[back to list of essays]

Markedness and Iconicity: Some Questions

1. INTRODUCTION

There has developed in linguistics a tradition for analyzing the members of grammatical categories and smaller, closed lexical categories in terms of features that are both privative and facultative, and that reveal in their design certain cross-linguistic tendencies that can be seen as iconic. It is part of what Greenberg (1966) characterizes as a "rich and complex set of notions . . . pertaining to marked and unmarked categories," originating in Prague school phonology but shown, primarily in the writings of Roman Jakobson, as capable of extension "to the study of grammatical categories and to semantics." Table 1 provides a summary of this set of notions, summarizing the characteristics set forth in Greenberg 1966, supplemented by the other sources indicated.

TABLE 1. NOTIONS PERTAINING TO MARKED AND UNMARKED CATEGORIES

This tradition has not been central to linguistics in America, either before or since 1957, and is sometimes not taken into account in discussions where it might afford insight. It is but one of several approaches to the identification of semantic features, some of which appear to have borrowed from it, while others seem oblivious to it. In this paper I take a brief look at several studies that have appeared since the mid-sixties and their relation to the tradition. I also note several questions that remain to be answered.

There are several things to be noted about the tradition. One has already been mentioned—its origins in phonology. There were insights to be gained from analyzing the contrastive elements making up the secondary articulation of language into the substantive features of sound that some of them possessed and others lacked; similar insights could be gained from a parallel analysis of contrastive elements in the primary articulation into substantive features of meaning. It is important to note that one began with formal, linguistic categories, and looked for substantive correlates, either phonetic or semantic. In both cases, the elements contrasted along a paradigmatic axis. In both cases they were members of relatively small, closed sets. On one hand, the phonemes; on the other, the grammatical categories within a paradigm (and possibly, also, similar categories within lexical sets). Such analysis transformed the elements from a collection of monolithic entities into interrelated groupings, based on shared qualities. Generalizations could be made, based on the qualities. Networks could be identified, new characteristics noted, and new questions posed, based on "the rich and complex set of notions" thus revealed. The investigation seemed to lead deeper and deeper.

Greenberg’s (1966) emphasis was on universals, on what was revealed about language generally. "What, if any, are the common features which would justify the equating of the concept of unmarked and marked categories in fields as diverse as phonology, grammar, and semantics? Is it possible to isolate some one characteristic which might serve as definitional for this notion which tends to take on Protean shapes? What is the connection between marked and unmarked categories and universals?" Greenberg pulled together the evidence that had been accumulating up until that time and posed questions such as these. He also added evidence concerning a most special closed lexical set, kin terms and their behavioral coordinates. But most of his questions, it would seem, remain unanswered at any deeper level.

This is not to say that the field has been vacant, or that his are the only questions that should be asked. Householder’s (1971) Linguistic Speculations contains several chapters that continue clearly in the direction set by Jakobson. Chapters 9 and 10 are commended by Matthews (1974:153) as a source on marked/unmarked oppositions in phonology. Chapter 12 is especially helpful in fully understanding the tradition with respect to semantic oppositions. Here Householder shows how a feature analysis of the English prepositions might proceed. Jakobson, "unfortunately, never gave a full exposition of his procedures so that . . . it is difficult for another linguist to duplicate his results. But we can easily see a number of them" (Householder 1971:228). Householder shows how "the one reliable criterion for discovery, contrast (or opposition)" might be employed in analyzing this particular closed set as an example.

Several authors in recent decades have approached the topic of iconicity in language quite generally, and have included markedness in their discussion (Haiman 1980, Matthews 1972, 1991). In the first edition of his textbook, Matthews (1972:150–153) gives brief mention to markedness in an appendix to his chapter on ‘Properties and their exponents,’ while also pointing the reader to Jakobson (1932, 1936), Lyons (1968), Householder (1971), and Benveniste(1966). He devotes an entire chapter in the second edition (1991) to iconicity in its syntagmatic as well as paradigmatic manifestations. He goes further in relating markedness to iconicity than other works I am aware of, but is slow to claim universal applicability. "To many scholars, it does not seem right to speak of explanation, even in what is plainly a branch of the humanities, unless there are laws that cover every instance. But we have stressed that there are no laws, only tendencies . . . A more realistic view [is that] the principle of iconicity will help to explain aspects of the history of languages" (1991:243–244).

2. AN INSTANCE OF INCONGRUITY

As an original contribution of this paper, I would like to call attention to a troublesome example from one language in which there is an incongruity in the picture associated with what seems to be a well-established tendency, that of plural to be marked and "singular" to be unmarked. This is to be found in the Marshallese demonstratives (Bender 1969:76), where a human-nonhuman distinction is maintained in the plural but suspended in the nonplural, just the opposite of what is to be expected if plural is the marked category (see Table 2).

Table 2. The Marshallese Demonstratives

Person categories serve as labels because of the close parallel with pronoun and subject marker distinctions: bok in ‘this book close to you and me’, bok e ‘this book close to me’, bok üe ‘that book close to you’; bok kaüe ‘those books close to you’, armej raüe ‘those people close to you’. This basic system is reinforced by a parallel set of emphatic demonstratives (bok üeüe ‘that book right there close to you’, bok kâkaüe ‘those books right there close to you’, armej râraüe ‘those people right there by you’, etc.), and also by sets of person demonstratives, locative demonstratives, and copular demonstratives, all of which have emphatic counterparts. These extensive interlocking sets of demonstratives all reflect the basic set in Table 2, with its unexpected suspension in the singular of a contrast made in the plural. Together they play a key role in noun phrase formation. It seems difficult, or impossible, to view them as in any way peripheral, or transitional between states that better conform to the expected markedness complex.

Even here, the plural has formal characteristics expected of marked members. Both the ‘1 incl’ and the ‘2’ forms have greater phonetic substance in the plural than in the singular, consisting as it does of kV- and rV- accretions to the nonplural. Also, the ‘1 excl’, ‘3’, and ‘remote’ forms have the full-consonant onsets in k and r, in contrast with the semi-consonant onset of /y/ in the nonplural. Thus this example from Marshallese is not one that we should identify as a counterexample to the strong tendency elsewhere for plural to be the marked category. Instead, we must note an incongruence in one of the associated tendencies in the "complex set of notions."

In most of the Micronesian sister languages I have examined (Gilbertese, Ponapean, Mokilese, Trukese, Woleaian, Ulithian—all except Kusaiean) there is a partially cognate demonstrative system distinguished by person and number, with a k element throughout the plural, but no human distinction in either singular or plural. Distinguishing human in demonstratives thus appears to be a Marshallese innovation whose motivation is obscure. The only possibly parallel phenomenon I am able to relate it to at present is the distinctive treatment given human objects of transitive verbs in the plural. The 3p object pronoun (er /yér/ ‘they’) must be used for human referents in anaphoric contexts, whereas singular human referents in parallel contexts are represented by an object marker suffixed to the verb, just as are both singular and plural nonhuman referents (see Bender 1984:447–448). It may not be coincidental that this plural pronoun has an r element distinguishing it from the 3s object pronoun e /yéy/. Both, like the 1^st and 2^nd object pronouns, are used only for human referents. Use of the 3s object pronoun in parallel anaphoric contexts is unusual, and thus emphatic.

3. SOME POSSIBLE ABBERRATIONS AND SOME FURTHER QUESTIONS

Less clearly central to the tradition set forth in section 1. is Chafe’s (1970) work on semantics, which gives overt mention to features only when they are present. That they are intended as privative, however, is made clear by his formalisms, as for example, "V - - > > state: The fact that the arrow has a broken shaft means that its application is optional. The fact that it has a double head means that it must be read is further specified as . . . This rule says that a verb, abbreviated V, may be specified optionally as a state. By implication, if a verb is not so specified it is a nonstate or, informally, a ‘happening’ or event." Although he does not mention Jakobson, at one point he indicates familiarity with the work of Hjelmslev, and in discussing the features by which nouns may be further specified, says that "A human noun is given ‘gender’; that is, its sex is specified. There are certain complexities in this area, but I shall assume here that a human noun, if not otherwise specified, is understood to be what we would normally term masculine. In other words, I shall take masculine to be the unmarked state of a human noun, and feminine, then, to be the marked state. In general, not knowing the sex of a human noun, we treat it conceptually as male." He thus intends not only the concepts "marked" and "unmarked," but also that a facultative relation exists between them. His work thus can be counted within the tradition under discussion here. The inventory of features he proposes for the further specification of nouns consists of count, potent, animate, human, feminine, and unique. Although developed in an English-centered discussion, this inventory seems intended for more universal application. None of these features needs to be replaced for Onondaga when he applies his model to that language, although additional features are added for the inflection of Onondaga nouns and the specification of pronouns—features of person (first, second, inclusive), number (plural, dual), and case (dative), for example.

Another approach, one referred to as componential analysis or semantic decomposition and one that may at first glance appear similar because of its use of privative features, is presented by O’Grady and Dobrovolsky in their textbook entitled Contemporary Linguistics. It is said to be "most useful for uncovering and representing similarities among semantically related words . . . A few simple features allow us to express the similarities and differences among subclasses of people—men, women, boys, and girls" (O’Grady 1997:251–252). An accompanying figure for these four nouns shows all four as being +human, two as being +male, and two as being +adult. (Those not labeled +male and +adult are labeled –male and –adult respectively.) The most recent edition adds a parenthetical statement, "Nothing depends on the choice of feature names here; the analysis would work just as well with the feature + female as + male" (251). Presumably this extends to the choice of "adult" over "young."

An exercise at the end of the chapter asks the reader to attempt to provide the semantic features associated with dog, puppy, cat, and kitten, and also asks how are the pairs dog–puppy and cat–kitten are different from man–boy and woman–girl. The instructor’s manual to the 2^nd edition (Aronoff et al. 1993:39) gives a solution that shows dog and cat as +adult, and puppy and kitten as –adult, and answers the second question by saying that "whereas man and woman are usually understood to be [+adult], dog and cat can be used for either [+adult] or [-adult] animals. Aside from the validity of this distinction, it is ironic that dog and cat are here characterized as what in markedness tradition would be the unmarked members, and puppy and kitten as marked, but with the specification [-adult]. From a markedness point of view, this is evidence that the feature has been wrongly named, and that it should be instead [+young].

Whether the human situation differs basically--that man and woman are usually understood to be [+adult] and cannot be used for either [+adult] and {-adult] humans--is open to some question. "Is a girl a special kind of woman, or a woman a special kind of girl?" and "Is a boy a special kind of man, or a man a special kind of boy?" might be better ways to state the question. "Young woman" and "young man" are certainly common enough appellations, and fairly neutral in connotation, whereas "old girl" and "old boy" are narrow, dissonant, and pejorative by comparison. Although this question clearly needs further study, classic markedness theory will probably sustain the view that the terms boy and girl, as well as puppy and kitten, are marked in comparison with man, woman, dog, and cat, respectively, and that the feature involved is thus more properly [+young].

Clearly, componential analysis as here presented is separate from and seemingly uninfluenced by markedness tradition. "This approach," it is said, "has long been used to analyze the meanings of certain types of nouns in terms of semantic features" and has as an obvious advantage that "it allows us to group entities into natural classes (much as we do in phonology)" (O’Grady 1997:251). Thus the use of features such as "male" and "female" rather than "masculine" and "feminine" may well be intentional, as part of a quest for what is natural biologically rather than linguistically. On the other hand, where any distinction can be made, markedness tradition is concerned with how semantic distinctions are treated grammatically and lexically, rather than how they inhere in nature, paralleling the distinctions that sometimes appear between folk taxonomies and scientific classifications.

The distinctive feature analysis of grammatical and lexical categories was modeled originally on the distinctive feature analysis of phonological categories/entities/classes, and has been further influenced by the new deployment given features by Chomsky and Halle (1968) in The Sound Pattern of English, and by the new contextual definitions given to "marked" and "unmarked" within their system. No longer do features simply identify what distinguishes the phonemes of a language from each other. They are given the additional function of spelling out what is nondistinctive as well—all the contextual phonetic detail—the allophones. Although still referred to as phonological "distinctive features," they have become in effect "phonetic features" in this latter interpretive role. Although it is possible to say that [+nasal] is the unmarked value for vowels when they precede [+nasal] consonants, but that [+nasal] is the marked value for such consonants because they bear this value in a context in which [-nasal] consonants can also occur, it is important to recognize that this diverts markedness from its original mission of dealing only with what is distinctive.

Some authors have attempted to make grammatical and lexical distinctive features contextual in parallel fashion, to make them into semantic features with nondistinctive interpretive functions. Thus Bierwisch . . .

[STILL TO BE COMPLETED]

As a case in point, we can look at the question as to how the category neuter ought to be treated. Chafe’s features reflect a commonly held view of the natural order in which the further specifications of animate, human, and feminine are built upon an inanimate, unmarked base. Although "it" may be count and (if an animate-like force of nature) potent, it is not specified as human, and certainly not as feminine. The following system is implied.

But is this what we find in language? Peinovich (1979) analyzes the grammatical gender of Old English nouns as follows "in accordance with Bierwisch’s universal markedness convention. The masculine nouns are the most numerous in the language, and since it is their inflectional system [that] survives into ME and NE, they will be designated the unmarked gender" (Peinovich 1979:57). . . .

[STILL TO BE COMPLETED]

REFERENCES

Aronoff, Mark, Michael Dobrovolsky, and Jane Rees-Miller. 1993. Instructor’s manual to accompany Contemporary Linguistics: An introduction, ed. by WilliamO’Grady, Michael Dobrovolsky, and Mark Aronoff. New York: St. Martin’s Press.

Bender, Byron W. 1969. Spoken Marshallese. Hawai‘i: University of Hawai‘i Press.

———. 1984. Object marking in Marshallese. In Studies in Micronesian linguistics, ed. by Byron W. Bender, pp. 443–465. Pacific Linguistics C-80. Canberra: Australian National University.

Bierwisch, Manfred. 1967. Syntactic features in morphology: General problems of so-called pronominal inflection in German. In To honor Roman Jakobson: Essays on the occasion of his seventieth birthday, vol. 1, pp. 239–270. Mouton: The Hague.

Chomsky, Noam, and Morris Halle. 1968. The sound pattern of English. New York. Harper and Row.

Hamp, Eric P., Fred W. Householder, and Robert Austerlitz, eds. 1966. Readings in Linguistics II. Chicago: University of Chicago Press.

Benveniste, E. 1966. Structure des relations de personne dan le verbe. In Problèmes de linguistique générale, pp. 225–236. Paris.

Blake, Barry. 1994. Case. Cambridge: Cambridge University Press.

Greenberg, Joseph H. 1966. Language universals. In Current trends in linguistics, vol. 3, Theoretical foundations, ed. by Thomas A. Sebeok, pp. 61–112. The Hague: Mouton.

Haiman, John. 1980. The iconicity of grammar: Isomorphism and motivation. Language 56:515–540.

———. 1985. Natural syntax. Cambridge: Cambridge University Press.

Hjelmslev, Louis. 1953. Prologomena to a theory of language. International Journal of American Linguistics, Memoir 7 (Indiana University Publications in Anthropology and Linguistics). Baltimore: Waverly.

Householder, Fred W. 1971. Linguistic speculations. Cambridge: Cambridge University Press.

Jakobson, Roman. 1932. Zur Struktur des russischen Verbums. Reprinted in Hamp et al., pp. 22–30.

———. 1936. Beitrag zur allegemeinen Kasuslehre. Reprinted in Hamp et al., pp. 51–89.

———. 1966. Implications of language universals for linguistics. In Universals of language, ed. by Joseph Greenberg, 2d ed., 263–278. Cambridge, Mass.: MIT Press.

Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University Press.

Matthews, P. H. 1974. Morphology: An introduction to the theory of word-structure. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press.

———. 1991. Morphology. 2d ed. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press.

O’Grady, William, and Michael Dobrovolsky, eds. 1997. Contemporary linguistics. 3^rd ed. (U.S. edition prepared by Mark Aronoff.) New York: St. Martins Press.

Peinovich, Michael P. 1979. Old English noun morphology: A diachronic study. North-Holland Linguistic Series 41. New York: Elsevier North-Holland.

[back to list of essays]