Place of Articulation in Consonants and Vowels: a Unified Theory

This paper is concerned with the following question: what phonological features are
required to characterize place of articulation in consonants and vowels? A significant result
of recent work in feature theory has been the discovery that features are organized into subclasses
which participate in phonological processes such as assimilation and deletion as a
single functional unit. One way of accounting for such behavior is to group features into
trees in such a way that all and only "natural classes" of features are constituents (see
Clements 1985, 1987, 1991; Sagey 1986; McCarthy 1988, 1991, and others for general
discussion). Among many other contributions, we owe to Sagey the notion of articulator
node, designating the participation of an active articulator (the lips, the tongue front, the
tongue body) in the production of a given segment, and to McCarthy the new feature
pharyngeal, designating constrictions fonned in the pharynx (broadly defined to include the
larynx).

This study offers a contribution to two areas. First, departing from most current feature
systems, it proposes that a single set of features characterizes place of articulation in
both consonants and vowels. This set includes the oral cavity features labial, coronal,
dorsal and possibly a pharyngeal cavity feature radical or constricted pharynx, located
under the pharyngeal node. Under this proposal, features such as back and round become
superfluous, and can be eliminated from feature theory. It will be shown that this simplification
of the set of place features allows us to capture generalizations about the relations
between consonants and vowels that earlier feature systems have failed to account for.

Second, this study offers a somewhat different model of feature grouping from that
found in much other work. In particular, it proposes that place features of vocoids (i.e.,
vowels and glides) are partially segregated from those of consonants, in the sense that they
are assigned to different regions or planes in phonological representation. Some amount of
segregation of this sort is required to express the fact that place features of vowels and
glides (which we informally term "V-place" features) spread more freely than place features
of consonants (hereafter termed "C-place" features). For instance, it is well known that Vplace
features are not blocked by the presence of intervening consonants in processes of
vowel harmony and assimilation.