出版社：Department of Linguistics, University of Toronto
摘要：A fundamental problem for phonologists concerns how the child arrives at the representations of a given phonological system. An investigation of inventories shows that the segmental make-up of a given inventory is not random: rather, smaller inventories contain statistically more common segments while larger inventories contain the statistically more common segments as well as those that are less frequent. As Maddieson 1984 shows, there are a number of segments which are almost universally present in inventories. In addition, many implicational statements can be made concerning the presence or absence of different segments based on what is present in a given inventory. The combination of universals and implicational statements suggests that there is a 'markedness' hierarchy informing acquisition and a path of acquisition based on contrasts present in an inventory. In work on consonant inventories, Lindblom 1988 recognizes three types of consonants, basic consonants, elaborated consonants, and complex consonants. Lindblom argues that: The number of segments a language uses in the basic or complex categories is predictable from the total size of its inventory. Small vowel or consonant inventories recruit only basic segments. (quoted in de Boysson-Bardies and Vihman 1991:316) The basic consonants include three places of articulation of voiceless stops (p, t, k), three places of articulation of voiceless fricatives (f, s, š), three places of articulation of nasals (m, n, ?), and two liquids (1, r). Lindblom argues that these are the most frequent consonants and any elaborations on these (additions of laryngeal features, subdivisions of places of articulation, etc.) are added only after all basic articulations are present. While Lindblom makes the observation that elaborated segments imply basic segments, he has no explanation for this nor can he offer an account for the range of asymmetries found within the basic consonants: for instance, why, if the number of places of articulation available to stops and nasals is not equal, do the stops always show a greater range of places of articulation? Why, within the sonorants, do nasals show a greater range of places of articulation than liquids? He also offers no explanation for why certain types of basic segments are elaborated before other basic segment are. For example, why, if the number of laryngeal contrasts available to stops and fricatives differ, do stops exhibit a greater number of laryngeal contrasts? In this paper we suggest that a theory of phonology should account for the types of observations noted above. We offer a formal account of segment structure which has as a consequence an account of the constraints on inventory shapes and segmental frequency as well as of the sequence of acquisition through an investigation of constraints on inventory shapes and segmental frequency. We propose that the observations about inventory shapes and segmental frequency .are a direct consequence of a theory of hierarchical segmental representation informed by a theory of markedness and coupled with a metric that evaluates segmental complexity.
