Dispersion Theory concerns the constraints that govern contrasts, the phonetic differences that can distinguish words in a language. Specifically it posits that there are distinctiveness constraints that favor contrasts that are more perceptually distinct over less distinct contrasts. The preference for distinct contrasts is hypothesized to follow from a preference to minimize perceptual confusion: In order to recover what a speaker is saying, a listener must identify the words in the utterance. The more confusable words are, the more likely a listener is to make errors. Because contrasts are the minimal permissible differences between words in a language, banning indistinct contrasts reduces the likelihood of misperception.
The term ‘dispersion’ refers to the separation of sounds in perceptual space that results from maximizing the perceptual distinctiveness of the contrasts between those sounds, and is adopted from Lindblom’s Theory of Adaptive Dispersion, a theory of phoneme inventories according to which inventories are selected so as to maximize the perceptual differences between phonemes. These proposals follow a long tradition of explaining cross-linguistic tendencies in the phonetic and phonological form of languages in terms of a preference for perceptually distinct contrasts.
Flemming proposes that distinctiveness constraints constitute one class of constraints in an Optimality Theoretic model of phonology. In this context, distinctiveness constraints predict several basic phenomena, the first of which is the preference for maximal dispersion in inventories of contrasting sounds that first motivated the development of the Theory of Adaptive Dispersion. But distinctiveness constraints are formulated as constraints on the surface forms of possible words that interact with other phonological constraints, so they evaluate the distinctiveness of contrasts in context. As a result, Dispersion Theory predicts that contrasts can be neutralized or enhanced in particular phonological contexts. This prediction arises because the phonetic realization of sounds depends on their context, so the perceptual differences between contrasting sounds also depend on context. If the realization of a contrast in a particular context would be insufficiently distinct (i.e., it would violate a high-ranked distinctiveness constraint), there are two options: the offending contrast can be neutralized, or it can be modified (‘enhanced’) to make it more distinct.
A basic open question regarding Dispersion Theory concerns the proper formulation of distinctiveness constraints and the extent of variation in their rankings across languages, issues that are tied up with the questions about the nature of perceptual distinctiveness. Another concerns the size and nature of the comparison set of contrasting word-forms required to be able to evaluate whether a candidate output satisfies distinctiveness constraints.
Daniel Currie Hall
The fundamental idea underlying the use of distinctive features in phonology is the proposition that the same phonetic properties that distinguish one phoneme from another also play a crucial role in accounting for phonological patterns. Phonological rules and constraints apply to natural classes of segments, expressed in terms of features, and involve mechanisms, such as spreading or agreement, that copy distinctive features from one segment to another.
Contrastive specification builds on this by taking seriously the idea that phonological features are distinctive features. Many phonological patterns appear to be sensitive only to properties that crucially distinguish one phoneme from another, ignoring the same properties when they are redundant or predictable. For example, processes of voicing assimilation in many languages apply only to the class of obstruents, where voicing distinguishes phonemic pairs such as /t/ and /d/, and ignore sonorant consonants and vowels, which are predictably voiced. In theories of contrastive specification, features that do not serve to mark phonemic contrasts (such as [+voice] on sonorants) are omitted from underlying representations. Their phonological inertness thus follows straightforwardly from the fact that they are not present in the phonological system at the point at which the pattern applies, though the redundant features may subsequently be filled in either before or during phonetic implementation.
In order to implement a theory of contrastive specification, it is necessary to have a means of determining which features are contrastive (and should thus be specified) and which ones are redundant (and should thus be omitted). A traditional and intuitive method involves looking for minimal pairs of phonemes: if [±voice] is the only property that can distinguish /t/ from /d/, then it must be specified on them. This approach, however, often identifies too few contrastive features to distinguish the phonemes of an inventory, particularly when the phonetic space is sparsely populated. For example, in the common three-vowel inventory /i a u/, there is more than one property that could distinguish any two vowels: /i/ differs from /a/ in both place (front versus back or central) and height (high versus low), /a/ from /u/ in both height and rounding, and /u/ from /i/ in both rounding and place.
Because pairwise comparison cannot identify any features as contrastive in such cases, much recent work in contrastive specification is instead based on a hierarchical sequencing of features, with specifications assigned by dividing the full inventory into successively smaller subsets. For example, if the inventory /i a u/ is first divided according to height, then /a/ is fully distinguished from the other two vowels by virtue of being low, and the second feature, either place or rounding, is contrastive only on the high vowels. Unlike pairwise comparison, this approach produces specifications that fully distinguish the members of the underlying inventory, while at the same time allowing for the possibility of cross-linguistic variation in the specifications assigned to similar inventories.
Chiyuki Ito and Michael J. Kenstowicz
Typologically, pitch-accent languages stand between stress languages like Spanish and tone languages like Shona, and share properties of both. In a stress language, typically just one syllable per word is accented and bears the major stress (cf. Spanish sábana ‘sheet,’ sabána ‘plain,’ panamá ‘Panama’). In a tone language, the number of distinctions grows geometrically with the size of the word. So in Shona, which contrasts high versus low tone, trisyllabic words have eight possible pitch patterns. In a canonical pitch-accent language such as Japanese, just one syllable (or mora) per word is singled out as distinctive, as in Spanish. Each syllable in the word is assigned a high or low tone (as in Shona); however, this assignment is predictable based on the location of the accented syllable.
The Korean dialects spoken in the southeast Kyengsang and northeast Hamkyeng regions retain the pitch-accent distinctions that developed by the period of Middle Korean (15th–16th centuries). For example, in Hamkyeng a three-syllable word can have one of four possible pitch patterns, which are assigned by rules that refer to the accented syllable. The accented syllable has a high tone, and following syllables have low tones. Then the high tone of the accented syllable spreads up to the initial syllable, which is low. Thus, /MUcike/ ‘rainbow’ is realized as high-low-low, /aCImi/ ‘aunt’ is realized as low-high-low, and /menaRI/ ‘parsley’ is realized as low-high-high. An atonic word such as /cintallɛ/ ‘azalea’ has the same low-high-high pitch pattern as ‘parsley’ when realized alone. But the two types are distinguished when combined with a particle such as /MAN/ ‘only’ that bears an underlying accent: /menaRI+MAN/ ‘only parsely’ is realized as low-high-high-low while /cintallɛ+MAN/ ‘only azelea’ is realized as low-high-high-high. This difference can be explained by saying that the underlying accent on the particle is deleted if the stem bears an accent. The result is that only one syllable per word may bear an accent (similar to Spanish). On the other hand, since the accent is realized with pitch distinctions, tonal assimilation rules are prevalent in pitch-accent languages.
This article begins with a description of the Middle Korean pitch-accent system and its evolution into the modern dialects, with a focus on Kyengsang. Alternative synchronic analyses of the accentual alternations that arise when a stem is combined with inflectional particles are then considered. The discussion proceeds to the phonetic realization of the contrasting accents, their realizations in compounds and phrases, and the adaptation of loanwords. The final sections treat the lexical restructuring and variable distribution of the pitch accents and their emergence from predictable word-final accent in an earlier stage of Proto-Korean.
Derivational morphology is a type of word formation that creates new lexemes, either by changing syntactic category or by adding substantial new meaning (or both) to a free or bound base. Derivation may be contrasted with inflection on the one hand or with compounding on the other. The distinctions between derivation and inflection and between derivation and compounding, however, are not always clear-cut. New words may be derived by a variety of formal means including affixation, reduplication, internal modification of various sorts, subtraction, and conversion. Affixation is best attested cross-linguistically, especially prefixation and suffixation. Reduplication is also widely found, with various internal changes like ablaut and root and pattern derivation less common. Derived words may fit into a number of semantic categories. For nouns, event and result, personal and participant, collective and abstract noun are frequent. For verbs, causative and applicative categories are well-attested, as are relational and qualitative derivations for adjectives. Languages frequently also have ways of deriving negatives, relational words, and evaluatives. Most languages have derivation of some sort, although there are languages that rely more heavily on compounding than on derivation to build their lexical stock. A number of topics have dominated the theoretical literature on derivation, including productivity (the extent to which new words can be created with a given affix or morphological process), the principles that determine the ordering of affixes, and the place of derivational morphology with respect to other components of the grammar. The study of derivation has also been important in a number of psycholinguistic debates concerning the perception and production of language.
The word accent system of Tokyo Japanese might look quite complex with a number of accent patterns and rules. However, recent research has shown that it is not as complex as has been assumed if one incorporates the notion of markedness into the analysis: nouns have only two productive accent patterns, the antepenultimate and the unaccented pattern, and different accent rules can be generalized if one focuses on these two productive accent patterns.
The word accent system raises some new interesting issues. One of them concerns the fact that a majority of nouns are ‘unaccented,’ that is, they are pronounced with a rather flat pitch pattern, apparently violating the principle of obligatoriness. A careful analysis of noun accentuation reveals that this strange accent pattern occurs in some linguistically predictable structures. In morphologically simplex nouns, it typically tends to emerge in four-mora nouns ending in a sequence of light syllables. In compound nouns, on the other hand, it emerges due to multiple factors, such as compound-final deaccenting morphemes, deaccenting pseudo-morphemes, and some types of prosodic configurations.
Japanese pitch accent exhibits an interesting aspect in its interactions with other phonological and linguistic structures. For example, the accent of compound nouns is closely related with rendaku, or sequential voicing; the choice between the accented and unaccented patterns in certain types of compound nouns correlates with the presence or absence of the sequential voicing. Moreover, whether the compound accent rule applies to a certain compound depends on its internal morphosyntactic configuration as well as its meaning; alternatively, the compound accent rule is blocked in certain types of morphosyntactic and semantic structures.
Finally, careful analysis of word accent sheds new light on the syllable structure of the language, notably on two interrelated questions about diphthong-hood and super-heavy syllables. It provides crucial insight into ‘diphthongs,’ or the question of which vowel sequence constitutes a diphthong, against a vowel sequence across a syllable boundary. It also presents new evidence against trimoraic syllables in the language.
Timothy J. Vance
The term rendaku, sometimes translated as sequential voicing, denotes a morphophonemic phenomenon in Japanese. In a prototypical case, an alternating morpheme appears with an initial voiceless obstruent as a word on its own or as the initial element (E1) in a compound but with an initial voiced obstruent as the second element (E2) in a two-element compound. For example, the simplex word /take/ ‘bamboo’ and the compound /take+yabu/ ‘bamboo grove’ (cf. /yabu/ ‘grove’) begin with voiceless /t/, but this morpheme meaning ‘bamboo’ begins with voiced /d/ in /sao+dake/ ‘bamboo (made into a) pole’ (cf. /sao/ ‘pole’). Rendaku was already firmly established in 8th-century Old Japanese (OJ), the earliest variety for which extensive written records exist, and subsequent sound changes have made the alternations phonetically heterogeneous. Many OJ compounds with eligible E2s did not undergo rendaku, and the phenomenon remains pervasively irregular in modern Japanese. There are, however, many factors that promote or inhibit rendaku, and some of these appear to influence native-speaker behavior on experimental tasks. The best known phonological factor is Lyman’s Law, according to which rendaku does not apply to E2s that contain a non-initial voiced obstruent. Many theoretical phonologists endorse the idea that Lyman’s Law is a sub-case of the Obligatory Contour Principle, which rules out identical or similar units if they would be adjacent in some domain. Other well-known factors involve vocabulary stratum (e.g., the resistance to rendaku of recently borrowed E2s) or the morphological/semantic relationship between E2 and E1 (e.g., the resistance to rendaku of coordinate compounds). Some morphemes are idiosyncratically immune to rendaku. Other morphemes alternate but undergo rendaku in some compounds while failing to undergo it in others, even though no known factor is relevant. In addition, many individual compounds vary between a form with rendaku and a form without, and this variability is often not reflected in dictionary entries. Despite its irregularity, rendaku is productive in the sense that it often applies to newly created compounds. Many compounds, of course, are stored (with or without rendaku) in a speaker’s lexicon, but fact that native speakers can apply rendaku not just to existing E2s in novel compounds but even to made-up E2s shows that rendaku as an active process is somehow incorporated into the grammar.
This is an advance summary of a forthcoming article in the Oxford Research Encyclopedia of Linguistics. Please check back later for the full article.
Autosegments were introduced by John Goldsmith in his 1976 MIT dissertation to represent tone and other suprasegmental phenomena. Goldsmith’s intuition, embodied in the term he created, was that autosegments constituted an independent, conceptually equal tier of phonological representation, with both tiers realized simultaneously like the separate voices in a musical score.
The analysis of suprasegmentals came late to generative phonology, even though it had been tackled in American structuralism with the long components of Harris 1944 and despite being a particular focus of Firthian prosodic analysis. The standard version of generative phonology of the era (Chomsky & Halle’s The Sound Pattern of English) made no special provision for phenomena that had been labeled suprasegmental or prosodic by earlier traditions.
An early sign that tones required a separate tier of representation was the phenomenon of tonal stability. In many tone languages, when vowels are lost historically or synchronically, their tones remain. The behavior of contour tones in many languages also falls into place when the contours are broken down into sequences of level tones on an independent level or representation. The autosegmental framework captured this naturally, since a sequence of elements on one tier can be connected to a single element on another. But the single most compelling aspect of the early autosegmental model was a natural account of tone spreading, a very common process that was only awkwardly captured by rules of whatever sort. Goldsmith’s autosegmental solution was the well-formedness condition, requiring, among other things, that every tone on the tonal tier be associated with some segment on the segmental tier, and vice-versa. Tones thus spread more or less automatically to segments lacking them. The condition of well-formedness, at the very core of the autosegmental framework, was a rare constraint, posited nearly two decades before optimality theory.
One-to-many associations and spreading onto adjacent elements are characteristic of tone but not confined to it. Similar behaviors are widespread in long-distance phenomena including intonation, vowel harmony, and nasal prosodies, as well as more locally with partial or full assimilation across adjacent segments. A major discovery, in Mark Liberman’s 1975 MIT dissertation, was that autosegmental tiers have hierarchical structure, with Goldsmith’s autosegments as the terminal elements of those structures.
The early autosegmental notion of tiers of representation that were distinct but conceptually equal soon gave way to a model with one basic tier—called the skeleton or CV tier—connected to tiers for particular kinds of articulation, including tone and intonation, nasality, vowel features, and others. This has led to hierarchical representations of phonological features in current models of feature geometry, replacing the unordered distinctive feature matrices of early generative phonology.
Autosegmental representations and processes also provide a means of representing nonconcatenative morphology, notably the complex interweaving of roots and patterns in Semitic languages.
Harry van der Hulst
The subject of this article is vowel harmony. In its prototypical form, this phenomenon involves agreement between all vowels in a word for some phonological property (such as palatality, labiality, height or tongue root position). This agreement is then evidenced by agreement patterns within morphemes and by alternations in vowels when morphemes are combined into complex words, thus creating allomorphic alternations. Agreement involves one or more harmonic features for which vowels form harmonic pairs, such that each vowel has a harmonic counterpart in the other set. I will focus on vowels that fail to alternate, that are thus neutral (either inherently or in a specific context), and that will be either opaque or transparent to the process. We will compare approaches that use underspecification of binary features and approaches that use unary features. For vowel harmony, vowels are either triggers or targets, and for each, specific conditions may apply. Vowel harmony can be bidirectional or unidirectional and can display either a root control pattern or a dominant/recessive pattern.
When the phonological form of a morpheme—a unit of meaning that cannot be decomposed further into smaller units of meaning—involves a particular melodic pattern as part of its sound shape, this morpheme is specified for tone. In view of this definition, phrase- and utterance-level melodies—also known as intonation—are not to be interpreted as instances of tone. That is, whereas the question “Tomorrow?” may be uttered with a rising melody, this melody is not tone, because it is not a part of the lexical specification of the morpheme tomorrow. A language that presents morphemes that are specified with specific melodies is called a tone language. It is not the case that in a tone language every morpheme, content word, or syllable would be specified for tone. Tonal specification can be highly restricted within the lexicon. Examples of such sparsely specified tone languages include Swedish, Japanese, and Ekagi (a language spoken in the Indonesian part of New Guinea); in these languages, only some syllables in some words are specified for tone. There are also tone languages where each and every syllable of each and every word has a specification. Vietnamese and Shilluk (a language spoken in South Sudan) illustrate this configuration. Tone languages also vary greatly in terms of the inventory of phonological tone forms. The smallest possible inventory contrasts one specification with the absence of specification. But there are also tone languages with eight or more distinctive tone categories. The physical (acoustic) realization of the tone categories is primarily fundamental frequency (F0), which is perceived as pitch. However, often other phonetic correlates are also involved, in particular voice quality. Tone plays a prominent role in the study of phonology because of its structural complexity. That is, in many languages, the way a tone surfaces is conditioned by factors such as the segmental composition of the morpheme, the tonal specifications of surrounding constituents, morphosyntax, and intonation. On top of this, tone is diachronically unstable. This means that, when a language has tone, we can expect to find considerable variation between dialects, and more of it than in relation to other parts of the sound system.
Matthew K. Gordon
Metrical structure refers to the phonological representations capturing the prominence relationships between syllables, usually manifested phonetically as differences in levels of stress. There is considerable diversity in the range of stress systems found cross-linguistically, although attested patterns represent a small subset of those that are logically possible. Stress systems may be broadly divided into two groups, based on whether they are sensitive to the internal structure, or weight, of syllables or not, with further subdivisions based on the number of stresses per word and the location of those stresses. An ongoing debate in metrical stress theory concerns the role of constituency in characterizing stress patterns. Certain approaches capture stress directly in terms of a metrical grid in which more prominent syllables are associated with a greater number of grid marks than less prominent syllables. Others assume the foot as a constituent, where theories differ in the inventory of feet they assume. Support for foot-based theories of stress comes from segmental alternations that are explicable with reference to the foot but do not readily emerge in an apodal framework. Computational tools, increasingly, are being incorporated in the evaluation of phonological theories, including metrical stress theories. Computer-generated factorial typologies provide a rigorous means for determining the fit between the empirical coverage afforded by metrical theories and the typology of attested stress systems. Computational simulations also enable assessment of the learnability of metrical representations within different theories.