0

The distribution of rhotics

Joan Mascaró

The class of rhotics is characterized by a rich variation in both phonetic properties , and in the set of other sounds with which they alternate . The central claim of this paper is that a considerable part of the allophonic variation can be derived from conditions imposed by (un)markedness of syllable structure.

A good ground to test this claim is languages in which rhotics contrast, as in some Romance languages (e.g. Catalan, Occitan, Portuguese, Spanish). There we find a characteristic quatripartite division of contexts which holds across languages and dialects:

Position\Variety / Port(Eu), Sp(Eu), Cat(C) / Cat(V), Occ,
Sp(CA) / Port. (Caipira) / P (Rio) / P(NE)
a) / between vocoids / / / / / / / / / /
b) / s-initial positions ≠ a) /  /  /  /  / 
c) / second position in onset /  /  /  /  / 
d) / In the coda /  /  /  /  / Ø

We will show that this distribution can be derived from independently motivated sonority preferences in different syllabic positions (Murray and Vennemann (1983), Vennemann (1988), Clements (1990)). In a syllable X1...V...Xk, where V is the nucleus, 1) the sonority increase between X1 and V tends to be maximal, while the decrease between V and Ck tends to be minimal; 2) the sonority distances between adjacent elements tend to be uniform, 3) in Xk]s[ sX1 the sonority decrease between Xk and X1 tends to be maximal. This explains the preference for less sonorous  in syllable initial position (b), and the preference for more sonorous  in second onset position, where the sonority distances between the rhotic and the initial C and the rhotic and the nucleus V are uniform in [sCV... than in [sCV..., or [sCV... In coda position syllabic sonority preferences dictate the more natural , following 1); this is what happens in many of the varieties attested, but others present the less sonorous allophone. This discrepancy can be attributed to an independent process of coda tensing which in some cases can be motivated independently. The presence of two factors (the general preference for more sonorous coda and the particular coda tensing) explains the great language and dialect variation in this position.

The next (important) question to be asked is why contrast is confined to intervocoid position, where the rhotic is at the beginning of the onset. Compare this position (contrast) and syllable initial after a true consonant (only less sonorous rhotic). In these positions there is the combination of two factors: the distance in sonority to the nucleus and syllable contact. Thus in [ a less sonorous rhotic is preferred in order to maximize the sonority decrease after [n], but in [] the preceding vocoid allows for both tense [] and the less sonorous [].

An account in terms of prosodic markedness constraints related to sonority preferences, and faithfulness constraints will be developped, and the analysis in Bradley (2001) of 7 languages of different families which also contrast rhotics (Basque, Kurdish, Austronesian (3) Nilo-Saharan, Afro-Asiatic) will be discussed.