"Markedness and the Contrastive Hierarchy in Phonology" is a research project directed by B. Elan Dresher and Keren Rice of the Department of Linguistics at the University of Toronto, supported by a grant from the Social Science and Humanities Research Council of Canada Social Science and Humanities Research Council of Canada (SSHRC Research Grant 410-2008-2645, 2008-2011). The project partially supports the research of the principal investigators and seven graduate student research associates. Project meetings have been attended by colleagues and collaborators from other departments at the University of Toronto and from York University, and the project has supported the visits of colleagues and collaborators from the United States, Germany, and the Netherlands.
The project begins with the assumption that contrast is fundamental to phonology. We have been building on the approach to segment structure that we have been pursuing at Toronto for more than a decade, which has come to be known as Modified Contrastive Specification (MCS). In this model, segments have internal structure of varying degrees of complexity: roughly, some segments are literally less marked, hence less complex, while others, the marked segments, have more complex structures. Since more complex representations are permitted only if they are needed to establish a contrast with a less complex structure, the theory of MCS leads us to expect a relation between the amount of segmental complexity a system allows and the number and nature of contrasts it has.
Contrasts, in turn, are determined by a series of binary cuts over a relevant domain; the ordering of the features in a language (the contrastive hierarchy of features) determines the order of the binary cuts, and hence the scope of a contrast. The number and nature of contrasts that a segment enters into influence, but do not determine, its phonetic realization. It follows that contrast and complexity play important roles in the course of acquisition: complexity at all levels increases monotonically under pressure of increasing contrast in the system. Our approach thus stresses the importance of understanding systems and how they work, as well as the relative abstractness of phonological representations with respect to their phonetic output. Our interest is to find unifying principles which play a role in both segmental and prosodic domains, and which have consequences for the phonology-phonetics interface as well as for the study of child phonology and learnability.
In the current phase of our project, we are investigating the following questions in particular:
Nontechnical Summary | Project Objectives | Detailed Description | The Reseach Team | References
This project description was written in October 2007 by B. Elan Dresher and Keren Rice.
Website updated October 16, 2008
Our proposal deals with the sounds of language. Speech sounds can be studied in two ways. A phonetic study seeks to describe each sound in as much detail as possible. Phonetic study shows that every language uses many different sounds. Not all differences between sounds are equally significant to speakers of a language. For example, in English the first sound in ton is different from the first sound in done. The difference between these sounds is significant in English, and English spelling uses different letters for them. The letter t also appears in the words stick, attic, and twelve, and most speakers of English would agree that they all contain the sound t, as in ton. However, from a phonetic point of view, these t-sounds are all different: in ton, the t is produced with a puff of air, or aspiration, written [th] (square brackets indicate sounds), that is lacking in stun ([t]); in attic, the t is produced with a tongue 'flap', written [D]; the t in twelve is pronounced with lip rounding, written [tw]. All these sounds have distinct symbols in the International Phonetic Alphabet (IPA). However, these differences are not contrastive in English, in the sense that they do not distinguish meanings. To indicate that English [t], [th], [D], and [tw] are all varieties of the 'same' sound, we can say they are all variants of a single phoneme, designated /t/. Phonemes are abstract representations of contrastive speech sounds. Phonology is the study of phonemes and their phonetic realizations, and of how sounds pattern in languages.
From phonetics, we know that sounds are related to other sounds in various ways. For example, [s] and [z] are identical in almost every respect, except that [z] is a voiced sound (made with vocal cord vibration), and [s] is voiceless. To capture these similarities and differences, phonologists propose that sounds should be represented in terms of a set of distinctive features, drawn from articulatory and acoustic parameters. Thus, we can say that [s] and [z] share features that specify place and manner of articulation, but differ in the feature [voiced]. In this case, we assume that [z] is marked as [voiced], but that voicelessness comes about when there is no such marking. Thus, /z/ is the marked member, and /s/ is the unmarked member of the /s/-/z/ contrast. These features are part of the mental representation of each phoneme in the minds of speakers and hearers who know the language.
This kind of markedness refers to the intrinsic content of features. But features also enter into relations with other features. For example, consider a language with only three contrastive vowel phonemes, /a, i, u/. Of these, /a/ is produced with a low tongue position, /i/ is high and front, and /u/ is high, back, and has lip rounding. But these features need not all be specified in order to distinguish the three vowels. Our hypothesis is that specifications are governed by ordering features into a contrastive hierarchy. Thus, if the feature [high] is at the top of the hierarchy, then /i/ and /u/ will be designated [high], thus distinguishing them from /a/, which need receive no mark at all. Now consider /i/ and /u/. They potentially differ in terms of several features, depending on which feature is next on the hierarchy. If [round] is next, then /u/ is specified as [round], and /i/ need not be specified any further. Or, if [front] is next on the hierarchy, then /i/ is specified [front], and /u/ need not be further specified. There are other possible specifications, corresponding to other possible orderings of features.
This method of assigning representations to phonemes thus combines the notions of markedness and the contrastive hierarchy. Our proposal aims to explore these notions, which have consequences that are fundamental to phonological theory. Some questions that we will attempt to answer are the following:
Our proposal thus addresses fundamental questions about the language faculty that are critical for our understanding of learning, perception, and human cognitive abilities.
The main objective of the proposed project is to study phonological features and their activity, with special attention to two fundamental questions:
1. What is consistent and invariant cross-linguistically with respect to features and their activity?
2. What is potentially variable about featural activity within a language and between languages?
The results of our previous research suggest an answer to these questions which we frame as the following hypothesis: Phonological systems display global uniformity and local variability.
By global uniformity we mean that languages, when viewed on a large scale (e.g., languages with large consonantal or vowel inventories) display a certain cross-linguistic consistency; by local variability we mean that individual languages may vary in ways that are not entirely predictable, particularly where the phonology is relatively sparse (e.g., in portions of the inventory where there are few contrasts).
This hypothesis arises from our ongoing research on markedness and contrast in phonology. One of the distinguishing characteristics of our approach is our emphasis on phonological activity as the key diagnostic of the contrasts and markedness relations that are actually operative in a given language. Taking phonological activity as a key concept, we have arrived at a number of surprising results:
Up to now, our focus has been on local variability, that is, on how much languages can differ in terms of contrast and markedness. Now we aim to try to understand what the limits of variation are, by attempting to characterize more precisely what global uniformity is.