In the Profession

Contrast in Language and Linguistics

Journal of English Linguistics Volume 36 Number 1 March 2008 93-98 © 2008 Sage Publications 10.1177/0075424207311720 http://eng.sagepub.com hosted at http://online.sagepub.com

O

ne of the things that become clearer to me with time is just how complex some of the fundamental notions in linguistics are.1 Another is just how much difficulty beginning students have in understanding them. These two points may not be unrelated: the concepts are not as easy as we tend to present them as being, and therefore a proper understanding of them may indicate a struggle with the notion rather than lack of mental engagement. To illustrate this point, I shall consider the notion of contrast, fundamental to the linguistics of the last century. Implicitly, contrast is still important today, but we can probably look at it in a slightly more nuanced way than we could have envisaged even a few years ago. Consider a simple example of the principle of contrast. The word tin in English is pronounced with a voiceless, alveolar, aspirated plosive. The word din, on the other hand, is pronounced with a partly voiced, alveolar, unaspirated plosive. The word tin conveys a different message from the word din: that is, the two are perceived as being different words, not homonyms like bank ‘side of a river’ and bank ‘financial institution.’ Therefore, the [d] and [th] contrast and belong to separate phonemes. We can construct stimuli based on tin and din but with initial consonants that are intermediate between [th] and [d], and they are perceived as being either tin or din, with an abrupt cut-off point between the two. Categorical perception of this kind leads to the notion that there are clearly distinct categories with which speakers and listeners operate. In other words, the notion of contrast is also a perceptual notion, one that is psychologically real. This notion of contrast transfers to applications of linguistics, too. English has a nonoptimal orthographic system because (among many other reasons) the letters can represent both [θ] and [ð], so that the spellings of earthy and worthy do not differentiate between sounds which contrast in words like mouth (noun) and mouth (verb). The ideal orthographic system would be phonemic. Linguists who are called upon to invent orthographies for unwritten languages tend to stick to this general principle. The principle is also invoked in language teaching, with minimal pairs used to encourage L2 learners to perceive and produce the distinctive (or contrastive) elements of the language they are learning. Books for ELT with titles such as Ship or Sheep (Baker 1981) may sound slightly old-fashioned now, but the method they instantiate is still widely used in the L2 classroom. The proof of the negative in Saussure’s famous dictum that “Dans la langue il n’y a que des differences” (Saussure 1916, 166) also comes to us from the languageteaching sector. Japanese speakers attempting words like probably appear, to English native speakers, to say something like /plaɐbəbri/. If this were true, it would represent 93

94

Journal of English Linguistics

a considerable feat on the part of Japanese learners of English: not only would they have to know which sound appeared in which word, they would also have to get it systematically wrong. The truth is much more prosaic. Japanese speakers say the same thing for /r/ and for /l/. Our perceptions make the difference. When we expect /r/, we say “that isn’t an /r/, therefore it must be /l/” and vice versa when we expect /l/. Our categorical perception is inherently a negative perception. We do not hear what we expect, so what we hear must be in the alternative category. It is the differences that are important. All these examples lead us to a picture of contrast which has a vital role to play in language: it is explanatory, it appears to correlate with actual behavior, and it is fundamental to the operation of language because it allows speakers to create new words—large numbers of them—which speakers do not confuse with each other. Yet the notion of contrast is not as straightforward as these examples might suggest. Let’s consider some potential problems. The first problem is the idea that contrast is an either/or matter: [t] and [d] contrast in English and Japanese, but not in Yingkarta and Passamaquoddy. There is no intermediate step, no “sort of contrast.” This is in line with the kind of phonological problem that first-year undergraduates in linguistics usually get. A single minimal pair in a data set is sufficient to establish the distinctiveness of phonemes. The implication is that minimal pairs are common and a longer list of them is not necessary in the context of an exercise. But consider minimal pairs between [ç] and [x] in German, which can quite literally be counted on the fingers on one hand: pairs like Tauchen ‘small rope’ [ç] and Tauchen ‘diving’ [x]. There are all sorts of problems with such pairs, not the least of which is that most native German speakers feel that [ç] and [x], though audibly different, belong to the same phoneme. So here we have a case where there is only a little bit of contrast, and it is extremely marginal. (For a book-length treatment of this particular problem, see Robinson 2000.) English stress provides another example. Languages like Modern Greek and Russian have phonemic stress: the position of stress in the word is variable and may lead to differences in meaning. In English, the position of stress is variable in the word, but it is largely predictable. In a pair like discount and discount, the stress is predictable on the basis of the part of speech of the word, an instance of predictability that also carries over to the difference between insight and incite, which are not etymologically related. There is one clear minimal pair distinguished by stress in English: billow versus below. Is a single example enough to establish contrast in such instances? Is this contrast or sort-of contrast? We find similar examples in morphology. English has a Latinate agentive suffix -or and a native agentive suffix -er. In some words there seems to be more or less free variation between them: superviser or supervisor, convener or convenor, predicter or predictor, inspector but respecter. As far as I am aware, there is only a single minimal pair: sailor ‘a person’ and sailer ‘a ship’ (Bauer 1990). Is this sufficient to prove contrast, or are we dealing with degrees of contrast?

Bauer / Language and Linguistics

95

This last example brings us to a second problem for notions of contrast, and that is variation. All classical notions of contrast depend on uniform and stable systems, at least implicitly. Yet one thing that has become much clearer since the 1960s is the huge amount of variability that is present in any real language usage. Different linguists treat this variation differently, ignoring it or making it central, but we must agree that variation complicates the notion of contrast. Take the example of the growth of so-called [θ]-fronting in English since the 1960s. This gives rise to pronunciations such as [brvə] (brother), [fiŋk] (think), [fri] (free or three), and so on. The difficulty is that [θ]-fronting is variable, even within the speech of individuals or individuals in given styles. So, in effect, we have two [f]-sounds which are not phonologically equivalent: one that—in the speech of the individual or in the community—alternates with [θ] and one that does not. Yet there is no way of hearing which one we have in a pronunciation like [fri]. This makes contrast a bit messy, but it probably does not threaten the fundamental notion. Rather worse, I think, are examples like [sphɔt] (support) and [skhm] (succumb), where we appear to have contrast with [spɔt] (sport) and [skm] (scum). But because there is variation between [sphɔt] and [səphɔt], but not between [spɔt] and *[səpɔt], this apparent contrast does not count as contrast. This certainly seems to coincide with speakers’ feelings about these pairs. Again, it seems to imply degrees of contrastiveness. The same is probably true of the standard examples of systematic non-contrastiveness of contrastive elements: lexically conditioned suspension of contrast, neutralization, and syncretism. By lexically conditioned suspension of contrast I mean examples like [e]conomics or [i]conomics, n[ai]ther or n[i]ther, [θ]ither or [ð]ither. I assume it is a fact about these lexical entries that, for example [ekənamiks] is not an error but an alternative for [ikənamiks] (or vice versa if you prefer), while [mezəlz] is at best an error and at worst totally incomprehensible for [mizəlz]. Neutralization is the failure of contrastive elements to contrast in a particular environment. Classic examples from English are the lack of contrast between DRESS and TRAP2 before /l/ in New Zealand English, so that Alan and Ellen become homophones; the lack of contrast between DRESS, FACE, and TRAP before /r/ in many North American varieties, so that merry, Mary, and marry become homophones; and the lack of contrast between [b] and [p] immediately following an initial [s] in English, so that only [sp] clusters are heard. There is a large literature on the problems involved here, including the nature of the output (sometimes a compromise, sometimes one sound or the other, sometimes a new sound—see e.g., Lass 1984) and the difference between neutralization and an accidental gap. Such instances do not prove that contrast as a linguistic notion is in doubt, but they do seem to indicate that some contrasts are much more thoroughly integrated into the language system than others. Syncretism is the morphological parallel to neutralization. Russian nouns, for instance, have singular and plural forms in six cases, but the nominative and accusative

96

Journal of English Linguistics

singular forms are the same for neuter nouns, and the dative and prepositional singular are the same for feminine nouns. Thus there are systematically fewer forms than there are slots in the inflectional paradigm. If the same slots were syncretized for all genders, we would say that there were fewer cases in Russian (i.e., fewer contrasts), but each gender shows syncretism in a different position in the paradigm. Another example involves what Chomsky (1964) identifies as breaches of linearity. Chomsky’s example is the contrast between writer and rider in varieties of English which flap intervocalic /t/ and /d/ and yet which keep the two words distinct. He transcribes the phonetic forms as [rajɾər] versus [rajɾər], and he claims that phonemic contrast in position 4 in the word is reflected by phonetic difference at position 2, which breaks a fundamental principle of phoneme theory. Descriptive linguists are increasingly finding that the exponents of phonological contrast can be found distributed across a wider range of phonetic material than was previously realized, even by Chomsky. Analyses done within a neo-Firthian framework are particularly good at drawing attention to such cases. One example is provided by Kelly and Local (1989, 159), where they look at the writer/rider question and show, for instance, that for their speakers the quality of the initial /r/ is different in the two cases, as are the quality and length of the diphthong and the quality of the final vowel, as well as the quality of the tap. If there is contrast here (which there presumably is, because the message is not the same), it is not neatly localized at all but rather contrast at the level of the word-form. We might want to make similar claims about instances such as the following. (1) Each morning they hit a ball for hours. (2) Last night they hit a ball through my window.

In morphosyntactic terms we might want to say that the two occurrences of they hit a ball are distinguished by a tense feature, but in textual terms, the difference between the two utterances is signaled by material elsewhere in the context, and possibly quite temporally distant from the presumed location of the morphosyntactic feature. Imperceptible contrast also provides a theoretical problem. The basis of contrast is that it is perceptually or psychologically real, that contrasting elements are perceived as distinct, that it therefore is important in creating orthographic systems, and the like. But there are instances where people make contrasts of which they are unaware and which they cannot hear. There has been a lot of discussion of the case of apparent instances of neutralization where speakers actually maintain a contrast of which they are not aware. This has been argued recently for the case of final voiced and unvoiced plosives in languages like German, Catalan, and Polish, traditionally described as having only voiceless plosives in word-final position. The plosives may be voiceless, but the distinction may nevertheless be carried by the length of the preceding vowel. The problem is that speakers and hearers of these languages are apparently unaware of the distinctions they (and others) are making (see e.g.,

Bauer / Language and Linguistics

97

Dinnesen 1985). There are also examples much closer to home. I have a distinction between band and banned, but I would never be able to say which of these sounds (if either) appears in a different word like mandatory. I also have a distinction between told and tolled. This one I can hear if I try hard enough, but it is not a distinction like that between toad and tod, which is clearly central in my system (see Bauer 1995). Examples like these seem to mock the notion of contrast. They illustrate contrast without any of the trappings of contrast: the contrast is there for a machine but not for the human speaker and listener. The converse of this situation is contrasting elements that never actually contrast. Here I am thinking about the classic puzzle of /h/ versus /ŋ/. Although we call these different phonemes, there is never any situation in which we can replace one with the other and get a different message. One situation that remains obscure to me is the situation with regard to synonyms. Assume that somebody is washing the floors. They might say (at least in some dialects): (3) I put my mop in the bucket.

or (4) I put my mop in the pail.

While bucket and pail would not necessarily be synonymous in all contexts (e.g., He filled the front-end-loader’s {bucket; *pail} with gravel), here the two sentences seem to refer accurately to the same set of possible events, to be stylistically equivalent, not to differ in connotations, etc. Since the notion of contrast demands different messages, it would seem that bucket and pail do not contrast in these examples. This is odd. Both contrast with cupboard; the phonology is different in the two cases. Since synonyms are rarely synonymous across the whole range of their use, we again seem to be in a position where some words contrast more than others do. Similar problems for the notion of contrast are raised by homonymy. Suppose we are asked what we did last night. We might reply (at least in Britain or in New Zealand): (5) [Being a sports fan] I watched the cricket.

or (6) [Being an entomologist] I watched the cricket.

Here the messages are different, though the form is identical. Can we say that the two iterations of I watched the cricket are in contrast with each other? If we think contrast involves messages, we should; if we think it involves form, we should not. Perhaps we need to expand the notion of contrast. What we see with the examples of synonymy and homonymy is that contrast may involve different messages (homonymy) or different form (synonymy) without any difference at the second level. Again this implies that not all contrasts are equal.

98

Journal of English Linguistics

So what can I conclude? Of course, contrast is important in linguistics, just as it is in home decoration and cookery. But I think that some contrasts are more fully built into the system of particular languages than are others. The contrast between /s/ and /t/ is much more central to the system of German than is the contrast between /ç/ and /x/. Contrast is a matter of how strong the contrast is, not just a matter of whether there is a difference. If we accept this proposition, then it ought to make a difference for L2 teaching. The whole notion of distributed contrast suggests that we should allow contrast to emerge from knowledge of words, and we should certainly spend more classroom time on some contrasts than on others. This is entirely in line with recent exemplarbased theories of the way in which language functions. Although all this may make the notion of contrast seem complex and mysterious, it should be some consolation that enough progress has been made in linguistics for us to be aware of the problems and to have started to build solutions to these problems into our theories. Applications of the theories always lag behind the evolution of theory, but I suspect that we can look forward to the application of exemplar theory in language teaching. In the meantime, if we ourselves recognize the complexity of the notion of contrast, we might have more sympathy for our students when they seem to have difficulty with it. Laurie Bauer Victoria University of Wellington

Notes 1. This paper was presented on the occasion of the retirement of Jim Miller from the Chair of Linguistics at the University of Auckland in June 2007. I have removed the references to Jim’s career (and to my own) and tried to make the style more appropriate for a written discussion paper; I have also shortened some of the discussion. 2. I use the names of lexical sets introduced by Wells (1982), whereby DRESS denotes the set of words containing the same vowel as the word dress, and by metonymy, the vowel itself.

References Baker, Ann. 1981. Ship or Sheep?: An Intermediate Pronunciation Course. 2nd ed. Cambridge, UK: Cambridge University Press. Bauer, Laurie. 1990. Level Disorder: The Case of -er and -or. Transactions of the Philological Society 88:97–110. ———. 1995. Another Marginal Phoneme of English. In Proceedings of the XIIIth International Congress of Phonetic Sciences, edited by Kjell Elnius and Peter Branderud, 3:354-357. Stockholm: KTH and Stockholm University. Chomsky, Noam. 1964. Current Issues in Linguistic Theory. The Hague; Paris: Mouton. Dinnesen, Daniel. 1985. A Re-examination of Phonological Neutralization. Journal of Linguistics 21:265–79. Kelly, John, and John Local. 1989. Doing Phonology. Manchester: Manchester University Press. Lass, Roger. 1984. Phonology. Cambridge: Cambridge University Press. Robinson, Orrin W. 2000. Whose German?: The ach/ich Alternation and Related Phenomena in Standard and Colloquial. Amsterdam; Philadelphia, PA: Benjamins. Saussure, Ferdinand de. 1916. Cours de Linguistique Générale. Paris: Payot. Wells, J.C. 1982. Accents of English. 3 vols. Cambridge: Cambridge University Press.