A Constraint Based Analysis of an Opacity Effect

Ben Hermans
Tilburg University

In this talk I propose an account of a Bulgarian opacity effect in the framework of Sympathy Theory and I demonstrate that alternative approaches are not explanatory.

Bulgarian uses Schwa Insertion to prevent a liquid from appearing in a peak. To see this, compare the underlying forms in column I in (1), next page, with their surface realizations in the singular (column II) and the plural (column III).

Schwa's position varies. It is inserted as follows: A) a tautosyllabic cluster is avoided and, if that is not possible, then B) a coda cluster is avoided. A) is exemplified in column III in (1). B) is exemplified in column II in (1). Due to A) we get e.g. v(schwa)rxove, rather than vr(schwa)xove(; due to B) we get vr(schwa)x, not v(schwa)rx.

A front yer changes this picture; if the syllable into which schwa is inserted precedes a front yer, then the position of schwa goes against A). In this environment, then, schwa's position is opaque. Examples are given in (2). In (2) the adjectivizing suffix -(e)n is added. This suffix contains a yer, as is evident from the fact that e alternates with zero (compare the columns IV and V in (2)). Notice now that the position of schwa in column IV in (2) is opaque; it goes against A). Only the front yer triggers this opaque behavior of epenthetic schwa in the syllable preceding it. A so called 'back'yer does not. This is shown in the definite forms in column IV of (1a), where -(schwa)t is added. In the morpheme -(schwa)t schwa is not epenthetic, but rather a 'back' yer. That it is a yer is evident from the fact that it alternates with zero (compare the def. sing. in the masc. forms in (1a) with the def. sing. in the fem. forms in (1b)).

Rule based theories can order rules to account for opacity. In this case we have to order Schwa Insertion before Yer Vocalization. This entails that a yer is still unsyllabified at the stage when schwa is inserted. Consequently, a consonant preceding a yer is syllabified in the coda. Hence, schwa is inserted in the onset, due to B). Unfortunately, however, this analysis will not work, because only a front yer can trigger opacity, and Yer Vocalization cannot distinguish between front and back yers.

Another device is cyclicity; we can try to apply Schwa Insertion cyclically. This approach, however, is also unacceptable, and for the same reason; it cannot explain why the rule is cyclic just before a front yer suffix. We conclude, then, that two ways to solve opacity in rule based theories are unacceptable.

The phenomenon is also problematic for two OT approaches: Paradigm Uniformity (Uniform Exponence (UE)) and OO-Faithfulness. They have the same drawback as the previous approaches, because they cannot explain why the opacity effect is only triggered by suffixes containing a front yer.

Only one possibility is left, then: Sympathy Theory, proposed in McCarthy 1997. This is our proposal. First we postulate, following Ito and Mester (1993), that a place node is a segment head. Secondly, we postulate that schwa is a placeless segment (also suggested in Ito(1986). Thirdly, a yer is underlyingly moraless (proposed in Yearly 1995). The idea now is that the constraint selecting the sympathetic candidate (the 'FLOWER'-candidate) is HEAD-IDENT(MORA).

Hence, the consonant preceding it is syllabified in the coda in the FLOWER-candidate. We now get the opacity effect by ranking faithfulness to the FLOWER-candidate higher than the constraint against complex onsets. Now opaque vr(schwa)xen is optimal, because it is more faithful to the FLOWER-candidate, vr(schwa)xn, than transparent v(schwa)rxen, and because faithfulness to the FLOWER-candidate dominates avoidance of a complex onset.

This account is the only one that can explain, rather than stipulate, why opacity occurs only before a front yer, and it does so with independently motivated constraints. It thus provides compelling evidence for Sympathy Theory as opposed to rule ordering, cyclicity, UE and OO-Faithfulness.

DATA (from Scatton 1975 and 1984)


Alderete, John. 1995. Faithfulness to prosodic heads. Manuscript, University of Massachusetts, Amherst.
Ito, Junko. 1986. Syllable Theory in prosodic phonology. Ph.D. Dissertation, University of Massachusetts, Amherst.
Ito, Junko and Armin Mester. 1993. Licensed segments and safe paths. In Constraints, violations, and repairs in phonology, ed. C. Paradis, D. LaCharitÈ and E. Nikiema. Special issue of the Canadian Journal of Linguistics
McCarthy, John. 1997. Sympathy and phonological opacity. Manuscript, University of Massachusetts, Amherst.
Scatton, E.A. 1975. Bulgarian phonology. Cambridge, Massachusetts, Slavica Publishers.
Scatton, E.A. 1984. A Reference grammar of modern Bulgarian. Cambridge, Massachusetts, Slavica Publishers.
Yearley, J. 1995. Jer vowels in Russian. In University of Massachusetts occasional papers in linguistics 18: Papers in Optimality Theory, 77-136. GLSA, University of Massachusetts, Amherst.

< Back to List of Abstracts | Back to PLC23 Home Page >

About the PLC23 Committee
Previously held Penn Linguistics Colloquium: PLC22 (1998), PLC21 (1997)

Penn Department of Linguistics
University of Pennsylvania