Phonological
processes were first systematically studied in the nineteenth century
under the rubric of sound laws relating the various Indo-European
languages. In the twentieth century, attention shifted to a synchronic
perspective, prompted by observations such as Edward SAPIR's
that as part of their grammatical competence mature speakers unconsciously
and effortlessly assign (sometimes radically) different pronunciations
to a lexical item drawn from memory and inserted in different grammatical
or prosodic contexts. For example, in the pronunciation of the word átom American
English speakers "flap" the intervocalic consonant
to [
] and reduce
the unstressed vowel to schwa [
] so
that it merges with Adam:
["æ m].
The underlying phonemes
emerge when the stress is shifted under affixation:
atóm-ic
[th"am- Ik].
Processes also figure in the neutralizations found in child language
such as the loss of the tongue-tip articulation of r so
that room merges with womb.
Phonological processes fall into
two broad categories: sound change and prosodic grouping. We briefly
illustrate each type. In in-articulate versus im-possible the
prefixal nasal assimilates the labial feature of the [p] thereby
changing from [n] to [m]. Dissimilation alters
neighboring sounds that share the same feature so that they become
more distinct from one another (see DISTINCTIVE FEATURES).
For example, the vocalic nucleus and offglide comprising the [au] diphthong
of how share a retracted tongue position in most English
dialects. In broad Australian English the nucleus is fronted to
the [æ] vowel of cat: h[æu].
Assimilation and dissimilation are subject to a strict locality
condition requiring that they apply in the context of the closest
sound with the appropriate feature. The phonological features that
define a sound are also subject to deletion and insertion: the
former typically operates in prosodically weak contexts (reduction
of unstressed vowels to schwa in át[
]m but [
]tóm-ic) and
the latter in strong contexts (aspiration of [t] before
stress in a[th]ómic).
Processes of prosodic grouping include
the organization of phonemes into syllables. In English a consonant
cluster such as [rt] easily combines with the
preceding vowel into a single syllable: monosyllabic mart.
But in order to syllabify the inverse cluster [tr],
a helping schwa is required: disyllabic me.t[
]r (cf. metr-ic).
Languages such as Japanese have simpler syllabic structures that
bar syllable-internal consonant clusters and place rigid restrictions
on syllable-final consonants. Accordingly, the [rt] cluster
in a loanword such as French courte [kurt] 'short' receives
two extra syllables when it is adapted into Japanese: kuruto (Shinohara
1997). At the next level of prosodic organization, syllables are grouped
into strong-weak (trochaic) or weak-strong (iambic) rhythmic units
known as metrical feet. Native Australian languages impose trochaic
rhythm so that words have a canonical SsSsSs . . . syllabic structure
in comparison to the iambic grouping sSsSsS . . . found in many
Native American languages. English has trochaic grouping as shown
by the strong-weak template imposed on nickname formation: Elízabeth shortens
to Ss Lísa; sS Elí is impossible.
Some linguists (e.g., Stampe 1979) distinguish between "processes" that reflect phonetically motivated limitations on which sounds can appear where in pronunciation and more arbitrary and conventional "rules" that are typically restricted to particular morphological or lexical contexts such as the voicing of [f] in the plural of leaf: leaves but reefs (*reeves) and verbal he leafs (*leaves) through the paper). A plausible but unsubstantiated hypothesis is that rules relate different lexical items stored in memory while processes operate online.
Phonological processes are often
phonetically motivated, seeming either to enhance the perceptibility
of a sound, especially in "strong" contexts or
more formal speaking styles (aspiration of prestressed [t] in a[t
h]ómic),
or to minimize articulatory gestures, especially in "weak" contexts
or fast tempos (flapping of the stop and reduction of the unstressed
vowel of átom ["æ
m]). Besides a typology based
on their formal properties, phonological processes are also usefully
viewed as different solutions to a common phonetic difficulty. For
example, the transition from the nasal to the fricative in the consonant cluster
of dense is relatively complex because it requires synchronization
of two independent gestures: raising the velum to shut off nasal
airflow and shifting the tongue tip from a closure to a constriction.
Common responses include insertion of a transitional stop den[t]se (to
rhyme with dents) or deletion of the tongue-tip closure d[
~]s. An
example from prosody is provided by the widespread tendency to avoid
syllables beginning with a vowel. When morphological or syntactic rules
juxtapose vowels, a variety of processes come into play to avoid
a syllable break between the vowels. These include deletion of one
of the vowels (Slavic), contraction of the vowels into a diphthong
(Polynesian) or long vowel (Sanskrit), or insertion of a consonantal
onset (British English intrusive [r] as in the
idea [r] is).
Processes are characteristically myopic
in the sense that in solving one phonetic problem they often create
another. Popular London (Cockney) deletion of initial [h] creates
a vowel cluster with the indefinite article ("a hedge"
[d
) -- a situation that is otherwise
avoided by substitution of the an allomorph (cf.
"an edge"
[n d];
Wells
1982). To take another example (data from Bethin 1992), in Polish [n] assimilates
the place of articulation of a following velar: ba[
]k 'bank'.
In the Southwestern dialect, the process is extended to clusters
arising from the deletion of a weak vowel, whereas in the Northeastern dialect,
such derived nk clusters remain unassimilated: ganek 'porch', ga[
]ka SW versus ga[n]ka NE
genitive singular. Many phonologists (e.g., Halle 1962) conclude
from examples like this that processes apply in a linear sequence:
in the Southwestern dialect, vowel deletion precedes nasal assimilation, whereas
in the Northeastern dialect, nasal assimilation precedes vowel deletion
(and so sees /ganek+a/ at its point of application).
An alternative interpretation (Donegan and Stampe 1979) sees all
processes as applying simultaneously to the input with each given
the option to iterate (Southwestern) or not (Northeastern).
Although myopic, phonological processes are typically not self-defeating in the sense of recreating the same problem they are called upon to solve. An example is provided by the liquid [l,r] dissimilation inherited from Latin (Steriade 1995), in which the suffixal [l] of nav-al, fat-al, mor-al is turned into [r] when the stem contains an [l]: stell-ar, lun-ar, column-ar, nucle-ar. The process systematically blocks when an [r] intervenes between the suffixal and stem [l]"s: flor-al, plur-al, later-al. If the point of the change is to avoid successive identical liquids, an output such as *flor-ar is no better than the input flor-al and hence the process is suspended.
Providing empirical substantiation to the notion "phonetic motivation" as well as determining the principles that underlie the interaction of rules and constraints remain outstanding research objectives.
Bethin, C. (1992). Polish Syllables. Columbus, OH: Slavica Publishers.
Donegan, P., and D. Stampe. (1979). The study of natural phonology. In D. Dinnsen, Ed., Current Approaches to Phonological Theory. Bloomington, IN: Indiana University Press, pp. 126-173.
Halle, M. (1962). Phonology in generative grammar. Word 18:54-72.
Makkai, V. B. (1972). Phonological Theory: Evolution and Current Practice. New York: Holt.
Sapir, E. (1925). Sound patterns in language. Language 1:37-51. Reprinted in Makkai 1972, pp. 13 - 21.
Sapir, E. (1933). The psychological reality of phonemes. Reprinted in Makkai 1972, pp. 22-31.
Shinohara, S. (1997). Analyse phonologique de l"adaptation japonaise de mots étrangers. Paris: Université de la Sorbonne Nouvelle, Paris 3.
Stampe, D. (1979). A Dissertation on Natural Phonology. New York: Garland.
Steriade, D. (1995). Underspecification and markedness. In J. Goldsmith, Ed., Handbook of Phonological Theory. Oxford: Blackwell, pp. 114-174.
Wells, J. (1982). Accents of English. Cambridge: Cambridge University Press.
Archangeli, D. and D. Pulleyblank. (1993). Grounded Phonology. Cambridge, MA: MIT Press.
Greenberg, J. (1978). Universals of Human Language. Stanford, CA: Stanford University Press.
Kenstowicz, M. (1994). Phonology in Generative Grammar. Oxford: Blackwell.
Copyright © 1999 Massachusetts Institute of Technology