16

MICHAEL HERZFELD

Disemia¹

The central argument of this paper is that the linguistic formulation of “diglossia”, whereby a single language may have both literary (“H”) and vernacular (“L”) “registers”, is part of a larger semiotic phenomenon in which individuals are able to negotiate social, national, ethnic, or political boundaries through a potentially inexhaustible range of co-domains. Language, though important, is not necessarily primary in this wider phenomenon, which I propose to call disemia. Disemia is thus a higher-order concept, not only than diglossia, but also than all such models as political polarization, class-based differentiation of behavior (including kinesics and proxemics), “folk” versus “urban” culture, and the like. In Ardener’s sense (1971), it is a paradigmatic structure, a category of formal social principle rather than of behavior type.²

Defining the Issue

Ferguson’s seminal presentation of the original diglossia model (1959) treats it as an ideal-type construct, or set of “continuous variables” (Johnson 1975: 39). Intermediate or “mixed” forms represent emic categories which permit a considerable range of variation in actual speaking practice. Like the “register” categories, the isolation of language as a discrete domain seems justified as a reflection of indigenous perceptions, at least as far as Greek (“the language question”) and Arabic are concerned, and the crucial role of literacy in determining the choice and range of any one speaker’s utterances reinforces that impression (Ferguson 1959; Fishman 1967; Shouby 1957). Yet this perspective shares the a priori logocentrism of many sociolinguistic categories (cf. Webber 1973; Crick 1976: 63-68), and demonstrates that a commitment to the reporting of emic categories does not adequately guarentee the representation of conceptual structure. If emic analysis stops at language, it is in fact rarely emic at all; it becomes instead a passive commitment to the etic view that a spoken language is capable of expressing the entire range of what its speakers are able to mean. It is a rejection of the very possibility of a “symbolic syntax” (Ardener 1970; Crick 1976: 111), or semiotic system, in which linguistic signs have a co-operative rather than a dominant function.

Thus, the alternative model of disemia poses a heuristic challenge. If the social “conditions of use” of diglossic variants in language can be identified, and if these conditions are not merely dependent functions of linguistic discourse, then they must represent principles which can be sought in non-linguistic behaviors also. If language exhibits diglossia, why not seek disemia in gesture, in architectonics, in music (surely a fertile field), in food habits?

The conditions of use for disemic variants generally are likely to be of two kinds. Conditions of the first group are historical or ideological; they consist of those preconditions which have been determined by the culture’s prior ideological experience, and which in turn determine the appropriateness of a register-marked usage. Those of the second group are immediate or communicative; as in Fillmore’s formulation (1971: 278), they are determined by the utterer’s intentions and communicative competence, and constrain the situational appropriateness of a particular parole. The two sets thus correspond to the two fundamental kinds of diachrony, that of linguistic (or cultural) evolution and that of particular utterances respectively, and are necessarily linked in an analogous and unceasing dialectic. In other words, the cultural presuppositions which govern an individual’s selection of register both influence, and are cumulatively influenced by, the particular choices that are actually made.

I shall attempt to demonstrate the model ethnographically, using the paradigm case of modern Greece. This does not mean, of course, that Greek disemia is necessarily motivated by the same antecedent principles as comparable phenomena elsewhere, but only that the generalization of a linguistic to a semiotic model offers more generous possibilities of explanation. Presumably, the potential of this revised model would then have to be worked out for other cultures on an empirical basis. It is a heuristic rather than an elaborately taxonomic device; otherwise, it could only entangle us in a thankless choice between reification and runaway typologizing—which is where, to some extent, the uncritically exploited diglossia model has already led. One cannot emphasize too strongly that a deterministic model, rather than one which is sensitive to the semiotics of social interaction, will lack heuristic efficacy, by focusing on disemia as rigid sets of rules rather than as a model for folk rhetorics that can be manipulated to achieve (“constitute”) all sorts of special effects—irony, contrast, humor, and much else (see Bouissac 1977, Kendall 1981, Schwimmer 1979).

The Greek Paradigm: First Approximations

Greek H (katharevousa)³ is not so much “purely Classical” as definitively European. While Turkish-and Arabic-derived forms were supplanted during the eighteenth-and early nineteenth-century emergence of Greek nationalism by Classical lexemes and syntactic markers, some syntactic elements (notably the genitive relative tou opiou) appeared to calque German or French prototypes. It is not clear how far such calquing represents conscious design, but—to anticipate the argument—it is at least consistent with the more general practice in nineteenth-century Greece of adopting Western European cultural realia⁴ on the grounds that they were “really” (i.e., derivatively) Hellenic. The self-view of the (modern) Greeks as “Hellenes” was directed to external consumption, and was a major ideological component in the articulation of Greek nationalism with Western European political and moral support (Herzfeld 1982). As a recent paper by Sotiropoulos (1977) demonstrates, the development of H-Greek served the economic, political, and ideological subordination of the emergent nation-state to the European Powers, by concentrating communicative control in the hands of an outward-directed, Western-educated elite. Although there may have been deliberate intention in this, given the peculiarly reflexive or metalinguistic properties of language, there is no reason to assume that something analogous was not taking place with regard to other, perhaps less obtrusive co-domains.

L-Greek, on the other hand, came to be associated with a more introverted ideology, one that was concerned with the Byzantine and Turkish roots of the modern culture, and with peasant society, rather than with the Periclean glories. In its extreme development, as dhimotikia,⁵ it became the clearest index of a speaker’s Marxist orientation—a “foreign dogma”, in the pronouncements of the H-using political Establishment, and certainly opposed to the “Hellenist” perspectives of the latter.

The principal axis along which these two ideologies were opposed was that of introversion/extroversion (or inclusion/exclusion). They thereby not only reproduce the definitive contours of a “culture text” (cf. Winner & Winner 1976: 106-107), but constitute the basis of a dialectical relationship between two explicitly recognized indigenous models.⁶ Moreover, the disemic registers encapsulate “constructions” (Goldstein 1976) of history, so that each social interaction reproduces the tension between extremes, and each utterance has the effect of creating anew (“performing”) the lines of social interaction between actors as a microcosm of the wider ideological issue.

Let me clarify, first of all, with a linguistic example. When a rural Greek wishes to explain something to a co-villager, he will normally use the term yati for “because”. When he wishes to imply some external authority for his opinion, he will use the Neo-Classical (H) form dhioti. Note that he never uses dhiati, “why”, for the L-form yati (“why” here); either the questioner assumes the superordinate position in the exchange (cf. Goody 1978: 17-23; Labov 1969: 57; Herzfeld 1980), or else a semiotic claim to the authority of external validation would violate “felicity conditions”. When the villager wishes to explain something to an outsider, such as the foreign ethnographer, he is more likely to use dhioti than yati for “because” as an index of his interlocutor’s external status and of his own desire to show familiarity with educated usage.

This is more than simply “code-switching”. The possibility of such connotative manipulation is part of the code itself. If connotation is not so much a “vaguer” kind of meaning than denotation, than it is a superimposed level of sign-production (Eco 1976: 54-57), the notion of rigidly differentiated codes simply reifies the taxonomy of subcodes and ignores the rich semiotic implications of so-called “mixed forms”—of ordinary discourse, in fact. It is not the lexical item that indicates which code is being used; rather, it is the use of a given lexical item which indicates (a) the burden of historical and ideological implication, and (b) the degree of social closeness between speaker and addressee. Thus, a “use theory” not only serves semantic analysis well in regard to encyclopedic knowledge (Eco 1976: 98-100, 1984: 46ff.), but also provides a more flexible (and less aprioristic) approach to the semiotics of diglossia. Phonologically identical units may and do appear in dictionaries of both registers with identical semantic descriptions. What distinguishes them is the way in which they are used, and this is knowledge of an encyclopedic kind.

Outside language, the same observations hold good. A Neo-Classical house-ornament which originally bespoke an H-orientation, rescued from the demolition of its original home and incorporated into the design of a small Piraeus house surrounded by skyscrapers would now usually convey, not the Hellenic pretensions of its new owner, but his extreme demoticism in the threatening face of “Western culture”. It is not the sign, but the way the sign is used, that tells us about its motivating ideology. When the once-despised underworld rebetika songs began to acquire popularity with the intelligentsia and then gradually with the Establishment generally, good Classical roots were suddenly “discovered” (or “constructed”) for them (Veinoglou 1976). Sung by a drunkard in a Piraeus dive, of course, they still connote “low” culture. Sung in an Athenian saloni, other cues will enable those who know about such things to determine whether they are “meant” ironically or pretentiously. These are indeed “illocutionary or perlocutionary acts that are unaccompanied by locutionary ones” (Cohen 1975: 8). They achieve their effects in part, at least, by alluding to the presupposition of shared historical knowledge.

An Architectonic Example

Certainly, when one has examined diglossia/disemia as a semiotic rather than as a purely linguistic problem, there is no reason to confine the construct to language at all. This is particularly the case when we consider that the informing ideology is largely the same for language as for the other co-domains. True, language is indigenously treated as a special and discrete problem (to ghlossiko zitima). It is also true that non-linguistic artifacts mostly seem to convey their messages as simultaneities rather than as sequential utterances. Such objections are nevertheless hardly radical. We are not necessarily denying the distinctiveness of language by refusing to treat it as fully autonomous. On the contrary, the shift from diglossia to disemia gives us a context in which the linguistic phenomenon can be treated as a paradigm case rather than as a unique isolate. What makes what we should then call linguistic disemia different is what in general makes language different from other cultural artifacts, no more and no less, and it would be merely tautologous to insist on representing it as a special case—what paradigm is not?—of the wider phenomenon.

An architectonic example shows how disemic “marking” can function as an index of ideological orientation and of immediate social relationships simultaneously, in a manner analogous but not identical to that found in language. In an account of the Neo-Classical architecture of Athens and Piraeus, we are told, “It was natural for the newly constituted “good” society—made up of the merchants, the leading local headmen of Turkish times, the klefts and the educated people from the West and from Constantinople—to concede the lead to the Bavarians⁷ in determining the measure of social standards. . . . Dances, dress, manners followed Bavarian dictates. The same was true of houses. . . . Thus, the Neo-Classical house came to be adopted. In addition, it had certain advantages of convenience. It created a false impression of Greekness⁸ on the exterior. It emphasized the owner’s economic pretensions, flattered his newly acquired urban outlook, and, at the same time, protected his private life” (Iakovidhis 1975).

Here, microcosmically, we find the extroversion of the “Hellenist” model and the introversion of the self-critical “Romeic” model combined in a single “utterance”. This is a doubling of the indexical function; the protection of the householder’s private life was not merely a mechanical function, it was a means of signifying his privacy to the rest of the world. To explain in more detail: the “Hellenistic” exterior was an ideological statement of cultural identity, while the interior harmonized well with the introversion of the “Romeic” ideology in that it was immune to inspection by outsiders. The ideological choice a Greek makes between the two national ideologies may be rhetorically stated in absolute terms. In everyday life, however, the signs of both ideologies are manipulated situationally, whether in language, architecture, mores, or any other medium.

Moral disemia is particularly interesting. A villager can praise the aggressive self-regard known as eghoismos to his fellow-villagers, then condemn someone as having exhibited precisely that moral quality when describing the latter to an official or a foreigner. Such behavioral disemia can be identified in other cultures hitherto restrictively classified as “diglossic”, notably those of the Caribbean. Thus, the evidence adduced by Abrahams and Bauman (1971) and Reisman (1970) shows how an ostensibly negative evaluative term is used of L-behavior that is, in the appropriate social contexts, regarded as positively indicative of “insider” status. Again, while it would be insulting to label a Greek villager as a “thief” or a “liar”, both terms can convey praise in contexts where the reference is to particular acts performed in symbolic defense of the in-group (e.g., sheep-stealing from other villages in the West Cretan highlands, lying to conceal a household’s inner workings) (see also Friedl 1962: 80; du Boulay 1974: 77-78). The sense of the “lie” in the latter instance is directly analogous to a style of architecture which protects each household from the prying eyes of the next.

Thus, the disemia model allows us to treat various distinctive codes or co-domains in directly analogous terms. This is infinitely preferable to treating each as so completely sui generis that it has to generate a sub-discipline of its own—“sociolinguistics”, “the anthropology of aesthetic form”, and so forth. Above all, it broadens our understanding of the potential significance of “conditions of use” for a general semiotic theory, and for the penetration of interpretative systems other than our own.

Heuristic Possibilities

Disemia has heuristic potential in two respects. First, it invites a more critical investigation of cultures where H-and L-styles are indigenously recognized, whether or not they include a linguistic component. Stirling, for example (1965: 283-289), evocatively describes such a pairing (urban-rural) in Turkish attitudes; we may now be led to inquire whether, as the Caribbean materials cited above might suggest, there are any situations in which the L-style is evaluated positively, or whether the classification is as static and socially context-free as the ethnograghic description implies.

The second heuristic potentiality of the disemia model is that it points up the marking of social boundaries (insiders vs. outsiders) in such a way as to lessen our dependence on linguistic indices. Pocius’ analysis of Newfoundland hooked rugs (1979) demonstrates this well, in the indexical correlation of iconicity (designs) with degrees of socially hierarchical differentiation appropriate to the rooms in which the rugs are situated. Here, the internal H/L distinction does not appear to be replicated in language, despite some evidence (Faris 1968) of semantic discrepancies between Newfoundland usage and “standard” English. Pocius’ description of the phenomenon does not suffer noticeably through the absence of supporting evidence of a linguistic nature.

One test of the model’s flexibility lies in its ability to accommodate some of the effects of tourism on so-called “traditional” cultures. Tourism effectively calls on local people to provide foreigners with an idealized view of their culture, while at the same time encouraging the incorporation of these idealized (or attenuated) cultural forms into the local aesthetic (see, e.g., Hirschfeld 1977 and Sherzer and Sherzer 1976 on San Blas molas; cf. also Cretan touristika, woven articles partly used domestically). The significance of cultural elements is thus contextually negotiated (cf. also Schwimmer 1979: 272-273). The metasemiotic rhetoric may change too. Especially in and around the coastal town of Rethimno, Crete, the practice has sprung up of scratching designs in cement house-fronts before the latter have dried. The local people claim that this is done “for beauty”, whereas the wealthier and worldlier Irakliots—who use “Western” tiles and facings to a large extent—distain such decoration, and say that they abandoned it many years ago (Herzfeld 1971)! Beauty is in the eye of the impecunious Rethimniot householder who mainly has to demonstrate domestic pride to his local neighbors, whereas the Irakliot is additionally concerned with the impression his town will make on the tourist hordes.

To summarize, the disemia model suggests a broadly holistic use of speech act theory without demanding the presence of (verbal) speech. By incorporating the ideological implications of insider/outsider distinctions, it abstracts the H/L opposition from the restrictive model of diglossia and applies it to the whole range of everyday realia. It thereby suggests a new and specifically semiotic perspective on the social basis of aesthetic concerns.