersection of partial languages.

Subordinate populations exchanging labor for subsistence have a peculiar kind of relationship with the dominant group. As with free trade, the exchange is typically between adult males.¹⁹ Here, however, the subordinate population is compelled to spend large amounts of time with the dominant group (or its intermediaries); and yet, the increased amount of contact does not mean that one group will learn the other’s language. By virtue of their differential social position, the dominant group cannot learn the language of the subordinate group, or vice versa.²⁰ This kind of contact produces linguistic systems we call pidgins, which are full fledged languages, although of a strange sort.

FUNCTIONALLY DERIVED DEFINITIONS

What features delimit classes of trade jargons and of pidgins? The following set of structural characteristics are not meant to be exhaustive (e.g., they do not mention the use of reduplication in pidgin), nor do they reflect the fact that many languages that are traditionally called trade jargons or pidgins did not arise under the specific social and economic conditions mentioned above. However, we should be able to explain languages that do not display these structural characteristics in terms of the actual social and economic conditions applicable in those situations.

STRUCTURAL CHARACTERISTICS

TRADE JARGONS

PIDGINS

The remainder of this section will explore the extent to which these features of trade jargons and pidgins are directly predictable from the functional needs of the speakers forging these languages.

Functional Definition of Trade Jargon

Trade jargons are linguistic interfaces that arise because the material objects in the language of one group of speakers’ experience (Mi) are different from the material objects in another group’s experience (M₂). The resulting linguistic interface, however, is not one language but two.²¹ Both share some vocabulary but derive the rest of their vocabulary and almost all their syntax from the basic language of the speaker. These shared elements, by themselves, do not constitute a language. We can understand this situation more clearly by introducing some simple formal symbols to talk about the two base languages and the resulting linguistic interfaces.

The base (native) language (L₁) spoken by one group of speakers (G₁) encodes the set of material objects which the speakers normally experience ({M₁}). Contained within the set {M₁} is a subset of objects {m₁}, which this first group of speakers is trading with speakers of a different language. Similarly, the second group of speakers (G₂) speaks a base language (L₂) which encodes the material objects they encounter in their normal experience ({M₂}). Contained within {M₂} is a subset of items to be traded, {m₂}.

When the two groups encounter each other, the first group will need to modify its language to incorporate names for the items they are receiving in trade from G₂. In some cases the items being received in trade may already have a native name (e.g., English has a word for “oil”). In these instances, we cannot consistently predict whether the word from L₁ or L₂ will be used in trading either by G₁ or G₂. For the sake of simplicity, all names for received items are assumed to come from the language of the group offering these trade items. The new modified language, L₃, will include most of the structure of L₁, that is, the nouns which are not names of items to be traded, the best of the vocabulary and syntax of L₁, plus the new names for the items received in trade from G2:

L₃ = L₁ + {m₂}

or, to separate out the items Gi is trading from the rest of their base language L₁:

L₃ = + {m₁} + {m₂}

The second group also develops a modified language, L₄, which incorporates names for the items they receive from G_t:

L₄ = L₂ + {m₁}

or

L₄ = + {m₂} + {m₁}

When we speak about trade jargons, we are really talking about the modified languages L₃ and L₄. However, if we need a shorthand for talking about both L₃ and L₄, we can define a trade jargon as L₀ (the residual lexicon and syntax from L₁ and L₂), plus the vocabularly for the items both groups are trading:

L_T = L₀ + {m₁} + {m₂}

This model is, of course, grossly oversimplified in its implication that the only shared vocabularly items in L_T are names for the items being traded. The basic model is constructed this way because the source of other shared vocabulary is, we predict, random (it cannot be deduced from the general structure of trading relationships) and because the vocabulary lists of trade jargons found in the literature are often unreliable (we often do not know whether they reflect L₃ or L₄ or both). Furthermore, because these lists, as in the case of Eskimo or American Indian trade languages, are published for English-speaking audiences, it is highly possible that English words that were used in trading were omitted, since the English-speaking audience already knew them. Stefânsson, for example, prefaces his vocabulary of the Eskimo Trade Jargon of Herschel Island with the admission that

Most of the confusion over whether trade jargons are actual languages (i.e., Do they have a stable pronunciation, lexicon, and syntax?) has arisen because of the kind of data on which we have based our descriptions. As Silverstein (1972) convincingly established in the case of Chinook Jargon, the use of informants who speak different base languages can lead one to construct radically different grammars for L_T. However, groups trading with each other are not necessarily aware that they are speaking different languages; in fact, they are sometimes convinced that they are speaking not their own base language or even some mixed jargon, but the language of the other group. The missionary Paul le Jeune is quoted as saying of the Franco-Amerindian contact vernacular that “the Frenchmen who spoke it supposed it to be good Indian, and the Indians believed it to be good French” (1632; in Hancock, 1971:512). The same phenomenon occurs with other kinds of interface languages; Cole mentions that speakers of Bantu languages who learn Fanagalo “fondly imagine that they are speaking the language of the White man” (1953:7). This belief is not unlike the conviction of the American tourist that he is speaking French when, in fact, he has merely “Frenchified” his pronunciation and added a few French words to his vocabulary.

To what extent are the structural characteristics proposed for trade jargons predictable from the human encounter of groups trading material objects on an equal footing? A trade jargon is not a stable language which is learned in toto (i.e., in which anyone can become saturated) because there are actually as many “languages” as there are base languages spoken by the groups engaging in trade. Because of the multiplicity of language bases, the vocabulary of Lr comes from both donor languages. Similarly, the syntactic structure of the language varies, depending on the base language of the informant. And because each group is using a (modified) form of its own base language, the syntax used may include some of the morphological inflections found in the base language.²²

Two other grammatical characteristics of trade jargons have been mentioned: unstable circumlocution (if any) and extralinguistic redundancy. If a speaker wants to refer to an object or activity for which he does not have a word in his lexicon, the most normal recourse is to describe the object or activity in terms of other words in his vocabulary; speakers do this all the time. If we want to talk about a particular person whose name we do not know, we might describe his height, weight, hair color, and so forth, hoping that the interlocutor will be able to identify the person from the description. However, if speaker and hearer spend a good deal of time talking to one another, and this nameless person is important in the conversation, by the simple rules of expediency, we shall either find out his name or coin one. In trading encounters, numerous situations arise in which one speaker is at a loss for a word that his hearer will understand. In some cases, the speaker resorts to circumlocution using shared vocabulary, but, because in these situations the period of contact between the trading parties is so limited, few of these spontaneous circumlocutions become crystallized in the language.²³ Frequently, however, when traders need to make their meaning clear, they resort to gesture to accompany their speech.²⁴ Since the exchange of material objects is the purpose of trade contact, in the early stages of trading, ostension (a form of gesturing) is sufficient to identify such objects. Like verbal circumlocution, gesture makes the total language system highly redundant in the information theoretic sense since a large number of units (in this case, words plus gestures) are needed to express a single piece of information.

Functional Definition of Pidgin

Pidgins arise when a dominant population imposes parts of its life style—which includes material objects (M₁) but, more importantly, particular kinds of activity (A₁)—on a subordinate population with a different life style (M₂ and A₂). A single language emerges whose lexical and syntactic composition reflects the relative power and different interests of the two groups involved.²⁵ As with trade jargons, the structure of pidgin language is more easily comprehended in terms of a formal representation of the donor languages and the resulting pidgin.

The dominant population (G₁) speaks a language (L₁) that encodes a set of material objects ({M₁}) and activities ({A₁)) which people in that group normally experience. Contained within {M₁} and {A_x} are two significant subsets, {m₁} and {a₁}, which are the objects and activities peculiar to the dominant group, and which the subordinate group will be encountering. Also contained within {M₁} and {A₁} are objects and activities that the dominant group already has in common with the subordinate population ({μ₁ }, {α₁}). If we label as the remainder of L₁ which is not relevant to contact between the two groups, we can define L₁ as the conjunction of and the two subsets of {M₁} and {A₁}:

The subordinate populatioh (G₂) speaks a language (L₂) that also encodes a set of material objects ({M₂}) and activities ({A₂}), very small subsets of which ({m₂} and {a₂}) are peculiar to G₂and will be relevant in speaking to the dominant group (e.g., native foods and customs), and other subsets of which ({/x₂} and {a₂}) are held in common with Gi. Labeling as L₂ the remainder of L₂ which is not relevant to contact, we can define L₂ as L₂ plus the two subsets of objects and activities:

What kind of language emerges out of this mixture? The characterization of pidgin proposed at the beginning of this section provides a convenient framework for answering this question.

Stability Pidgins are stable languages over which speakers from either the dominant or subordinate group can gain mastery, although it is frequently said that the subordinate population “knows” the language better than the dominant group (e.g., Hall, 1955:22). The stability of pidgin follows directly from the fact that the social situation itself is stable: for years—even decades, day after day, the subordinate population performs the same tasks and maintains the same relationship with the dominant group. In some instances (e.g., Hawaii) this dominant/subordinate relationship has given way comparatively quickly as the dominated group began leaving the plantations and learning the dominant group’s language (see Reinecke, 1969:100). In other cases (e.g., New Guinea) the inequality—and the use of pidgin—has persisted much longer.

Source of Vocabulary The majority of the vocabulary comes from the dominant language L₁ because of the position of authority held by speakers of G_lt The dominant group introduces a number of new concepts—{m₁} and {a₁}—which form the basis of the pidgin in that they represent the new kinds of objects and activities that the subordinate group must confront in working for G₁. The subordinate group also has a few items of experience {m₂} and {a₂} unknown to G ₁ that find their way into the pidgin lexicon, but, because the subordinate group must fit its life style to that of the dominant group (and not vice versa), there are comparatively few of these items.

The rest of the pidgin vocabulary refers to experience that both groups shared before contact (eating, birth, death). However, because the dominant group has the upper hand, it can retain its own words for these aspects of experience, thereby forcing the subordinate group to accommodate itself either by learning these parts of the vocabulary of Lq or by using circumlocution (see below).

Yet despite the overwhelming proportion of pidgin vocabulary taken from the dominant language, the pidgin vocabulary remains much smaller than that of the dominant language itself.²⁶ The primary reason for this restriction is that, in the typical social situation in which pidgin is used, the laboring population is not given the opportunity to learn L₁²⁷—not only because it is felt to be unnecessary, but also because the dominant group may believe that the “natives” are not smart enough to learn the language or because the dominant group wishes to maintain social distance between themselves and the “lower orders.”²⁸

Source of Syntax Trade jargons exhibit a fairly sophisticated level of syntax that can be traced directly to the base languages. However, in the case of pidgin, the question of syntax assumes a different cast. Pidgin has a novel syntactic structure, chiefly characterized by highly simplified morphology and word order, circumlocution, and syntactic redundancy (the latter two resulting from the paucity of vocabulary). Nevertheless, many linguists (e.g., Bateson, 1943:138) have commented that pidgin syntax, to the extent that its origins can be traced, seems to derive more from the subordinate language (L₂) than from the dominant language (L₁).

We need to distinguish between the syntactic structure of a language (in this case, pidgin), and particular syntactic features which are used as surface markers of underlying syntactic and semantic relationships. We must nevertheless remember that these specific syntactic features, however pervasive they may be in texts (e.g., Neo-Melanesian -im marking transitive verbs), are not integral parts of the syntactic structure itself. Neo-Melanesian Pidgin would still be a pidgin without its -im, because the choice of syntactic features has little (if anything) to do with the new activities of the subordinate population, or with the relationship between the dominant and subordinate groups.

To the extent that the syntax of pidgin delineates common grammatical properties of both L₁ and L₂ (distinguishing between transitive and intransitive verbs, aspectual distinctions, singular and plural number), it is structurally irrelevant whether L₁ or L₂ is the source of explicit syntactic constructions in pidgin (to the extent they exist—see “Complexity of Syntax”). However, there are several reasons which make L₂ (the subordinate language) the logical choice. Since the dominant group (missionaries generally excepted) is primarily interested in using pidgin to get its orders carried out—not in educating the “natives” —they will arrive at the simplest language possible for the hearer (the subordinate group) to comprehend in the shortest amount of time. That is, the relation HRL takes precedence over SRL (see “People, Language, and Experience”). To the extent that vocabulary conveys the dominant group’s intentions, there is no time—or need—to worry about niceties of syntax. For the subordinate population—just as for the trader—it is far easier, as hearer, to learn the vocabulary than the syntax of another language. Maintenance of scattered syntactic features from the subordinate group’s language is irrelevant to the dominant group since these features play little part in conveying the dominant group’s instructions to its subordinates. While these features are equally useless in making the subordinate group understood by the dominant groups (who do not understand the lower group’s native language[s]), their use is just as “natural” as the traders’ use of syntax from their base languages—which the group with whom they are trading does not understand. Finally, to the extent that syntactic elements of the subordinate language become part of the pidgin, such markers will be easily learnable because of simplicity and high frequency (e.g., Neo-Melanesian transitive verb marker -im or predicate marker i).

But why does the dominant group not introduce syntactic features from its own language? I hypothesize that, as with their restricted use of vocabulary, the dominant group may either believe the “natives” too stupid to learn their syntax, or that they may wish to maintain linguistic distance between themselves and their laborers. Furthermore, nearly all the dominant languages were, historically, western European languages (especially English), which have comparatively little surface syntactic marking anyway.

Complexity of Syntax As mentioned above, one of the chjef characteristics of pidgin is its scarcity of morphological features and syntactic rules. Rather, it is a highly “simplified” language—typically having a set word order, not marking gender or number distinctions, differentiating between aspects but not tenses. These characteristics are predictable from the situations in which pidgins are used. Generally speaking, the simpler the morphology and syntax, the shorter time required to learn (or create) a language. It is advantageous to both parties to understand each other as quickly as possible: the ruling groups stand to increase their profits as they increase the efficiency of their laborers —which requires the laborers to understand what they are being asked to do; and members of the laboring group may expect better treatment from their superiors if they can make themselves understood.²⁹

Circumlocution Having just said that pidgin is a very simple language with respect to its limited vocabulary and syntactic machinery, I must now explain the presence of circumlocution. Circumlocution—perhaps the most distinctive characteristic of pidgin—is an extremely cumbersome way of getting ones point across. However, once again the human context in which pidgins are created and used helps explain the usefulness of circumlocution.

The dominant group, as in the case of syntax, is not concerned with the propagation of its mother tongue but that members of the subordinate group understand its desires. If the native doesn’t understand the word never but comprehends i no gat wanpela taim (Neo-Melanesian), simplicity is sacrificed for the sake of comprehension. If, by tupela han wanpela fut, the subordinate group understands fifteen, there is no reason to extend one’s knowledge of the subordinate group’s vocabulary.

Consider the similar position of a speaker from the subordinate group, to whom it is of paramount importance to be understood by the overseer who may hold the power of life and death over him. If, for example, while working in the fields, a laborer needs to relieve himself, he must make absolutely sure that his superior understands that he is not running away. It matters little whether both his native language and that of the overseer have simple lexical items for diarrhea. If the native does not know the word from L^ and is not sure the overseer knows the word from L₂, then it is most expedient to describe the problem in no uncertain terms:pekpek no gut olsam wara (again, Neo-Melanesian). The circumlocution may save him from a beating —or worse. Likewise, if the laborer injures himself and runs for a tourniquet, he wants to be certain the overseer will understand that he is not shirking his duties but only looking for strongpela rop bilong pasim rot bilong blut.

Yet there is a problem with the use of circumlocution. Every language encounters an optimization problem. The fewer the vocabulary items in a language (whether it is a natural language or an interface language like pidgin), the simpler the language. However, if the vocabulary is insufficient for expressing what its speakers want to say (either because no appropriate word exists in the language—as in the case of natural languages—or because the speaker does not know the word in the foreign language and is not certain the foreigner knows the speaker’s word—as in the case of pidgin), the speaker may either add a new word, thereby complicating the vocabulary but reducing potential syntactic redundancy—see “Redundancy”), or may resort to circumlocution, which keeps the vocabulary small but increases syntactic redundancy.

It is difficult, if not impossible, to determine which solution the language will select in a particular case. And so, in looking at contact situations which generate pidgin, we can’t predict which particular words will be adopted directly from the dominant (or, less often, from the subordinate) language, and which will be expressed phrasally. However, when pidgins are stable for a relatively long time, circumlocutions are often replaced by single lexical items, usually from the dominant language.

Nevertheless, we might ask whether there are any principles governing this vocabulary distribution. Consider first the situation in which the item or activity talked about is part of the previously shared experience of both speech communities and for which both languages already have a single word in their lexicons. At first glance we might hypothesize that, since experiences can here be easily translated, speakers from one group would simply adopt the lexical item from the other (usually from Li). However, a quick look at Neo-Melanesian disproves this hypothesis. The most mundane objects and activities—for which we assume the native language of the subordinate population has items in its lexicon—are expressed by circumlocution, for example: deaf (i no save harim tok) or gratitude (pasim bilong givim tenkyu long arapela man). Thus, it seems that, for pidgin, the optimization balance is weighted in favor of circumlocution—presumably for the reasons we have discussed.

But what about the areas of experience that the dominant population imposes on the subordinate? Should we expect these to be expressed through circumlocution rather than lifted directly from the dominant languages lexicon? Because it is sometimes extremely difficult to translate these new experiences into the indigenous language, and because these are the terms which the dominant population is likely to speak most frequently to the “natives,” we would expect words for these new aspects of experience to be borrowed directly from the dominant language. Judging from Neo-Melanesian, this hypothesis generally seems to hold—automobile (motoka) , battery (bateri), fountain pen (fauntenpen) , screwdriver (skrudraiva)—although there are some exceptions, e.g., dynamo (masin bilong mekim lektrik).

Redundancy As we have just seen, pidgin has a high level of syntactic redundancy in that circumlocutions require a large number of items to express a single piece of information. I have already discussed the situational factors that make this syntactic phenomenon highly functional.

In this section and the previous one, I have shown how certain kinds of linguistic structure can be deduced from a small number of nonlinguistic features of human interaction, especially the nature of one’s experience and the relative footing on which one confronts speakers of a different language. To test the usefulness of the theory of interface languages I have proposed, it will be necessary to compile data from a large number of trade jargons and pidgins. Before undertaking such research, however, we can benefit from a historical perspective on the actual situations in which interface languages have arisen.

PAST, PRESENT, AND FUTURE

The empirical analysis of trade jargons and pidgins cannot be undertaken by looking under t or p in an appropriate compendium. Using the above characterizations—either linguistic or nonlinguistic—we find that the names traditionally given many interface languages are misleading. A quick glimpse at Hancock’s “Survey of the Pidgins and Creoles of the World” (which appears as an appendix to Hymes, 1971) emphasizes the problem: Pidgin Eskimo and Pidgin Spanish are not pidgins but trade jargons; New Jersey Amerindian Trade Pidgin is a contradiction in terms; Hawaiian Pidgin English and Chinese Pidgin English may have been pidgins at some point in their history, but we cannot be sure until we sort more carefully through the linguistic and nonlinguistic data.³⁰

If we can’t rely on linguistic labeling to tell us which interface languages are potential trade jargons or pidgins, then we need some other means of finding the right empirical domain to explore. The answer is not difficult to find: if we are correct in saying that trade jargons and pidgins are linguistic responses to conditions under which humans confront each other, we can then ask, historically, when and where these language have arisen. For example, the history of late mercantilism and early capitalism will explain why pidgins (as defined here) did not really appear before the eighteenth and nineteenth centuries, why they almost invariably grew up in coastal regions of non-European countries, and why the dominant language in almost every instance is a western European language.³¹

We can also use history to understand why so many of these languages have died out or have been replaced by other types of languages (specifically, trade jargons giving way to pidgins, and pidgins either giving way to Creoles, or dying out completely). By looking at the numerous shifts in the nineteenth and early twentieth centuries from trading relationships to relationships of empire, we can understand the rapid rise of pidgins during the height of Western colonialism; by observing the growing independence of former colonies, we predict the demise of pidgin. Furthermore, unless the structure of the world economy drastically alters in the next decades, trade jargons may continue to emerge on a limited basis between speakers from nonmodernized countries (where speakers do not share a common third language like English, French, or Swahili); pidgins, on the other hand, probably will not. There are no longer the proper mixtures of unprotected “natives,” untapped natural resources, and dominant populations who are allowed the upper hand in the expropriation of labor and materials.