Academia.eduAcademia.edu

Figure 22 – uploaded by Jeremy Perkins

See full PDFdownloadDownload figure
Fig. 20. Experiment | & 2 Results — Control Comparisons The significance of the [+constricted glottis]-low tone preference is confirmed in the

Figure 20 Experiment | & 2 Results — Control Comparisons The significance of the [+constricted glottis]-low tone preference is confirmed in the

Related Figures (88)

They both have /ate H-tone targets. On the other hand, the falling-tone sequence involves
They both have /ate H-tone targets. On the other hand, the falling-tone sequence involves
or : : : 3 Consonant-Tone Restrictions in Thai — Ruangjaroon, 2006 * [2] is just another voiceless unaspirated obstruent, and is treated as such here and elsewhere. * Here and throughout, “T” refers to any voiceless unaspirated stop or affricate (notably all obstruent coda: are voiceless unaspirated stops due to neutralization in codas). “D” refers to any voiced stop. “h” refers to itself. “C..” refers to all other consonants other than these. ““N” refers to the class of sonorant consonants. Table 2
or : : : 3 Consonant-Tone Restrictions in Thai — Ruangjaroon, 2006 * [2] is just another voiceless unaspirated obstruent, and is treated as such here and elsewhere. * Here and throughout, “T” refers to any voiceless unaspirated stop or affricate (notably all obstruent coda: are voiceless unaspirated stops due to neutralization in codas). “D” refers to any voiced stop. “h” refers to itself. “C..” refers to all other consonants other than these. ““N” refers to the class of sonorant consonants. Table 2
Fig. 1 — Glottal Stop Onset
Fig. 1 — Glottal Stop Onset
Lexical Type Frequency for Thai Monosyllabic Words in the Dictionary
Lexical Type Frequency for Thai Monosyllabic Words in the Dictionary
Lexical Type Frequency for Thai Monosyllabic Words in the ORCHID Corpus
Lexical Type Frequency for Thai Monosyllabic Words in the ORCHID Corpus
Lexical Type Frequency for Monosyllabic English Loan Words Table 6 Lexical Type Frequency for Thai Monosyllabic Onomatopoeia
Lexical Type Frequency for Monosyllabic English Loan Words Table 6 Lexical Type Frequency for Thai Monosyllabic Onomatopoeia
The proportions of words that violate high-tone restrictions are larger in loans and
The proportions of words that violate high-tone restrictions are larger in loans and
Lexical Type Frequency for Monosyllabic Native Thai Words There are two notable observations that are apparent in Table 7. First, [h] is under-
Lexical Type Frequency for Monosyllabic Native Thai Words There are two notable observations that are apparent in Table 7. First, [h] is under-
Summary of Consonant-Tone Restrictions in Thai by Lexical Stratum An important observation concerns a difference between the status of lexical gaps Table 8
Summary of Consonant-Tone Restrictions in Thai by Lexical Stratum An important observation concerns a difference between the status of lexical gaps Table 8
Lexical Type Frequency for Non-Roots in Thai Lexical Frequencies for Roots and Non-Roots in Thai
Lexical Type Frequency for Non-Roots in Thai Lexical Frequencies for Roots and Non-Roots in Thai
Table 9 illustrates that there is very little difference between roots and non-roots in
Table 9 illustrates that there is very little difference between roots and non-roots in
Fig. 2. Spectral Tilt Schematized (H1 is the 1“ harmonic of FO, H2 the 2" harmonic of FO, A1 is the 1“ formant).
Fig. 2. Spectral Tilt Schematized (H1 is the 1“ harmonic of FO, H2 the 2" harmonic of FO, A1 is the 1“ formant).
Fig. 3. Spectral Tilt following creaky [p] (left) and following modal [m] (right).
Fig. 3. Spectral Tilt following creaky [p] (left) and following modal [m] (right).
Fig. 4. Two examples from the slideshow. Stimuli are separated by spaces. The stimulus “th’(above), oa pronounced [p'a:], means “to cut”. The distractor “ou” (below), pronounced [lwa:j], means “to creep”.
Fig. 4. Two examples from the slideshow. Stimuli are separated by spaces. The stimulus “th’(above), oa pronounced [p'a:], means “to cut”. The distractor “ou” (below), pronounced [lwa:j], means “to creep”.
Phonetic hypotheses for each onset type” To enable statistical testing, paired comparisons were constructed by identifying near
Phonetic hypotheses for each onset type” To enable statistical testing, paired comparisons were constructed by identifying near
Pairwise comparisons for onsets Not all of the stimuli formed in the manner described above corresponded to a Thai
Pairwise comparisons for onsets Not all of the stimuli formed in the manner described above corresponded to a Thai
Table 12 — Experimental word stimuli A single Thai frame sentence, written in Thai orthography, was used with the
Table 12 — Experimental word stimuli A single Thai frame sentence, written in Thai orthography, was used with the
Fig. 5. Example of vowel segmentation.
Fig. 5. Example of vowel segmentation.
These results indicate that the nonce words were not very effective at eliciting the
These results indicate that the nonce words were not very effective at eliciting the
Section 3.3.1 presents evidence that voiced and unaspirated stops and the unaspirated
Section 3.3.1 presents evidence that voiced and unaspirated stops and the unaspirated
Fig. 6. Mean spectral tilt measurements for each speaker categorized by onset.
Fig. 6. Mean spectral tilt measurements for each speaker categorized by onset.
In every comparison, spectral tilt measurements were significantly affected by onset.
In every comparison, spectral tilt measurements were significantly affected by onset.
In three of the four comparisons, onset type affected FO. Significant effects were data containing only the relevant pairs of onsets for each comparison. comparison. The baseline comparison between [p"] and [m] yielded an effect, just as it significantly affects FO. Table 18 below summarizes the results for ANOVA’s run on
In three of the four comparisons, onset type affected FO. Significant effects were data containing only the relevant pairs of onsets for each comparison. comparison. The baseline comparison between [p"] and [m] yielded an effect, just as it significantly affects FO. Table 18 below summarizes the results for ANOVA’s run on
T-test results for FO comparisons among oral stops While not statistically significant, mean FO for both [b] and [p] is lower than [p*] for
T-test results for FO comparisons among oral stops While not statistically significant, mean FO for both [b] and [p] is lower than [p*] for
The high spectral tilt values for [?] indicate that it is not inducing creakiness, unlike
The high spectral tilt values for [?] indicate that it is not inducing creakiness, unlike
Table 20 summarizes t-test results. While FO differences reach statistical significance
Table 20 summarizes t-test results. While FO differences reach statistical significance
T-test results for FO comparisons between [p] and [?] Turning now to the comparison between the two glottals, [?] and [h], Figure 10
T-test results for FO comparisons between [p] and [?] Turning now to the comparison between the two glottals, [?] and [h], Figure 10
Fig. 10. Mean spectral tilt — [?] vs. [h] across speakers Table 21 below shows that higher mean spectral tilt is seen following [h] than [?]
Fig. 10. Mean spectral tilt — [?] vs. [h] across speakers Table 21 below shows that higher mean spectral tilt is seen following [h] than [?]
T-test results for spectral tilt comparisons between [?] and [h] The FO results show a divergence from the spectral tilt results, however. [?] is
T-test results for spectral tilt comparisons between [?] and [h] The FO results show a divergence from the spectral tilt results, however. [?] is
Only a single significant difference was found in Table 22 however: Speaker T has
Only a single significant difference was found in Table 22 however: Speaker T has
T-test results for FO comparisons between [?] and [h]
T-test results for FO comparisons between [?] and [h]
onset in Figure 12 below. Results for t-tests are summarized in Table 23.
onset in Figure 12 below. Results for t-tests are summarized in Table 23.
Fig. 13. Glottal Stop Onsets ‘ ‘ ‘ : 26 indicative of creakiness.
Fig. 13. Glottal Stop Onsets ‘ ‘ ‘ : 26 indicative of creakiness.
Stimulus Design for Unaspirated-high ban In addition to the unaspirated onset-high tone combination, three separate parallel
Stimulus Design for Unaspirated-high ban In addition to the unaspirated onset-high tone combination, three separate parallel
Lexical Gap Status in Unchecked Monosyllables The predictions among each comparison category for each experiment are shown in
Lexical Gap Status in Unchecked Monosyllables The predictions among each comparison category for each experiment are shown in
Experimental Predictions by Comparison Type”
Experimental Predictions by Comparison Type”
Fig. 14. Experiment 1 Results — Test Comparisons
Fig. 14. Experiment 1 Results — Test Comparisons
Neither high tone nor rising tone was significant in comparisons varying manner, and Table 27
Neither high tone nor rising tone was significant in comparisons varying manner, and Table 27
Fig. 15. Response Means — Control Comparisons
Fig. 15. Response Means — Control Comparisons
Fig. 16. Responses categorized by LND Difference
Fig. 16. Responses categorized by LND Difference
Experiment | - Summary of Results & Predictions
Experiment | - Summary of Results & Predictions
Fig. 17. Experiment 1 Results — Test Comparisons
Fig. 17. Experiment 1 Results — Test Comparisons
Fig. 18. Loan Distribution in the Thai Lexicon classified by Onset-Tone Combination
Fig. 18. Loan Distribution in the Thai Lexicon classified by Onset-Tone Combination
Fig. 19. Experiment 2 Results — Test Comparisons dispreferred in comparisons varying tone, but not in comparisons varying manner.
Fig. 19. Experiment 2 Results — Test Comparisons dispreferred in comparisons varying tone, but not in comparisons varying manner.
In comparisons varying manner, a significant negative effect was discovered for both
In comparisons varying manner, a significant negative effect was discovered for both
Fig. 21. Effects of LND in Both Experiments
Fig. 21. Effects of LND in Both Experiments
Fig. 22. Experiment | & 2 Results — Test Comparisons
Fig. 22. Experiment | & 2 Results — Test Comparisons
In comparisons varying manner, the fact that no significant interactions existed below for comparisons varying manner.
In comparisons varying manner, the fact that no significant interactions existed below for comparisons varying manner.
Logistic Regression — Between Experiments (Comparisons Varying Tone) Table 31 First, a significant interaction between experiment and voiced manner was
Logistic Regression — Between Experiments (Comparisons Varying Tone) Table 31 First, a significant interaction between experiment and voiced manner was
Summary of Results & Predictions Across Experiments
Summary of Results & Predictions Across Experiments
Perception of Tones Of particular importance here is the tendency for high and rising tone to be
Perception of Tones Of particular importance here is the tendency for high and rising tone to be
Consonant-Tone Restrictions in Thai This chapter has presented a pair of experiments that showed all four onset-tone in Thai is given below in Table 34, based on the findings of this chapter.
Consonant-Tone Restrictions in Thai This chapter has presented a pair of experiments that showed all four onset-tone in Thai is given below in Table 34, based on the findings of this chapter.
7) Locality in Consonant-Tone Interaction in Thai Unchecked Syllables
7) Locality in Consonant-Tone Interaction in Thai Unchecked Syllables
Coda-Tone Restrictions in Checked Syllables Recall that only low tone is allowed in all checked syllables. In addition, high tone
Coda-Tone Restrictions in Checked Syllables Recall that only low tone is allowed in all checked syllables. In addition, high tone
than one mora (*[yp]T).
than one mora (*[yp]T).
It is not possible to preserve both the H and L tones in the output. The constraint *[TT]u
It is not possible to preserve both the H and L tones in the output. The constraint *[TT]u
32) Only low tone surfaces in checked syllables with [+CG] onsets 5.2.6 Rising and High Tone Restrictions
32) Only low tone surfaces in checked syllables with [+CG] onsets 5.2.6 Rising and High Tone Restrictions
Assuming Richness of the Base (Prince & Smolensky 1993/2004), it is important to
Assuming Richness of the Base (Prince & Smolensky 1993/2004), it is important to
35) = *[+CG]-[H]p2, MAX[H] or MAX[L] >> LINEARITY (TONE), *H, *L Candidate (35a) is the faithful candidate, but the sequence of the [+CG] consonant
35) = *[+CG]-[H]p2, MAX[H] or MAX[L] >> LINEARITY (TONE), *H, *L Candidate (35a) is the faithful candidate, but the sequence of the [+CG] consonant
Without a [+CG] onset, ALIGN-R would select right-aligned H preferentially (MH)
Without a [+CG] onset, ALIGN-R would select right-aligned H preferentially (MH)
In addition, Lic-T-Rt takes care of the basic fact that tones are right-aligned
In addition, Lic-T-Rt takes care of the basic fact that tones are right-aligned
violation of * [+CG]-[H]p2. violates LINEARITY (TONE)zoan allowing candidate (58b) to be optimal, despite the
violation of * [+CG]-[H]p2. violates LINEARITY (TONE)zoan allowing candidate (58b) to be optimal, despite the
The faithful candidate (59b) fatally violates *[+voice]-LH. The winning candidate (59a) satisfies *[+voice]-LH by switching the linear order of the L and H tones, surfacing with LINEARITY (TONE)roan. This is shown in tableau (59).
The faithful candidate (59b) fatally violates *[+voice]-LH. The winning candidate (59a) satisfies *[+voice]-LH by switching the linear order of the L and H tones, surfacing with LINEARITY (TONE)roan. This is shown in tableau (59).
The faithful rising-tone candidate (63b) fatally violates *[LH-Obs]o since the rising tone
The faithful rising-tone candidate (63b) fatally violates *[LH-Obs]o since the rising tone
However, in the loan stratum, high tone surfaces faithfully. The only way to account
However, in the loan stratum, high tone surfaces faithfully. The only way to account
Both the H and L tones must be preserved since MAX[H]roan and MAX[L]roan
Both the H and L tones must be preserved since MAX[H]roan and MAX[L]roan
Fig. 23. Experiment 2 Results Recall that in comparisons varying tone (UH-UL & VH-VL), a more significant trend experiments in Chapter 4, shown again in Figure 23 below.
Fig. 23. Experiment 2 Results Recall that in comparisons varying tone (UH-UL & VH-VL), a more significant trend experiments in Chapter 4, shown again in Figure 23 below.
Violation profiles by comparison To illustrate how the violation profiles will map to a predicted response score, an
Violation profiles by comparison To illustrate how the violation profiles will map to a predicted response score, an
Fig. 24. Predicted response scores by comparison
Fig. 24. Predicted response scores by comparison
Fig. 25. Transformed predicted response scores by comparison with scaling factor (h=1/3) included the actual mean responses in Figure 25 below. comparisons varying tone. The results of the transformed predicted mean are plotted with
Fig. 25. Transformed predicted response scores by comparison with scaling factor (h=1/3) included the actual mean responses in Figure 25 below. comparisons varying tone. The results of the transformed predicted mean are plotted with
Fig. 26. Transformed predicted response scores by comparison (without *L)
Fig. 26. Transformed predicted response scores by comparison (without *L)
# Translates Thai tone marks to JPA given a tone-determining onset consonant tdons_ # mid tone = 0; low tone = 1; falling tone = 2; high tone = 3; rising tone = 4
# Translates Thai tone marks to JPA given a tone-determining onset consonant tdons_ # mid tone = 0; low tone = 1; falling tone = 2; high tone = 3; rising tone = 4
Related topics:
PhonologyPsycholinguisticsOptimality TheoryThai Language
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved