CHAPTER SIXTEEN

Remarks on Markedness

Linda R. Waugh

One of the most important issues in phonology, which was addressed directly by one of the papers in this volume (Houlihan and Iverson) and only indirectly by others, is that of markedness.¹ Markedness follows directly from the nature of opposition: opposition is a binary relation of mutual implication where the two poles of the opposition educe one another and where they are in a hierarchical arrangement vis-à-vis one another, which hierarchical arrangement is given by the terms marked and unmarked. Phonological elements, then, just as semantic elements, far from being disparate and equal, are hierarchically organized—they set up equivalences but they do not create equals. In both phonology and semantics, the marked term is opposed to the unmarked term by its constraining, focusing character, by its concentration on a certain characteristic. The constraining, focusing character of the marked term of any grammatical opposition is directed toward a more closely specified conceptual item—thus in the general meanings of coupled grammatical categories one of them signals a certain grammatical concept which the other one leaves unsignalled. In phonology, in the distinctive features, the marked term is opposed to the unmarked one by a closer concentration on a certain, whether positive or negative, perceptual sound property polar to that of the unmarked term; and accordingly it is characterized by a restrictive selection of contextual conditions of occurrence (whether a sequential or concurrent context). In this respect, then, in the obstruents, as was pointed out in Houlihan and Iverson, who based their remarks on implicational statements by Trubetzkoy (1969) and Jakobson (1968), voicing is marked whereas unvoicing is unmarked. Similarly, continuancy is marked in the obstruents (e.g., fricatives) whereas non-continuancy is unmarked (e.g., stops). Aspiration (one of the forms of tenseness) is marked with respect to non-aspiration (laxness),² and so forth.

One thing which it is important to understand is that the concept of markedness applies at the level of features and not at the level of segments. In other words, if we say that /n/ is marked vis-à-vis /t/ then that is a shorthand for saying that /n/ is markedly nasal while /t/ is non-markedly oral (non-nasal). In other words, if one were to ask whether /k/ is marked or unmarked, the question would have no sense, for /k/ is marked for being compact (velar) but unmarked for being non-continuant, non-nasal. Similarly, to ask what the marked counterpart of /t/ is has no sense because /n/ is its marked counterpart with respect to nasality, /d/ its marked counterpart with respect to voicing, /p/ and /k/ its marked counterparts with respect to gravity and compactness respectively, and so forth.

The ways in which these statements have been formulated reveals another characteristic of markedness—it is context-sensitive. It is not the case, for instance, that voicing is marked everywhere, but rather that it is marked in obstruents but unmarked in vowels. Thus, /+ voice/ in the concurrent context of /+ consonantal/ is marked whereas /+ voice/ in the concurrent context of /+ vocalic/ is unmarked. Other examples of context-sensitivity include the markedness of compactness in consonants—/k/ and /t/³ are marked as against /t/ and /p/—while compactness in vowels is unmarked—/a/ vs. /u/, /i/. Similarly, tenseness is marked in consonants and unmarked in vowels.

These two context-sensitive types of markedness point up another important phenomenon—namely that the distinction between vowel and consonant is a fundamental one. Markedness values are often reversed within the two systems. (But not always—nasality is marked both in vowels and in consonants.) And even within one of these two major categories markedness reversals can occur: thus in the consonants, within the unmarked diffuse consonants, gravity is marked—the bilabial stop is marked in contradistinction to the dental stop. However, within the compact consonants, gravity is unmarked: the velar stop is unmarked in contradistinction to the palatal stop. With respect to this it should be pointed out that markedness relations have often been based on the order of acquisition of features in child language and this has sometimes been adduced as evidence for the unmarkedness of the bilabial stop (i.e., gravity with diffuseness) vs. the dental stop (i.e., acuteness with diffuseness). However, if one looks at consonantal systems around the world, it is generally the dental series which is the more elaborated, i.e., the less focused, and the more frequent in terms of lexical and text frequency.

A further characteristic which is important in terms of markedness is the difference between distinctiveness and redundancy. In English and French, for example, stops and fricatives (e.g.,p,f,t,s,k,š vs. b,v,d,z,g,ž) are distinctively opposed as tense to lax (as shown by, e.g., Malécot 1970, 1977 and Fischer-Jørgensen 1954, 1972), while voiceless vs. voiced is redundant. The redundant nature of the latter is particularly apparent in cases of assimilation where tense ~ lax remains as distinctive and voiced~voiceless is assimilated: cf. French acheter with tense and voiceless [š] vs. à jeter with lax and voiceless [ž]. In French in certain contexts there may be voicing of the ‘unvoiced’ stops or devoicing of ‘voiced’ stops, but since tense~lax is the distinctive opposition, the pairs are still differentiated. In English, the tenseness of the tense stops, for example, is implemented by various phonetic correlates—aspiration, length, intensity, etc.—depending on its position in the word. (In general, the tense stop is stronger the closer it is to ‘strong position’—either word-initial or syllable-initial before the stressed vowel.) And this is corroborated by the fact that its lax opposite is correspondingly weaker in all of these positions. The phonemes /p/, /t/, /k/ then are markedly tense (and redundantly voiceless)— and all of their specific phonetic manifestations are as it were rule-governed. In certain Swiss German dialects the only distinction is tense~lax, with both phonemes basically voiceless. Russian, on the other hand, displays a voiced ~ voiceless distinctive opposition with only minimal reinforcement by tenselax. In such cases, it is also important to realize that the markedness values are again reversed. In languages with a voiced voiceless opposition, it is voicing which is marked while in languages with a tense~lax opposition it is tenseness which is marked. This is borne out not only by child language studies, implicational universals, frequency counts, historical changes, assimilative effects, etc., but also by what happens in languages which use both oppositions autonomously with a three-termed system. In such languages, it is the marked combination of marked tenseness with marked voicing which is unused; i.e., such languages have unmarked laxness with unmarked voicelessness unmarked laxness with marked voicing, and marked tenseness with unmarked voicelessness: Only rarely does a four-term system occur: It is such data which show that Voice Onset Time (so-called VOT), while helpful in discerning certain phonetic differences between languages, is not a primitive since it often combines the effect of voiced ~ voiceless with tense ~ lax.

In some cases, however, it is not clear which feature is distinctive and which is redundant—and then it is the combination (or syncretism) of the two features which is distinctive; there is no direct marked~unmarked relation. Such examples include the typical five-vowel system with two acute non-flat vowels, two grave flat vowels, and one compact vowel:

For some languages, but not all, it is the syncretism of acuteness and non-flatness which is distinctively opposed to the syncretism of gravity and flatness. In languages with bifurcations of such systems—i.e., with flat acute vowels and/or non-flat grave vowels—it becomes clear that in flat vowels acuteness is marked while in grave vowels non-flatness is marked. Furthermore, the /a/ (the unmarked vowels) of such systems is neither acute nor flat. In such cases, then, there is an equipollent opposition between the acute (marked) non-flat (unmarked) and grave (unmarked) flat (marked):

Neither series is unmarked vis-à-vis the other.

Cases like these syncretisms also result when one of the features in question is distinctive and the other redundant. In the examples mentioned above, a lax and thus unmarked stop is redundantly voiced (marked) whereas the tense and thus marked stop is redundantly voiceless (unmarked). Likewise, the optimal stop (unmarked for being non-continuant) is non-strident (marked) whereas the optimal fricative (marked for being continuant) is strident (unmarked). Bifurcations of these combinations into strident non-continuants (affricates) or non-strident continuants (weak fricatives) are rarer in languages of the world. Apparently, the differential marking of the phonemes—their syncretism —is perceptually and systematically more salient and thus unmarked.

There is also a tendency in language for the non-accumulation of marks, i.e., for the superposition of markedness on unmarkedness. This is borne out by the data on the scission of the syncretism of the tenseness feature and voicing feature mentioned above: the doubly marked /d^h/ (voiced and tense) is rarely found. Furthermore, the marked compact consonants only rarely are subdivided into grave vs. acute (velar vs. palatal) while the unmarked diffuse consonants are so subdivided, near-universally or perhaps universally. In other words, if the two features are autonomous (both distinctive), very often three-way systems are found and the phoneme which is lacking is that which exhibits the accumulation of marks. Furthermore, no system evidences the accumulation of marks (e.g., a /d^h/) without the other three possibilities. And no system displays the opposition grave ~ acute in compacts without at the same time having the opposition grave ~ acute in the diffuse phonemes. The (marked) nasal consonants tend to be less elaborated than their (unmarked) non-nasal counterparts (especially the stops). Moreover, nasality is almost universally distinctive in the (unmarked) consonants, whereas it is much more rarely distinctive in the (marked) vowels. And there is a general tendency for consonantal systems to be more elaborated (i.e., to use more features distinctively) than vocalic systems.

The work of Thomas Gamkrelidze and Irina Melikišvili has established that for obstruents the opposition of voiced and voiceless is most prevalent in the (unmarked) dentals, whereas the labials (marked for gravity) may have only the voiced series and the velars (marked for compactness) may have only the voiceless series. Such typological observations await explanation from the systematic analysis of markedness effects in combinations.

With respect to the relation of markedness and neutralization, it is not always the case that the unmarked segment always stands in the place of neutralization. In French, for example, while both [e] and [e] can occur in open syllables, only [e] can occur in a closed syllable,⁴thus producing alternations of the sort premier ~ première. Yet, in the vowels, tenseness is unmarked (as shown by the prevalence of the typical five-vowel system) while laxness is marked. Thus, while there is a tendency to have the unmarked in the place of neutralization, such is not always the case. Furthermore, German, in at least some of its dialects, has the following inventory:

Thus, in both initial and final position, the tense ~ lax opposition is neutralized (or better, an ‘incomplete’ phoneme occurs, being deprived of the tense ~ lax opposition, with the lax contextual variant of this incomplete phoneme in initial position and the tense variant in final position.). In addition, as it happens, the non-distinctively tense -s# may correlate with both the distinctively tense -s- or the distinctively lax -z-, depending on the morpheme—in other words, the neutralization has grammatical consequences. Furthermore, the interrelation between markedness, productivity and neutralization should not be seen in absolute terms. In Russian, before /e/ there is productive neutralization of the opposition between palatalized and non-palatalized consonants: only the palatalized (sharp-marked) series regularly occur. Non-palatalization is a marker of foreignisms, of special vocabulary like interjections, acronyms, names of letters of the alphabet, etc. In other words, it performs a special, stylistic function. The regular, productive way of making a foreign word native or of situating a word within the normal stock of vocabulary of Russian is through the use of a palatalized consonant before /e/. Non-palatalization signals that the word is of special status, while palatalization does not. Thus, before /e/ the marked (sharp) consonants occur; the presence of the unmarked (non-sharp) consonants is a marker of the special status of the word itself. Markedness then assumes an important role—in particular, it is only in such grammatically marked elements as interjections, foreignisms, acronyms, or alphabet names that the combination of the unmarked non-palatalized consonant + /e/ occurs.

It seems to me then that further research in phonology will most profitably take into account the problem of markedness, as well as the very interesting question of the interrelation between markedness in sound systems and markedness in grammatical systems.