Читать книгу The Handbook of Speech Perception - Группа авторов - Страница 74

The theoretical foundations for features come from linguistics, and in particular from phonology (the study of the sound systems of language). Phonological evidence for features as a theoretical construct is based on consideration of (1) the nature of the phonological inventories across languages; (2) the phonological processes of languages synchronically, that is, at a particular point in time; and (3) the processes by which sound inventories in language change historically, that is, over time.

To begin, it is assumed that the sound segments of language are composed of a smaller set of features or parameters of sound (at this point, we will not focus on whether the features are acoustically or articulatorily based, nor will we describe them in detail). These parameters reflect the fact that the phonetic segments (phonetic categories) of language can be further broken down into general classes. These classes are attributes that characterize how a sound may be produced or perceived, including whether the phonetic segment is a consonant or vowel; its manner of articulation (e.g. stop, nasal, fricative); where in the mouth the articulation occurs, for example, at the lips (labial), behind the teeth (alveolar), or at the velum (velar); and the laryngeal state of the production (e.g. voiced or voiceless). Thus, segments (phonetic categories) that share features are closer in articulatory or acoustic space than those that do not. As we shall see, this has ramifications for the performance of listeners in the perception of speech.

The basic notion that underlies the study of phonology is that sound inventories of language are not composed of a random selection of potential phonetic categories. Rather, sound segments tend to group into natural classes reflecting shared sets of feature parameters, for example, manner of articulation (obstruent, continuant, nasal), place of articulation (labial, alveolar, palatal, velar), or the state of the glottis (voicing). Speech inventories across languages also tend toward symmetry. For example, a language inventory may have voiced and voiceless stop labial, alveolar, and velar consonants, as in English, but would typically not have an inventory consisting solely of a voiceless labial, voiced alveolar, and voiceless palatal stop consonant.

Synchronically, languages have phonological processes (called morphophonemic rules) in which phonemic changes occur to morphemes (minimal units of sound–meaning relations) when they combine with other morphemes to form words. Such changes are typically systematic changes to particular features. For example, in English, a plural morpheme {s} is realized phonetically as an [s] when it is preceded by a word ending in a voiceless obstruent, for example, book [s]; it is realized as a [z] when it is preceded by a voiced obstruent, for example, dog [z]; and it is realized as the syllable [əz] when it is preceded by a fricative, for example, horse [əz]. Interestingly, the same morphophonemic processes occur in English for possessives, for example, Dick’s book, Doug’s book, Gladys’ book, and for the third singular, for example, he kicks/jogs/kisses.

Historically, language systems change over time. Again, such changes typically involve changes in features. Moreover, many changes involve changes not to one phonetic segment in the language inventory but to a class of sounds. Two such changes are the Great Vowel Shift and Grimm’s law. In the Great Vowel Shift, as Modern English developed from Middle English, features of vowels systematically changed: low vowels became mid‐vowels, mid‐vowels became high vowels, and high vowels became diphthongs. Grimm’s law affected features of consonants: voiceless stops became voiceless fricatives (a change in the feature [obstruent]); voiced stops became voiceless stops, and voiced aspirated stops became voiced stops or fricatives (both changes of laryngeal features) as German developed from Proto‐Indo‐European.

Taken together, linguistic evidence suggests that features are the building blocks for the sounds of language and for characterizing its phonological processes (Jakobson, Fant, & Halle, 1951). That the speaker or hearer has internalized these properties in the daily use of language in knowing and in applying (even if unconsciously) the phonological rules of the language suggests that features are a part of the human cognitive apparatus. The next section will consider this claim by examining evidence that features are representational units in the functional architecture of language, and underlie speech perception and lexical access processes.