OCPDriven variation in American English schwa production

1
OCP-Driven variation in American English schwa
production

• Mary Ann Walter
• MIT

2
Introduction

• Languages often avoid sequences with identical
  elements in close proximity.
• This generalization has been expressed in the
  linguistics literature as the Obligatory Contour
  Principle, or OCP.
• Its effects may be seen through constraints on
  underlying representations, triggering
  alternations, and blocking alternations.

3
Introduction

• Two OCP effects
• Epenthesis (English inflectional schwa)
• pat pats pass passes
• pæt pæts, pæt?s pæs pæs?z, pæs
• Antigemination
• digibté digbé she/I married
• xararté xararé s/he burned
• V-deletion except between 2 identical consonants
• Claimed not to have a phonetic counterpart
• (McCarthy 1986, Blevins 2005)

4
Introduction

• Recent work has shown OCP effects that are
  gradient as well as categorical
• Properties of the lexicon (Frisch 2004 Frisch,
  Pierrehumbert and Broe 2004 Berkley 2000)
• Perception of consonantal place (Coetzee 2005)
• Goodness ratings (Berent and Shimron 1997
  Berent, Everett and Shimron 2001 Coetzee 2005)
• Even phonotactically licit OCP violations have
  processing effects.

5

• Goal
• Determine extent to which phonetic productions in
  OCP-violating contexts differ from what is
  otherwise expected.
• Prediction
• Vowels will have longer durations when between
  identical consonants than non-identical ones.

6
Exp 1 Method

• Subjects (n9) read aloud the following stimuli
  from visual presentation in random order in a
  soundproof booth, in isolation and in a frame
  sentence.
• b a b a b o t
• ee d d ee
• oo g g oo
• Bababot. Do you know what a bababot is?

7
Exp 1 Results

• Schwa duration is significantly longer when
  produced between two identical consonants than
  non-identical ones (RM ANOVA, plt.001).
• This asymmetry holds for each individual subject.

8
Exp 2 Method

• Subjects (n8) read aloud the following stimuli
  from visual presentation in random order in a
  soundproof booth, in isolation and in a frame
  sentence.
• p a p ee n
• k k oo l
• t t
• She papeens a lot. Shes a papeener now.

9
Exp 2 Results

• Again, schwa duration is significantly longer
  when produced between two identical consonants
  than non-identical ones (RM ANOVA, plt.001).
• Again, the asymmetry is consistent across
  subjects.

10
Exp 2 Results

• Schwa deletion is quite common between the two
  voiceless stops of Exp. 2.
• Such deletion is significantly positively
  correlated with being between two identical
  consonants
• (Pearsons correlation coefficient.07,
  two-tailed significance p.058) .

11
Discussion

• An OCP effect on schwa duration is robust and
  replicable (even in the more prominent
  word-initial syllable).
• Schwas are longer between identical consonants
  than otherwise.
• Similarly, schwas are less likely to be deleted
  between identical consonants than otherwise.

12
Conclusions

• Even in the absence of categorical or
  grammaticalized repairs to OCP violations,
  speakers manipulate low-level acoustic variables
  in a gradient fashion in order to ameliorate
  them.
• Those documented here are phonetic counterparts
  to antigemination.
• This undermines claims that it is not a true OCP
  effect, and offers a potential diachronic origin
  for the phenomenon.