Anglophone perceptions of Arabic syllable structure

1
Anglophone perceptions of Arabic syllable
structure

• Azra N. Ali, Michael Ingleby and David Peebles
• University of Huddersfield
• England

2
Introduction

• Existing studies have shown that the syllable has
  an internal hierarchical structure
• syll (onset) rhyme
• rhyme nucleus (coda)
• e.g. flame f l (aI m)

3
Detecting syllables in the mental models

• The syllable structure has been tested
  empirically using word games with concealed
  parts.

(Mehler et al, 1981)
(Ferrand Segue, 1996 Segui Ferrand, 2002)
4
Alternative Method

• We outline here an alternative method for probing
  syllable structure.
• We have sought for an empirical probe that
  engages only one mode of perception
• that concerned with audiovisual stimuli
• probed by response of groups of participants to
  audiovisually incongruent data (McGurk effect).

5
McGurk Effect

• When humans are presented with conflicting (or
  incongruent) audio and visual stimuli, the speech
  percept elicited may differ from that in either
  channel ..because of phonological fusion.
• McGurk and MacDonald asked their recording
  technician to create a videotape with the audio
  syllable /ba/ dubbed onto a visual /ga/.
• most normal adults reported fusion percept /da/
  or /Da/
• symbolically, AUD(baga)VIS ? (da)PER(Da)PER

6
McGurk fusion rates

• Fusion rate measured as the proportion of
  participants that perceive a fusion.

7
Probing syllable structure using the McGurk effect

• McGurk fusion rates have been used successfully
  to probe syllable structure of English words
  (Ali, 2003 Ali Ingleby 2005)
• Findings showed significant coda / onset
  differences of fusion rate
• Codas fuse more readily than onsets
• AUD(mapmack)VIS ? (mat)PER 75
• AUD(balegale)VIS ? (dale)PER 45

8
Coda-Onset Difference

• Such differences survive in branching
  constituents, polysyllabic words and words
  embedded in natural phrase contexts.

(Ali 2003 Ali Ingleby 2004)
9
CV Languages

• In a language without codas, however, the
  quantitative distinction between fusion rates in
  codas and onsets obviously cannot persist.
• Although no-coda languages are known, Arabic
  syllabic structure is contested. In such a
  contested case.
• one could use incongruent stimuli to open up the
  contest to experimental test,
• seeking to falsify the no-coda hypothesis by
  finding the segments that show too much fusion to
  be onsets.

10
Arabic Language (1)

• The Arabic tradition of Sybawaih, on which the
  (phonetic) alphabet for Arabic is founded, uses
  CV units symbolised orthographically by a
  consonant with a vowel diacritic ?? /fa/
  ?? /fi/ ?? /fu/
• In the Western tradition, however, classical
  scholars, treating Arabic like Latin and Greek,
  have postulated that there are closed syllables
  CVC, CVVC, CVCC.
• Recently, Baothmans study (2002) of Arabic
  speech patterns amasses evidence that Arabic has
  a CV phonology. This suggests that the Western
  tradition should be questioned.

11
Arabic Syllable Structure (2)

• Thus, an preliminary aim of this paper is to
  question empirically whether native Arabic
  speakers have codas in their mental
  representation of Arabic speech.
• We used word stimuli with an incongruent phonetic
  segment that would be a coda in the Western
  tradition but an onset in the Sybawaih view, and
  other stimuli incongruent at undisputed onset
  sites
• Null H0 that fusion rates are the same at onset
  and coda sites.
• Alternative H1 that fusion rates are not the
  same for onsets and codas.

12
Results of Preliminary Aim

• Table 1 fusion rates

13
Arabic Syllable Structure (3)

• Primary Aim
• Having confirmed that fusion rates are the same
  at onset and coda sites for Arabophones. We
  subjected Anglophones to the same Arabic word
  stimuli that had confirmed no-coda perceptions
  amongst Arabophones
• Null hypothesis H0 Anglophones fuse Arabic word
  stimuli at the same rate with incongruity at
  onset and coda sites
• Alternative H1 Anglophones fuse Arabic word
  stimuli at significantly different rates at these
  sites

14
Results of Primary Aim

• Table 2 fusion rates

15
Results (2)
onsets
codas
16
A more recent study Geminates in Arabic

• Arabophones show bimodally distributed fusion
  rates between singleton and geminate consonants
• Significant differences in fusion rates between
  singletons (fusing readily), and geminates
  (resistant to fusion).
• But when the same Arabic stimuli are put to
  Anglophones, their fusion rates are unimodally
  distributed
• fusion rates for singletons and geminates were
  similar, no significant differences.

17
Summary (1)

• These new results suggest that Arabophones use
  mental models of speech that differ markedly from
  the mental models that mediate speech perception
  in Anglophones.
• Further research questions
• Subject Arabophones to English stimuli

18
Summary (2)

• Further research questions
• For an Anglophone learning Arabic, at what point
  on the learning curve, does a student acquire the
  no-coda mental model that characterises
  Arabophones?
• What are the fusion habits of those bilingually
  fluent from an early age ?.
• Probing syllable structure of Moroccan Arabic.