Roberts/HT 2014 Week 3

Phonology

Exercise — Opacity in Catalan

(Based on an exercise by Ricardo Bermúdez-Otero)

  1. Consider the data in (1), and the set of constraints in (2):

    (1)
    /ɡat-/ ‘cat’
    a.gat [ɡat] ‘he-cat’
    b.gata [ɡatə] ‘she-cat’
    c.gat feliç [ɡat.fə.lis] ‘happy cat’
    d.gat gelós [ɡad.ʒə.los] ‘jealous cat’
    /ɡɾɔɡ-/ ‘yellow’
    a.groc [ɡɾɔk] ‘yellow.m
    b.groga [ɡɾɔ.ɣə] ‘yellow.f
    c.grogor [ɡɾu.ɣo] ‘yellowish’
    d.grogenc [ɡɾu.ɣɛŋk] ‘rather yellow’
    (2)
    • Agree-[voice]

      Assign a violation mark for every pair of adjacent segments {a, b} such that a and b differ in specification with respect to the [voice] feature.

    • VOP (Voiced Obstruent Prohibition)

      Assign a violation mark for every voiced obstruent in the output candidate.

    • Ident-[voice]

      Assess a violation for every pair of segments a, a′ such that a is an input segment, a′ is its correspondent in the output, and a and a′ differ in specification with respect to the [voice] feature.

    • OnsIdent-[voice]

      Assess a violation for every pair of segments a, a′ such that a′ is in the onset of a syllable and the pair {a, a′} violate Ident-[voice]

    (See Lombardi 1999 for background).

    Rank the constraints in (2) to predict the pattern of voicing illustrated in (1), and illustrate your ranking with tableaux.

    (Don’t worry about the fact that [ɣ] is an allophone of /ɡ/: it’s still voiced, at least, so just treat it as though it were [ɡ])

  2. /ʒ/ in Catalan surfaces as [dʒ] or [tʃ] under conditions illustrated in the following data:

    (2)
    /bɔʒ-/ ‘mad’
    a.boja [bɔ.ʒə] ‘madwoman’
    b.bogeria [bu.ʒə.ɾi.ə] ‘madness’
    c.boig [bɔtʃ] ‘madman’
    d.boig petit [bɔtʃ.pə.tit] ‘small madman’
    e.boig gran [bɔdʒ.ɣɾan] ‘big madman’
    f.boig estrany [bɔ.dʒəs.tɾaɲ] ‘strange madman’

    (Note that /ʃ/ does not surface as [tʃ]).

    1. What is the domain of /ʒ/-affrication?
    2. What is the domain of voicing assimilation?
    3. How, therefore, do you expect the two generalisations to interact?
  3. Finally, consider the following data. Do they confirm or deny the prediction you’ve just made about the interaction between /ʒ/-affrication and voicing assimilation?

    (3)
    /məteʃ-/ ‘same’
    a.mateix [mə.teʃ] ‘same-masc
    b.mateixa [mə.te.ʃə] ‘same-fem.
    c.mateix corb [mə.teʃ.kɔɾp] ‘same raven’
    d.mateix gos [mə.teʒ.ɣɔs] ‘same dog’

    Note in particular that we do not have *[mə.teʃ.ɣɔs] or *[mə.tedʒ.ɣɔs] for ‘same dog’.

  4. References

    Lombardi, Linda (1999) “Positional faithfulness and voicing assimilation in Optimality Theory.Natural Language and Linguistic Theory 17:267‒302.