Linguists have long been interested in the study of phonotactic patterns, or restrictions on the concatenation of sounds. It has long been recognized that palatalized consonants do not show the same distribution as their non-palatalized counterparts. For instance, a survey of both Slavic (Bulgarian, Polish, Russian, etc.) and non-Slavic languages (Irish, Lithuanian, Nenets, etc.) with contrastive palatalization shows that in these languages palatalized consonants are found only in a subset of positions where “plain” segments occur (Kochetov 1999). The most common environment for a palatalized consonant to occur is in a syllable onset (initial and medial prevocalic) and the least favourable one is a preconsonantal coda. It appears that if a language maintains the contrast in a coda environment, it is necessarily has it in the onset position. Also, if a language allows a plain-palatalized contrast before a consonant, it does so in the final coda position. Among the three places of articulation, palatalized coronals are more likely to be found in coda position than labials and velars (place markedness). The place of articulation of the following consonant is also of importance.
This paper investigates some typologically common phonotactic restrictions on phonemic palatalized consonants and proposes a phonetic (articulatory and perceptual) explanation for these phenomena. The proposal is tested experimentally on data from Russian. The results suggest that the peculiarities of distribution of palatalized consonants can be explained by phonetic, articulatory and perceptual, factors interacting with an Optimality Theory-type phonological grammar.
To test this view we investigated articulatory and perceptual properties of Russian plain and palatalized stops in a series of experiments. In an articulatory magnetometer study a Russian speaker read nonsense phrases of the type taC (C)apy, where C stands for any of the consonants /p/, /t/, /p′/, and /t′/. This gave us all possible types of clusters (16) with respect to palatalization and homo-/hetero-organicity (e.g. pt, pt′, p′t, p′ t′, pp′, p′ p′, etc.). For each cluster articulatory gestures, or movements of the lips, tongue tip, and tongue body, were recorded and measured. The results show that the main indicator of secondary palatal articulation is tongue body raising and fronting (cf. Bolla 1981, Bondarko 1981). No complete assimilation to the following consonant (C′C → CC or CC′ → C′C′ was found: a palatalized C′1 (_C) was significantly different from the respective plain C1 (_C) in terms of tongue body height. However, this difference was less substantial than before vowels (_V). A preconsonantal C′ was found to retain its tongue body height to a greater extent before a hetero-organic C2. This was particularly true for /t′/. Even less articulatory difference was observed for plain and palatalized C1 before a palatalized C2. Homorganicity and place of articulation were also important factors here.
In a perceptual study the nonsense stimuli obtained in the previous experiment were presented to twelve Russian listeners. The subjects were asked to identify the consonants of interest in phrases taC (C)apy by pushing the keys corresponding to the phonemes /p/, /t/, /p′/, and /t′/. As was expected, least identification errors were found in environments where articulatory differences between plain and palatalized consonants were most robust (e.g. prevocalic onset) and most mistakes were made in the least salient contexts (e.g. preconsonantal coda, esp. where C2 is homorganic). The contrast between /p/ and /p′/ was less reliably identified than the distinction between /t/ and /t(′)/. More errors were made before a palatalized C2 rather than before a plain one. Thus, assimilation with respect to palatalization (C′C → CC and CC′ → C′C′ was found on the perceptual level, but it was based on articulatory differences. The assimilation pattern corresponds to the observed phonotactic restrictions on the plain-palatalized contrast.
The results suggest that specific restrictions on plain-palatalized contrast found in the phonological grammar of Russian and other languages are motivated by articulatory and perceptual factors. It is argued, however, that the synchronic grammars do not encode the knowledge of phonetic substance (contra the view taken by Optimality Theory (cf. positional markedness; Prince and Smolensky 1993 or licensing by cue, Steriade 1997). Rather the observed phonotactic patterns result from the interaction of the phonetic factors with the phonological grammar during language acquisition.
Bolla, K. 1981. A Conspectus of Russian Speech Sounds. Cologne: Boelau.
Bondarko, L.V. 1981. Fonetičeskoe opisanie jazyka i fonologičeskoe opisanie reči. Moscow: Nauka.
Kochetov, Alexei. 1999. “Phonotactic Constraints on the Distribution of Palatalized Consonants.” Toronto Working Papers in Linguistics. Holland Academic Graphics.
Prince, Alan and Paul Smolensky 1993. Optimality Theory: Constraint Interaction in Generative Grammar. Ms., Rutgers University.
Steriade, Donca. 1997. Phonetics in Phonology: The Case of Laryngeal Neutralization. Ms., University of California, Los Angeles.