Three puzzles in the phonetics and phonology of Thai Tone Elizabeth Zsiga, Georgetown University in

1
Three puzzles in the phonetics and phonology of
Thai ToneElizabeth Zsiga, Georgetown
Universityin collaboration with Bruce Morén,
Cornell University
2

• What role should abstract, formal
  representations play in describing and explaining
  sound patterns?

3

• Established result a system of interacting
  constraints is a useful analytical framework.
• Constraints should be grounded.
• V-stop-V Expend as little energy as possible
• Onset Cues to a Cs place of articulation are
  clearest when the C is released into a vowel.

4

• If constraints need to be grounded in facts
  outside the formal grammatical system, why not
  reference those facts directly? Why bother with
  a formal representation at all?

V-stop-V Be Lazy?
Onset
Maximize Perceptibility?
5

• In my view
• While phonological constraints may arise from
  physical dimensions, they do not reference those
  dimensions directly.
• Phonological constraints refer to abstract
  categories, which are not always directly
  realized in the speech stream.
• In this talk
• Evidence that reference to abstract categories
  (H, L, µ) are necessary to solve some interesting
  puzzles in the phonetics and phonology of Thai
  tone.

6
(No Transcript)
7
H L HL LH s s s s
s Mid High Low Falling Rising

• Exactly five lexical contrasts
• Simple, elegant, phonetically straightforward?

8

• Our data show that the tonal system of Thai is
  much more complex and interesting than is usually
  assumed.
• 3 Puzzles
• Unexpected tonal contours in citation form
• Paradoxical tone distribution in obstruent-final
  forms
• Surprising differences in tone realization
  between citation forms and connected speech.

9
(No Transcript)
10
H
M
L
F
R
11
H L HL LH s s s s
s Mid High Low Falling Rising
Straightforward phonetic implementation?
12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19

• Interim finding
• Tones are linked to moras, not syllables, in
  Thai.
• Provides an example of a match between
  phonological and phonetic representations.
• Wheres the evidence for abstraction?

20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
Claim ObsCodaL is the phonologization of a
phonetic tendency, not the phonetic tendency
itself.
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
The phonologically simple H tone is phonetically
a contour, which is fully realized, even on short
CVO syllables. It is phonological complexity
thats ruled out on these syllables, not phonetic.
30
sentence
token
31
(No Transcript)
32
(No Transcript)
33
(No Transcript)
34
(No Transcript)
35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
Again, the complexity ruled out by higher ranking
of TTs is phonological, not phonetic.
41
(No Transcript)
42
Conclusions Puzzle 1 unexpected tonal contours
can be explained if we assume that tones are
associated to moras. Puzzle 2 restrictions on
tone patterns in obstruent-final syllables can be
explained if we assume a constraint that requires
glottalized codas to be associated with low
tones. Puzzle 3 in connected speech, the tones
associated to the second mora may be deleted.
43
(No Transcript)
44

• Why bother with a formal representation?
• The tonal puzzles of Thai can only be solved by
  utilizing abstract, formal entities (tonal
  autosegments and moras).
• Generalizations are based in phonetic principles.
• They are discovered through careful phonetic
  analysis.
• Theyre just not always phonetically true.

45

• For further research
• The larynx articulatory and aerodynamic
  modeling
• Perception studies
• Relation of casual speech variation to
  non-variable lexical forms.