# What does it mean to be "checked": case study of Xiapu Min ### Yuan Chai ### University of California Los Angeles <yuanchai@g.ucla.edu> <https://yuanchaiyc.github.io/website/presentation/NYCU-talk-YC.html> --- ### Thanks to - Ye Shihong 叶施宏 as my research collaborator for the experiments on Xiapu Min; - Wang Jinmin 王锦敏 as my primary language consultant for the documentation of Xiapu Min; - Grace Kim, Eliana Marvizon, Kaylina Moreno, Lesly Nunez, Samuel Parker, Marilyn Sun, Daniela Vasquez, Chunxiao Zhang, and Zhien Zhang for being my RA on the field work recording transcrption and the language survey. --- ### Fundamental goal in phonetics and phonology: - To come up with a set of sounds that are **necessary** & **sufficient** to describe all the possible sounds in language. - "Checked" is a phenomenon that raises a debate in this goal: - "Checked" constituent encompasses multiple dimensions of phonetic and phonological features/categories. - e.g. Duration, Voice quality, Tone, Syllable type - Is it necessary to propose a "checked" feature/category? --- ### What do we mean when we say a constituent is "checked"? Meixian Hakka 梅县客家话 tones <table> <tbody style = "font-size: 23px;"> <tr> <td>(1)</td> <td style="color: red;">Checked</td> <td style="color: red;">/it 55/</td> <td style="color: red;">"wing"</td> <td style="color: red;">Checked</td> <td style="color: red;">/it 31/</td> <td style="color: red;">"benefit"</td> </tr> <tr> <td></td> <td>Unchecked</td> <td>/i 31/</td> <td>"rain"</td> <td>Unchecked</td> <td>/i 33/</td> <td>"clothes"</td> </tr> <tr> <td></td> <td>Unchecked</td> <td>/i 53/</td> <td>"idea"</td> <td>Unchecked</td> <td>/i 11/</td> <td>"aunt"</td> </tr> <tr> <td> </td> <td style="font-size: 20px;" colspan=6> (Lee & Zee, 2009) </td> </tr> </tbody> </table> <img width = "700px" src="image/hakka_wing_benefit_rain.svg"/> <audio controls><source src="audio/4_wing.wav" type="audio/wav"></audio><audio controls><source src="audio/5_benefit.wav" type="audio/wav"></audio><audio controls><source src="audio/6_rain.wav" type="audio/wav"></audio> - <span style="font-size: 28px;">"Checked" refers to syllables closed by /-p, -t, -k/ which also bear different tones from open and nasal-closed syllables.</span> --- ### What do we mean when we say a constituent is "checked"? Phonetic properties of "checked" constituents in the literature: <table> <tbody style = "font-size: 23px;"> <tr> <td>Meixian Hakka</td> <td>"Checked" syllable</td> <td>V{p,t,k}</td> <td>Short, Modal or Final-glottalized</td> </tr> <tr> <td></td> <td>"Checked" tone</td> <td>55, 31</td> <td>High or Mid-falling pitch</td> </tr> </tbody> </table> - Checked constituents have an obstruent-like constriction at the end of vowels, creating a percept that the voicing is shut off abruptly. - Phonetic dimensions associated with checked constituents: - Pitch - Duration - Voice quality - Do we need to add a "checked" feature or category to the existing phonological labels? - Or can "checked" constituents be reduced to an existing feature and/or category? --- #### What do we already have in IPA and what can "checked" constituents be reduced to? Can "checked" tone be reduced to tone? <img width = "400px" src="image/ipa_tone.png"/> --- #### What do we already have in IPA and what can "checked" constituents be reduced to? Can "checked" constituents be reduced to duration? <img width = "400px" src="image/ipa_short.png"/> --- #### What do we already have in IPA and what can "checked" constituents be reduced to? Can "checked" phonation (Vˀ) be reduced to creaky/glottalized phonation? <img width = "700px" src="image/ipa_creaky.png"/> --- ### Roadmap - <span style="color: #585858">1: What do we mean when we say a constituent is "checked"? </span> - 2: Phonetics of "checked" constituent - Production and perception of Xiapu Min checked tones in citation and sandhi forms - 3: Phonology of "checked" constituent - Survey of Chinese language families --- ### Phonetics of checked constituents Fieldwork on Xiapu Min: Checked syllables and tones --- ### Significance of the phonetic analysis of checked constituents - Describe the multi-dimensionality of the acoustics of checked constituents - Pitch - Duration - Voice quality - Amplitude of voicing - Degree of glottal constriction - Degree of noisiness - Provide the first comprehensive quantitative analysis of checked constituents - Time-series data; - Dimentionality reduction analysis; - Resynthesized stimuli. --- ### Phonetics of "checked" constituent #### Xiapu Min 霞浦闽语 <img src="image/xm_map.svg" style = "height: 500px;" /> --- ### Phonetics of "checked" constituent #### Xiapu Min - Existing documentation: Wen (2015); - Phonological inventory; Tone sandhi process - Conducted fieldwork in Xiapu County in Summer 2019 and Summer 2022; - Collected 38 hours of recordings; - Conducted production study on 10 speakers; - Conducted perception study on 30 listeners; --- ### Phonetics of "checked" constituent #### Xiapu Min <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T5</td> <td style="vertical-align: middle; text-align:left">/θiʔ 5/*</td> <td style="vertical-align: middle; text-align:left">湿</td> <td style="vertical-align: middle; text-align:left">"wet"</td> <td rowspan="7" style="vertical-align: middle; text-align:right; width: 800px;"><img src="image/xm_f0_check.svg" style="height:380px;text-align:right"></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T2</td> <td style="vertical-align: middle; text-align:left">/θiʔ 2/*</td> <td style="vertical-align: middle; text-align:left">实</td> <td style="vertical-align: middle; text-align:left">"concrete"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T44</td> <td style="vertical-align: middle; text-align:left">/θi 44/</td> <td style="vertical-align: middle; text-align:left">诗</td> <td style="vertical-align: middle; text-align:left">"poem"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T11</td> <td style="vertical-align: middle; text-align:left">/θi 11/</td> <td style="vertical-align: middle; text-align:left">时</td> <td style="vertical-align: middle; text-align:left">"time"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T35</td> <td style="vertical-align: middle; text-align:left">/θi 35/</td> <td style="vertical-align: middle; text-align:left">四</td> <td style="vertical-align: middle; text-align:left">"four"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T23</td> <td style="vertical-align: middle; text-align:left">/θi 23/</td> <td style="vertical-align: middle; text-align:left">寺</td> <td style="vertical-align: middle; text-align:left">"temple"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T42</td> <td style="vertical-align: middle; text-align:left">/θi 42/</td> <td style="vertical-align: middle; text-align:left">死</td> <td style="vertical-align: middle; text-align:left">"die"</td> <td></td> </tr> </tbody> </table> <p style="font-size: 23px;">*Checked tones are by convention annotated with one Chao's numeral in Chinese literature to indicate its short duration</p> --- ### Phonetics of "checked" constituent #### Xiapu Min <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T5</td> <td style="vertical-align: middle; text-align:left">/θiʔ 5/</td> <td style="vertical-align: middle; text-align:left">湿</td> <td style="vertical-align: middle; text-align:left">"wet"</td> <td rowspan="7" style="vertical-align: middle; text-align:right; width: 800px;"><img src="image/xm_f0_check_level.svg" style="height:380px;text-align:right"></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T2</td> <td style="vertical-align: middle; text-align:left">/θiʔ 2/</td> <td style="vertical-align: middle; text-align:left">实</td> <td style="vertical-align: middle; text-align:left">"concrete"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T44</td> <td style="vertical-align: middle; text-align:left">/θi 44/</td> <td style="vertical-align: middle; text-align:left">诗</td> <td style="vertical-align: middle; text-align:left">"poem"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T11</td> <td style="vertical-align: middle; text-align:left">/θi 11/</td> <td style="vertical-align: middle; text-align:left">时</td> <td style="vertical-align: middle; text-align:left">"time"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T35</td> <td style="vertical-align: middle; text-align:left">/θi 35/</td> <td style="vertical-align: middle; text-align:left">四</td> <td style="vertical-align: middle; text-align:left">"four"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T23</td> <td style="vertical-align: middle; text-align:left">/θi 23/</td> <td style="vertical-align: middle; text-align:left">寺</td> <td style="vertical-align: middle; text-align:left">"temple"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T42</td> <td style="vertical-align: middle; text-align:left">/θi 42/</td> <td style="vertical-align: middle; text-align:left">死</td> <td style="vertical-align: middle; text-align:left">"die"</td> <td></td> </tr> </tbody> </table> --- ### Phonetics of "checked" constituent #### Xiapu Min <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T5</td> <td style="vertical-align: middle; text-align:left">/θiʔ 5/</td> <td style="vertical-align: middle; text-align:left">湿</td> <td style="vertical-align: middle; text-align:left">"wet"</td> <td rowspan="7" style="vertical-align: middle; text-align:right; width: 800px;"><img src="image/xm_f0_check_level_rising.svg" style="height:380px;text-align:right"></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T2</td> <td style="vertical-align: middle; text-align:left">/θiʔ 2/</td> <td style="vertical-align: middle; text-align:left">实</td> <td style="vertical-align: middle; text-align:left">"concrete"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T44</td> <td style="vertical-align: middle; text-align:left">/θi 44/</td> <td style="vertical-align: middle; text-align:left">诗</td> <td style="vertical-align: middle; text-align:left">"poem"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T11</td> <td style="vertical-align: middle; text-align:left">/θi 11/</td> <td style="vertical-align: middle; text-align:left">时</td> <td style="vertical-align: middle; text-align:left">"time"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T35</td> <td style="vertical-align: middle; text-align:left">/θi 35/</td> <td style="vertical-align: middle; text-align:left">四</td> <td style="vertical-align: middle; text-align:left">"four"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T23</td> <td style="vertical-align: middle; text-align:left">/θi 23/</td> <td style="vertical-align: middle; text-align:left">寺</td> <td style="vertical-align: middle; text-align:left">"temple"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T42</td> <td style="vertical-align: middle; text-align:left">/θi 42/</td> <td style="vertical-align: middle; text-align:left">死</td> <td style="vertical-align: middle; text-align:left">"die"</td> <td></td> </tr> </tbody> </table> --- ### Phonetics of "checked" constituent #### Xiapu Min <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T5</td> <td style="vertical-align: middle; text-align:left">/θiʔ 5/</td> <td style="vertical-align: middle; text-align:left">湿</td> <td style="vertical-align: middle; text-align:left">"wet"</td> <td rowspan="7" style="vertical-align: middle; text-align:right; width: 800px;"><img src="image/xm_f0_full.svg" style="height:380px;text-align:right"></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Checked</td> <td style="vertical-align: middle; text-align:left">T2</td> <td style="vertical-align: middle; text-align:left">/θiʔ 2/</td> <td style="vertical-align: middle; text-align:left">实</td> <td style="vertical-align: middle; text-align:left">"concrete"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T44</td> <td style="vertical-align: middle; text-align:left">/θi 44/</td> <td style="vertical-align: middle; text-align:left">诗</td> <td style="vertical-align: middle; text-align:left">"poem"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T11</td> <td style="vertical-align: middle; text-align:left">/θi 11/</td> <td style="vertical-align: middle; text-align:left">时</td> <td style="vertical-align: middle; text-align:left">"time"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T35</td> <td style="vertical-align: middle; text-align:left">/θi 35/</td> <td style="vertical-align: middle; text-align:left">四</td> <td style="vertical-align: middle; text-align:left">"four"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T23</td> <td style="vertical-align: middle; text-align:left">/θi 23/</td> <td style="vertical-align: middle; text-align:left">寺</td> <td style="vertical-align: middle; text-align:left">"temple"</td> <td></td> </tr> <tr> <td style="vertical-align: middle; text-align:left">Unchecked</td> <td style="vertical-align: middle; text-align:left">T42</td> <td style="vertical-align: middle; text-align:left">/θi 42/</td> <td style="vertical-align: middle; text-align:left">死</td> <td style="vertical-align: middle; text-align:left">"die"</td> <td></td> </tr> </tbody> </table> <audio controls><source src="audio/22_T5.wav" type="audio/wav"></audio><audio controls><source src="audio/23_T2.wav" type="audio/wav"></audio><audio controls><source src="audio/24_T44.wav" type="audio/wav"></audio><audio controls><source src="audio/25_T11.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/26_T35.wav" type="audio/wav"></audio><audio controls><source src="audio/27_T23.wav" type="audio/wav"></audio><audio controls><source src="audio/28_T42.wav" type="audio/wav"></audio> --- #### Xiapu Min What acoustic parameters distinguish checked tones from unchecked tones most effectively? --- Acoustic measures: F0: Higher the F0, higher the pitch <img width = 550px src="image/f0_100_200.png" /><br> <audio controls><source src="audio/F0_100hz.wav" type="audio/wav"></audio><audio controls><source src="audio/F0_200hz.wav" type="audio/wav"></audio> --- Acoustic measures: Residual H1* **(Chai & Garellek, 2022)**: The lower the energy of H1, the more glottal constriction <img src="image/h1_illustration.png" /> <audio controls><source src="audio/spectral_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/spectral_creaky.wav" type="audio/wav"></audio> --- Acoustic measures: Harmonic-to-noise ratio (HNR): The lower the HNR, the noisier <img width = 650px src="image/hnr_illustration_modal.png"><img width = 650px src="image/hnr_illustration_noisy.png"> <audio controls><source src="audio/no_noise_02082023.wav" type="audio/wav"></audio><audio controls><source src="audio/all_noise_02082023.wav" type="audio/wav"></audio> <p style="font-size:23px">(Generated using UCLA Voice Synthesizer; Kreiman et al., 2006)</p> --- Acoustic measures: Amplitude of voicing - Measured by Strength of Excitation (SoE): The amplitude of voicing is usually low during glottalization. <img width = 650px src="image/soe_wave_illustration.svg"> <audio controls><source src="audio/a_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/a_creaky.wav" type="audio/wav"></audio> --- Acoustic measures: <p class="subpoint">F0</p> <p class="subpoint">Duration</p> <p class="subpoint">Amplitude of fundamental frequency (Residual H1*);</p> <p class="subpoint">Harmonic-to-noise ratio (HNR);</p> <p class="subpoint">Amplitude of voicing (SoE)</p> Calculated using VoiceSauce (Shue et al., 2011) **Question**: What acoustic parameters distinguish checked tones from unchecked tones most effectively? - Stimuli: 90 monosyllabic words produced by 10 speakers - Carrier phrase: /wa42 e11 kaŋ42 TARGET tɕja42 ka44 tɕi35/<br>我会讲XXX这个字<br>("I know how to say the segment TARGET") --- Results of Linear Discriminant Analysis (LDA) <table style="width: 120%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 24px"> <tr> <td style="vertical-align: middle;text-align:right; width:300px">23.39% </td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/xm_lda_2112344.svg"></td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/xm_lda_5354244.svg"></td> <td style="vertical-align: middle;text-align:left; width:300px">16.31% </td> </tr> <tr> <td></td> <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">72.11%<br> <span style="color: #fc8d62">T2</span> vs. <span style="color: #a6cee3">T44</span>, <span style="color: #1f78b4">T23</span>, <span style="color: #5e3c99">T11</span><br> Amplitude of voicing (final)<br>Duration<br>Residual H1* (final) <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">81.5%<br> <span style="color: #b2df8a">T5</span> vs. <span style="color: #a6cee3">T44</span>, <span style="color: #66c2a5">T42</span>, <span style="color: #8da0cb">T35</span><br> F0 (initial, medial)<br>Duration </td> <td></td> </tr> </tbody> </table> --- Results of Linear Discriminant Analysis (LDA) T2 vs. T44, 23, 11: Amplitude of voicing (final), Duration, Residual H1* (final) <img width = 650px src="image/xm_soez_t2112344_annotate.png" /><img width = 650px src="image/place_holder.png"/> --- Results of Linear Discriminant Analysis (LDA) T2 vs. T44, 23, 11: Amplitude of voicing (final), Duration, Residual H1* (final) <img width = 650px src="image/xm_soez_t2112344_annotate.png" /><img width = 650px src="image/xm_h1cr_t2112344_annotate.png"/> --- Results of Linear Discriminant Analysis (LDA) T5 vs. T44, 42, 35: F0 (initial, medial), Duration <img width = 650px src="image/xm_f0_logz_5354244_annotate.png" /><img width = 550px src="image/place_holder_xmduration.png" /> --- Results of Linear Discriminant Analysis (LDA) T5 vs. T44, 42, 35: F0 (initial, medial), Duration <img width = 650px src="image/xm_f0_logz_5354244_annotate.png" /><img width = 550px src="image/xm_duration_annotate.png" /> --- T2 is distinguished from unchecked tones by glottalization and short duration. - T5 is distinguished from unchecked tones by high F0 and short duration. --- #### Xiapu Min checked tone perception What acoustic parameters do listeners of Xiapu Min use when identifying checked tones? - Parameters manipulation - F0 shape and height - Glottalization (Spectral tilt, Noise, and Amplitude of voicing) - Duration --- <img width = 600px src="image/xm_stimuli_duration.svg"/><img width = 600px src="image/xm_stimuli_duration_f0_placeholder.svg"/> --- <img width = 600px src="image/xm_stimuli_duration.svg"/><img width = 600px src="image/xm_stimuli_duration_f0.svg"/> - <img width = 600px src="image/xm_stimuli.svg"/> --- Example stimuli for condition of High-falling F0: <img src="image/percep_exp1_spectro_5longshort.svg" width = "700"> <img src="image/percep_exp1_spectro_5gllongshort.svg" width = "700"> <audio controls><source src="audio/41_xm_5_long_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/42_xm_5_short_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/43_xm_5_long_gl.wav" type="audio/wav"></audio><audio controls><source src="audio/44_xm_5_short_gl.wav" type="audio/wav"></audio> --- Word identification task: <img src="image/xm_percept_exp_page_crop.svg" width = "1000"> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_1.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_3.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_4.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_5.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_6.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_7.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2_8.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br><span style="color: white;">Random forest importance of variables:<br>F0 > Duration > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_1.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_3.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_4.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_5.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_6.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_7.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_8.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_9.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_10.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_11.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5_12.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px"><span style="color: white;">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">High F0, short duration, glottalization → T5.<br><span style="color: white;">Random forest importance of variables:<br>Duration, F0 > Glottalization</span> </td> </tr> </tbody> </table> --- <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T2 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block"><img width = "850px" src="image/xm_response_T2.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">Low F0, short duration, glottalization → T2.<br>Random forest importance of variables:<br>F0 > Duration > Glottalization </td> </tr> </tbody> </table> <table style="width: 120%;"> <tbody style = "font-size: 25px;margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tr> <td colspan="2"> Percentage of T5 responses in different F0 (facet), duration (color), glottalization (color) conditions </td> </tr> <tr> <td style="vertical-align: middle;text-align: center;display: inline-block;"><img width = "850px" src="image/xm_response_T5.svg" ></td> <td style="vertical-align: middle;text-align:left; width: 750px">High F0, short duration, glottalization → T5.<br>Random forest importance of variables:<br>Duration, F0 > Glottalization </td> </tr> </tbody> </table> --- #### Production correlates ≠ perceptual cues <table style="width: 100%;"> <tbody> <tr> <td></td> <td>T2</td> <td>T5</td> </tr> <tr> <td>Production</td> <td>Glottalization, Duration</td> <td>F0, Duration</td> </tr> <tr> <td>Perception</td> <td>F0 > Duration > Glottalization</td> <td>F0, Duration > Glottalization</td> </tr> </tbody> </table> --- #### Xiapu Min Neutralization between checked and unchecked tones in sandhi forms <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left;">Rule</td> <td style="vertical-align: middle; text-align:left;">Citation</td> <td colspan="2" style="vertical-align: middle; text-align:left;">Gloss</td> <td style="vertical-align: middle; text-align:left;">Sandhi</td> <td colspan="2" style="vertical-align: middle; text-align:left;">Gloss</td> </tr> <tr> <td rowspan="3" style="vertical-align: middle; text-align:left; width: 1000px;">{T2, T23, T44} → T44 / ___ X</td> <td style="vertical-align: middle; text-align:left; width: 200px;">/xuʔ 2/</td> <td style="vertical-align: middle; text-align:left; width: 100px;">服</td> <td style="vertical-align: middle; text-align:left; width: 200px;">"clothes"</td> <td style="vertical-align: middle; text-align:left; width: 300px;">[xu 44 tsɔŋ 44]</td> <td style="vertical-align: middle; text-align:left; width: 100px;">服装</td> <td style="vertical-align: middle; text-align:left; width: 200px;">"clothes"</td> </tr> <tr> <td style="vertical-align: middle; text-align:left;">/xu 23/</td> <td style="vertical-align: middle; text-align:left;">护</td> <td style="vertical-align: middle; text-align:left;">"protect"</td> <td style="vertical-align: middle; text-align:left;">[xu44 kain44]</td> <td style="vertical-align: middle; text-align:left;">护工</td> <td style="vertical-align: middle; text-align:left;">"caretaker"</td> </tr> <tr> <td style="vertical-align: middle; text-align:left;">/xu 44/</td> <td style="vertical-align: middle; text-align:left;">肤</td> <td style="vertical-align: middle; text-align:left;">"skin"</td> <td style="vertical-align: middle; text-align:left;">[xu44 tɕin42]</td> <td style="vertical-align: middle; text-align:left;">肤浅</td> <td style="vertical-align: middle; text-align:left;">"superficial"</td> </tr> </tbody> </table> <audio controls><source src="audio/29_cite_cloth.wav" type="audio/wav"></audio><audio controls><source src="audio/30_cite_protect.wav" type="audio/wav"></audio><audio controls><source src="audio/31_cite_skin.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/32_sandhi_clothes.wav" type="audio/wav"></audio><audio controls><source src="audio/33_sandhi_caretaker.wav" type="audio/wav"></audio><audio controls><source src="audio/34_sandhi_superficial.wav" type="audio/wav"></audio> - <table style="width: 100%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle; text-align:left;">Rule</td> <td style="vertical-align: middle; text-align:left;">Citation</td> <td colspan="2" style="vertical-align: middle; text-align:left;">Gloss</td> <td style="vertical-align: middle; text-align:left;">Sandhi</td> <td colspan="2" style="vertical-align: middle; text-align:left;">Gloss</td> </tr> <tr> <td rowspan="3" style="vertical-align: middle; text-align:left; width: 1000px;">{T5, T35, T42} → T55 / ___ X</td> <td style="vertical-align: middle; text-align:left; width: 200px;">/iʔ 5/</td> <td style="vertical-align: middle; text-align:left; width: 100px;">一</td> <td style="vertical-align: middle; text-align:left; width: 200px;">"one"</td> <td style="vertical-align: middle; text-align:left; width: 300px;">[i 55 tiaŋ 23]</td> <td style="vertical-align: middle; text-align:left; width: 100px;">一定</td> <td style="vertical-align: middle; text-align:left; width: 200px;">"certain"</td> </tr> <tr> <td style="vertical-align: middle; text-align:left;">/i 35/</td> <td style="vertical-align: middle; text-align:left;">意</td> <td style="vertical-align: middle; text-align:left;">"mind"</td> <td style="vertical-align: middle; text-align:left;">[i 55 ŋuai 23]</td> <td style="vertical-align: middle; text-align:left;">意外</td> <td style="vertical-align: middle; text-align:left;">"accident"</td> </tr> <tr> <td style="vertical-align: middle; text-align:left;">/i 42/</td> <td style="vertical-align: middle; text-align:left;">以</td> <td style="vertical-align: middle; text-align:left;">"by"</td> <td style="vertical-align: middle; text-align:left;">[i 55 xeu 23]</td> <td style="vertical-align: middle; text-align:left;">以后</td> <td style="vertical-align: middle; text-align:left;">"later"</td> </tr> </tbody> </table> <audio controls><source src="audio/35_cite_one.wav" type="audio/wav"></audio><audio controls><source src="audio/36_cite_mind.wav" type="audio/wav"></audio><audio controls><source src="audio/37_cite_by.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/38_sandhi_certain.wav" type="audio/wav"></audio><audio controls><source src="audio/39_sandhi_accident.wav" type="audio/wav"></audio><audio controls><source src="audio/40_sandhi_later.wav" type="audio/wav"></audio> --- #### Xiapu Min tones in sandhi forms For tones that are phonologically neutralized in sandhi forms, are they still distinguishable at phonetic level? - Acoustic measures: - F0 - Duration - Amplitude of voicing (Strength of Excitation SoE): **Smaller** SoE, **more** <span style="color:red">glottalization</span> - Amplitude of fundamental frequency (Residual H1*): **Smaller** H1*, **more** <span style="color:red">glottalization</span> - Harmonic-to-noise ratio (HNR): **Smaller** HNR, **more** <span style="color:red">glottalization</span> - Stimuli: 21 minimal pairs, 41 compounds, produced by 10 speakers. - Carrier phrase: /wa42 e11 kaŋ42 TARGET tɕja42 la44 θø11/<br>我会讲XXX这个词<br>(“I know how to say the word TARGET”) --- Results of Linear Discriminant Analysis (LDA) T44 vs. T23 vs. T2 in sandhi forms <table style="width: 120%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 24px"> <tr> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_44_sandhi_244.svg"></td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_44_sandhi_223.svg"></td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_44_sandhi_2344.svg"></td> </tr> <tr> <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 79.53% (p < .001)<br> duration<br>HNR (initial) <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 65.35% (p = .003)<br> duration </td> <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 67.57% (p = .068)<br> duration </td> </tr> </tbody> </table> --- T44 vs. T23 vs. T2 in sandhi forms <img width = 580px src="image/xm_sandhi_44_dur.png" /><img width = 680px src="image/place_holder.png"/> --- T44 vs. T23 vs. T2 in sandhi forms <img width = 580px src="image/xm_sandhi_44_dur.png" /><img width = 680px src="image/xm_sandhi_44_hnr_annotate.png"/> --- Results of Linear Discriminant Analysis (LDA) T42 vs. T35 vs. T5 in sandhi forms <table style="width: 120%; margin-bottom: 0px; margin-top: 0px; margin:0;display:block;"> <tbody style = "font-size: 24px"> <tr> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_55_sandhi_542.svg"></td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_55_sandhi_535.svg"></td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "430px" src="image/lda_55_sandhi_3542.svg"></td> </tr> <tr> <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 83.02% (p < .001)<br> duration<br>HNR (initial) <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 78.16% (p < .001)<br> duration<br>HNR (initial) </td> <td style="vertical-align: top; text-align:center;display: inline-block; width: 430px">Accuracy: 59.56% (p = .035)<br> </td> </tr> </tbody> </table> --- T42 vs. T35 vs. T5 in sandhi forms <img width = 580px src="image/xm_sandhi_55_dur.png" /><img width = 680px src="image/place_holder.png"/> --- T42 vs. T35 vs. T5 in sandhi forms <img width = 580px src="image/xm_sandhi_55_dur.png" /><img width = 680px src="image/xm_sandhi_55_hnr_annotate.png"/> --- #### Xiapu Min tones in sandhi forms For tones that are phonologically neutralized in sandhi forms, are they still distinguishable at phonetic level? - The neutralization is partial for several contrasts: - T2 vs. T44 - T5 vs. T42 - T5 vs. T35 - The most effective parameters for distinguishing these neutralized tones are <span style="color:red;">duration</span> and <span style="color:red;">HNR</span>. Checked tones tend to have a <span style="color:red;">shorter</span> duration and a <span style="color:red;">noisier</span> beginning in sandhi forms. --- #### Xiapu Min checked tone in sandhi forms perception Are listeners of Xiapu Min able to distinguish checked tones from unchecked ones in their neutralized sandhi forms? - Use natural stimuli with 40 tokens (5 contrasts * 2 words in contrast * 2 speakers * 2 tokens). - 5 contrasts: T2-T44, T23-T44, T5-T35, T5-T42, T35-T42 --- #### Sample stimuli <img style = "margin-top: -20px; margin-bottom: -20px;" width = 630px src="image/xm_perception_sandhi_stimuli_p1.svg" /> <audio controls><source src="audio/xm_perception_sandhi_cite_19A.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_cite_20A.wav" type="audio/wav"></audio> --- #### Sample stimuli <img style = "margin-top: -20px; margin-bottom: -20px;" width = 630px src="image/xm_perception_sandhi_stimuli_p2.svg" /> <audio controls><source src="audio/xm_perception_sandhi_35A.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_36A.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/xm_perception_sandhi_35A.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_36A.wav" type="audio/wav"></audio> --- #### Sample stimuli <img style = "margin-top: -20px; margin-bottom: -20px;" width = 630px src="image/xm_perception_sandhi_stimuli.svg" /> <audio controls><source src="audio/xm_perception_sandhi_cite_19A.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_cite_20A.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/xm_perception_sandhi_35A.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_36A.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/xm_perception_sandhi_35A_noise.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_sandhi_36A_noise.wav" type="audio/wav"></audio> --- Word identification task: <img src="image/xm_percept_expII_page.svg" width = "1000"> --- Results <img src="image/exp2_response_annotate_sig.png" style = "margin-top: -30px;" width = "800px"> --- Results <img src="image/exp2_response_annotate.png" style = "margin-top: -30px;" width = "800px"> --- Results - What acoustic cue do listeners use to distinguish T5 from T35 in sandhi forms? - glmer(Response ~ F0 + Duration + H1*-H2* + HNR + (1|Participant), family = binomial(link = "logit")) - Shorter duration leads to a higher likelihood of T5 response. Longer duration leads to a higher likelihood of T35 response. --- #### Summary of the production and perception of checked tones in Xiapu Min Production correlates ≠ perceptual cues <table style="width: 100%;"> <tbody> <tr> <td></td> <td></td> <td>Production</td> <td>Perception</td> </tr> <tr> <td>Citation</td> <td>T2</td> <td>Glottalization, Duration</td> <td>F0 > Duration > Glottalization</td> </tr> <tr> <td></td> <td>T5</td> <td>F0, Duration</td> <td>F0, Duration > Glottalization</td> </tr> </tbody> </table> --- #### Summary of the production and perception of checked tones in Xiapu Min Production correlates ≠ perceptual cues <table style="width: 100%;"> <tbody> <tr> <td></td> <td></td> <td>Production</td> <td>Perception</td> </tr> <tr> <td>Citation</td> <td>T2</td> <td>Glottalization, Duration</td> <td>F0 > Duration > Glottalization</td> </tr> <tr> <td></td> <td>T5</td> <td>F0, Duration</td> <td>F0, Duration > Glottalization</td> </tr> <tr> <td>Sandhi</td> <td>T2-T44</td> <td>Duration, Glottalization</td> <td>NA</td> </tr> <tr> <td></td> <td>T5-T35</td> <td>Duration, Glottalization</td> <td>Duration</td> </tr> <tr> <td></td> <td>T5-T42</td> <td>Duration, Glottalization</td> <td>NA</td> </tr> </tbody> </table> --- ### Phonetics of "checked" constituent #### Is "checked" constituent a necessary phonological feature/category? - Phonetically, the checked constituents in Xiapu Min are distinguished from their unchecked counterparts: - It is reasonable to propose "checked" and "unchecked" labels to distinguish two phonetically distinct constituents. - Should the phonetic difference between checked and unchecked constituents be counted as a distinct phonological feature/category? --- ### Survey of checked constituents Chinese - 94 varieties <table> <tbody style = "font-size: 20px;"> <tr> <td>Gan 赣</td> <td>Anyi 安义, Chongyang 崇阳, Chaling 茶陵, Duchang 都昌, Jianning 建宁, Jishui Luotian 吉水螺田, Liling Baitutan 醴陵白兔潭, Liling Lujiang 醴陵渌江, Liuyang Jiaoxi 浏阳焦溪, Nancheng 南城, Xinyu Shatu 新余沙土, Shaowu 邵武, Susong Heta 宿松河塔, Pingjiang Chengguan 平江城关, Pingjiang Meixian 平江梅仙, Pingjiang Nanjiang 平江南江, Wuning 武宁, Xiushui 修水, Xupu Longtan 溆浦龙潭, Yangxin Guohe 阳新国和, Yiyang 戈阳, Yongxin 永新, Yueyang Xinqiang 岳阳新墙, Yugan 余干 </td> </tr> <tr> <td>Hakka 客家</td> <td>Changting 长汀, Dayu 大余, Ganxian Panlong 赣县蟠龙, Heyuan 河源, Hong Kong Xigong Hakka 香港西贡客家话, Jiexi 揭西, Meixian 梅县, Ningdu 宁都, Ninghua 宁化, Wengyuan 翁源, Liannan 连南, Luchuan 陆川, Luoyuan She 罗源畲族, Qingxi 清溪, Sandu 三都, Xihe 西河, Xin'an 新安, Xiuzhuan 秀篆, Wuping Yanqian 武平岩前 </td> </tr> <tr> <td>Jin 晋</td> <td> Baochang 宝昌, Baotou 包头, Dongsheng 东胜, Fengzhen 丰镇, Huhetaote 呼和浩特, Shangdu 商都, South Taiyuan 太原南郊 </td> </tr> <tr> <td>Min 闽</td> <td>Amoy 厦门, Danzhou Hainan 海南儋州, Ding'an 定安, Fuzhou 福州, Gutian Pinghu 古田平湖, Hainan Wenchang 海南文昌, Jianyang 建阳, Jianou 建瓯, Northern Taiwan 台湾北部, Pingxi 平溪, Pucheng 浦城, Putian Daitou 莆田埭头, Putian Jiangkou 莆田江口, Shaxia Gaizhu 沙县盖竹, Shaowu 邵武, Taiwanese 台湾, Yun'ao 云澳, Zhangzhou 漳州 </td> </tr> <tr> <td>Wu 吴</td> <td>Luqiao 路桥, Ningbo 宁波, Shanghainese 上海, Shaoxing 绍兴, </td> </tr> <tr> <td>Xiang 湘</td> <td>Changsha 长沙, Chishan 赤山 </td> </tr> <tr> <td>Yue 粤</td> <td>Cantonese, Dongkeng 东坑, Gaobuxiansha 高埗冼沙, Hengli 横沥, Humenchigang 虎门赤岗, Shijie 石碣, Taishanese 台山, Tangxia 塘厦, Wangniudun 望牛墩, Wanjiang 万江, Zhongtangjiangnan 中堂江南 </td> </tr> <tr> <td>Jianghuai Mandarin<br>江淮官话</td> <td> Anqing 安庆, East Hefei 肥东, Nanjing 南京, Nantong 南通, Ningbo 宁波, Rugao 如皋, Wuhu 芜湖 </td> </tr> <tr> <td>Southwest Mandarin<br>西南官话</td> <td> Kunming 昆明, Lizhuang 李庄 </td> </tr> </tbody> </table> --- ### Survey of checked constituents #### Chinese - Checked refers to a phonotactic restriction between syllable and tone in Chinese: - Obstruent-closed syllables are associated with a different set of tones from open syllables and nasal-closed syllables. - <table style="width: 100%;"> <tbody style = "font-size: 26px"> <tr> <td></td> <td colspan="6">Meixian Hakka (Lee & Zee, 2009)</td> <td></td> <td style="color: white;" colspan="6">Shanghainese (Zee & Xu 2017)</td> </tr> <tr> <td>(2)</td> <td colspan="3">Checked</td> <td colspan="3">Unchecked</td> <td style="color: white;">(6)</td> <td style="color: white;" colspan="3">Checked</td> <td style="color: white;" colspan="3">Unchecked</td> </tr> <tr> <td></td> <td>Tone</td> <td>Word</td> <td>Gloss</td> <td>Tone</td> <td>Word</td> <td>Gloss</td> <td></td> <td style="color: white;">Tone</td> <td style="color: white;">Word</td> <td style="color: white;">Gloss</td> <td style="color: white;">Tone</td> <td style="color: white;">Word</td> <td style="color: white;">Gloss</td> </tr> <tr> <td></td> <td>55</td> <td>si<span style="color: red">p</span> 55</td> <td>"to collect"</td> <td>31</td> <td>i 31</td> <td>"rain"</td> <td></td> <td style="color: white;">5</td> <td style="color: white;">pɪ<span>ʔ</span> 5</td> <td style="color: white;">"pen"</td> <td style="color: white;">51</td> <td style="color: white;">pi 51</td> <td style="color: white;">"edge"</td> </tr> <tr> <td></td> <td></td> <td>i<span style="color: red">t</span> 55</td> <td>"wing"</td> <td>33</td> <td>i 33</td> <td>"clothes"</td> <td></td> <td style="color: white;">12</td> <td style="color: white;">bɪ<span style="color: white">ʔ</span> 12</td> <td style="color: white;">"other"</td> <td style="color: white;">13</td> <td style="color: white;">bi 13</td> <td style="color: white;">"skin"</td> </tr> <tr> <td></td> <td></td> <td>sa<span style="color: red">k</span> 55</td> <td>"stone"</td> <td>53</td> <td>i 53</td> <td>"idea"</td> <td></td> <td></td> <td></td> <td></td> <td style="color: white;">34</td> <td style="color: white;">pi 34</td> <td style="color: white;">"flat"</td> </tr> <tr> <td></td> <td>31</td> <td>se<span style="color: red">p</span> 31</td> <td>"garbage"</td> <td>11</td> <td>i 11</td> <td>"aunt"</td> <td rowspan="7"></td> </tr> <tr> <td></td> <td></td> <td>i<span style="color: red">t</span> 31</td> <td>"benefit"</td> <td></td> <td></td> <td></td> <td rowspan="7"></td> </tr> <tr> <td></td> <td></td> <td>sɔ<span style="color: red">k</span> 31</td> <td>"rope"</td> <td></td> <td></td> <td></td> <td rowspan="7"></td> </tr> </tbody> </table> --- ### Survey of checked constituents #### Chinese Checked refers to a phonotactic restriction between syllable and tone in Chinese: Obstruent-closed syllables are associated with a different set of tones from open syllables and nasal-closed syllables. <table style="width: 100%;"> <tbody style = "font-size: 26px"> <tr> <td></td> <td colspan="6">Meixian Hakka (Lee & Zee, 2009)</td> <td></td> <td colspan="6">Shanghainese (Zee & Xu 2017)</td> </tr> <tr> <td>(2)</td> <td colspan="3">Checked</td> <td colspan="3">Unchecked</td> <td>(3)</td> <td colspan="3">Checked</td> <td colspan="3">Unchecked</td> </tr> <tr> <td></td> <td>Tone</td> <td>Word</td> <td>Gloss</td> <td>Tone</td> <td>Word</td> <td>Gloss</td> <td></td> <td>Tone</td> <td>Word</td> <td>Gloss</td> <td>Tone</td> <td>Word</td> <td>Gloss</td> </tr> <tr> <td></td> <td>55</td> <td>si<span style="color: red">p</span> 55</td> <td>"to collect"</td> <td>31</td> <td>i 31</td> <td>"rain"</td> <td></td> <td>5</td> <td>pɪ<span style="color: red">ʔ</span> 5</td> <td>"pen"</td> <td>51</td> <td>pi 51</td> <td>"edge"</td> </tr> <tr> <td></td> <td></td> <td>i<span style="color: red">t</span> 55</td> <td>"wing"</td> <td>33</td> <td>i 33</td> <td>"clothes"</td> <td></td> <td>12</td> <td>bɪ<span style="color: red">ʔ</span> 12</td> <td>"other"</td> <td>13</td> <td>bi 13</td> <td>"skin"</td> </tr> <tr> <td></td> <td></td> <td>sa<span style="color: red">k</span> 55</td> <td>"stone"</td> <td>53</td> <td>i 53</td> <td>"idea"</td> <td></td> <td></td> <td></td> <td></td> <td>34</td> <td>pi 34</td> <td>"flat"</td> </tr> <tr> <td></td> <td>31</td> <td>se<span style="color: red">p</span> 31</td> <td>"garbage"</td> <td>11</td> <td>i 11</td> <td>"aunt"</td> <td rowspan="7"></td> </tr> <tr> <td></td> <td></td> <td>i<span style="color: red">t</span> 31</td> <td>"benefit"</td> <td></td> <td></td> <td></td> <td rowspan="7"></td> </tr> <tr> <td></td> <td></td> <td>sɔ<span style="color: red">k</span> 31</td> <td>"rope"</td> <td></td> <td></td> <td></td> <td rowspan="7"></td> </tr> </tbody> </table> --- ### Survey of checked constituents #### Chinese - Can checked constituents in Chinese be reduced to <strong>a category of syllable</strong>? - Yes. - Checked syllables in Chinese are obstruent-closed syllables. - <table> <tbody style = "font-size: 23px"> <tr> <td>Coda</td> <td>Example</td> <td>Coda </td> <td>Example</td> </tr> <tr> <td>/-p, -t, -k/</td> <td>Jianning 建宁 Gan</td> <td>/-p, -t, -k, -ʔ/</td> <td>Hainan Wenchang 海南文昌 Min</td> </tr> <tr> <td>/-p, -t/</td> <td>Xiuzhuan 秀篆 Gan</td> <td>/-p, -t, -ʔ/</td> <td>Anyi 安义 Gan</td> </tr> <tr> <td>/-p, -k/</td> <td>Zhongtang Jiangnan 中堂江南 Yue</td> <td>/-p, -k, -ʔ/</td> <td>Yun'ao 云澳 Min</td> </tr> <tr> <td>/-t, -k/</td> <td>Wengyuan 翁源 Hakka</td> <td>/-t, -k, -ʔ/</td> <td>Sandu 三都 Hakka</td> </tr> <tr> <td>/-k/</td> <td>Wangniudun 望牛墩 Yue</td> <td>/-t, -ʔ/</td> <td>Jishui (Luotian) 吉水螺田 Gan</td> </tr> <tr> <td>/-tŋ, -kŋ/</td> <td>Yugan 余干 Gan</td> <td>/-k, -ʔ/</td> <td>Dongkeng 东坑 Yue</td> </tr> <tr> <td></td> <td></td> <td>/-k, -l/</td> <td>Duchang 都昌 Gan</td> </tr> <tr> <td></td> <td></td> <td>/-ʔ/</td> <td>Wuning 武宁 Gan</td> </tr> </tbody> </table> --- ### Survey of checked constituents #### Chinese - Can checked constituents in Chinese be reduced to <strong>a category of tone</strong>? - Yes. - Checked tones refer to tones that are borne by obstruent-closed syllables. - <table> <tbody style = "font-size: 23px"> <tr> <td>Number of checked tones</td> <td>Language</td> <td>Tonal value</td> <td>Example</td> </tr> <tr> <td>Four</td> <td>Duchang 都昌 Gan</td> <td>45, 24, 3, 21</td> <td> /pɔk 45/ 博 "broad"<br> /lɔk 24/ 托 "support"<br> /mɔk 3/ 莫 "not"<br> /bɔk 21/ 薄 "thin" </td> </tr> <tr> <td>Three</td> <td>Hong Kong Cantonese</td> <td>5, 33, 2</td> <td> /jek 5/ 益 "benefit"<br> /jak 33/ 喫 "eat"<br> /jek 2/ 亦 "also" </td> </tr> <tr> <td>Two</td> <td>Wengyuan 翁源 Hakka</td> <td>2, 5</td> <td> /tak 2/ 答 "answer"<br> /thak 5/ 踏 "step" </td> </tr> <tr> <td>One</td> <td>Ganxian (Panlong) 赣县蟠龙 Gan</td> <td>5</td> <td>/taʔ 5/ 答 "answer"</td> </tr> </tbody> </table> --- ### Survey of checked constituents #### Chinese - Can checked constituents in Chinese be reduced to <strong>duration</strong>? - Maybe. - In the survey, the majority of checked syllables have a shorter duration than unchecked syllables. - But we also see cases where obstruent-closed syllables have comparable or even longer duration than open syllables: Zhangzhou 漳州 Min (Huang, 2018); Hong Kong Cantonese (Zhu et al., 2008) --- ### Is being "checked" a necessary phonological feature/category in Chinese languages? - No - "Checked" constituent in Chinese can be reduced to obstruent-closed syllable and their associated tones. - "Checked" constituent describes the phonological restriction between syllable shape and tone. --- ### Survey of checked constituent #### Chinese <table style="width: 90%;"> <tbody> </tr> <tr> <td style="width: 300px">Abstraction</td> <td style="width: 300px">V{p, t, k, ʔ} + tone</td> </tr> <tr> <td style="width: 300px">Necessary & Sufficient distinct category?</td> <td style="width: 300px">No</td> </tr> <tr> <td style="width: 300px">Phonological nature</td> <td style="width: 300px">A descriptor of phonological restriction</td> </tr> <tr> <td style="width: 300px">Reduced to</td> <td style="width: 300px">Obstruent-closed syllable + Specific tone(s)</td> </tr> <tr> <td style="width: 300px">Interaction with tone</td> <td style="width: 300px">Tone-dependent</td> </tr> </tbody> </table> --- ### Future direction - To collect articulatory production data (e.g. electroglottograph; airflow; larynscope) - To create more balanced stimuli for the sandhi tone difference analysis in the production experiment - To use more frequency-balanced stimuli for the sandhi tone perception experiment --- ### What do we mean when we say a constituent is "checked"? Burmese tones <table> <tbody style = "font-size: 23px;"> <tr> <td>(3)</td> <td style="color: red;">Checked</td> <td style="color: red;">/maʔ 51/</td> <td style="color: red;">"March"</td> <td>Falling-Creaky</td> <td>/maˀ 51/</td> <td>"female"</td> </tr> <tr> <td></td> <td>High</td> <td>/ma 44/</td> <td>"towering"</td> <td>Low</td> <td>/ma 22/</td> <td>"hard"</td> </tr> <tr> <tr> <td> </td> <td style="font-size: 20px;" colspan=6> (Watkins, 2001) </td> </tr> </tbody> </table> <img width = "700px" src="image/burmese_march_female_towering.png"/> <audio controls><source src="audio/7_march.wav" type="audio/wav"></audio><audio controls><source src="audio/8_female.wav" type="audio/wav"></audio><audio controls><source src="audio/9_towering.wav" type="audio/wav"></audio> - <span style="font-size:28px;">“Checked” refers to /ʔ/ closed syllables bearing only Falling tone (vs. other syllables can bear High, Low, Falling-creaky three tones).</span> --- <img src="image/xm_perception_manipulation.png" class = "padding"/><img src="image/xm_perception_pitchtier.png" class = "padding"/> <img width = "1000px" src="image/xm_perception_praatscript.png"/> <audio controls><source src="audio/xm_perception_original_demonstration.wav" type="audio/wav"></audio><audio controls><source src="audio/xm_perception_gl5_demonstration.wav" type="audio/wav"></audio> --- #### H1*-H2* <img width = "1000px" src="image/h1h2_illustration.png"/> - H1*-H2* can be noisy - Wrong F0 tracking - Wrong formant tracking - Affected by nasal resonance (nasal pole) - Instead of subtracting H2 as a method of controlling for voicing energy, we control for energy in linear models directly. --- #### Residual H1* As dependent variable: - H1* ~ predictors + RMS Energy - As independent variable: - H1* = b0 + b1 * RMS Energy - Residual H1* = H1* - b1 * RMS Energy --- ### Phonetics of checked constituents #### Yateé Zapotec <img src="image/yatee_map.svg" /> --- ### Phonetics of checked constituents #### Yateé Zapotec - Existing documentation of the language: Jaeger (1983); Jaeger & Van Valin (1982) - Focus on the consonants in the language - Fieldwork with two speakers of Yateé Zapotec since May 2020; - Collected 70 hours of field recording (remote and in-person); - Conducted a structured production experiment; - Ongoing elicitation and data collection; - Forthcoming perception study. --- ### Phonation and tone system in Yateé Zapotec Three phonations: Modal, Rearticulated, Checked - <svg width="600" height="500"> <text x="10" y="40" class="small">Modal</text> <text x="10" y="70" class="smallred">V</text> <text x="10" y="100" class="small">no gl</text> <text x="10" y="200" class="small">Rearticulated</text> <text x="10" y="230" class="smallred">VˀV</text> <text x="10" y="260" class="small">Mid-phased gl</text> <text x="10" y="360" class="small">Checked</text> <text x="10" y="390" class="smallred">Vˀ</text> <text x="10" y="420" class="small">Late-phased gl</text> <text x="300" y = "15" class="tiny">beginning</text> <text x="570" y = "15" class="tiny">end</text> <rect x="300" y="30" width="300" height="50" style="fill: white; stroke:black;stroke-width:2;" /> <text x="400" y="60" class="small">Vowel</text> <text x="300" y = "175" class="tiny">beginning</text> <text x="570" y = "175" class="tiny">end</text> <rect x="300" y="190" width="300" height="50" style="fill: white; stroke:black;stroke-width:2;" /> <text x="400" y="220" class="small">Vowel</text> <rect x="375" y="250" width="150" height="50" style="fill: white; stroke:black;stroke-width:2;" /> <text x="410" y="285" class="small">[+c.g.]</text> <text x="300" y = "335" class="tiny">beginning</text> <text x="570" y = "335" class="tiny">end</text> <rect x="300" y="350" width="300" height="50" style="fill: white; stroke:black;stroke-width:2;" /> <text x="400" y="380" class="small">Vowel</text> <rect x="450" y="410" width="150" height="50" style="fill: white; stroke:black;stroke-width:2;" /> <text x="480" y="445" class="small">[+c.g.]</text> </svg> --- ### Phonation and tone system in Yateé Zapotec <table style="width: 90%;"> <tbody style = "font-size: 25px"> <tr> <td></td> <td colspan = "3" style="font-weight: bold;">Modal</td> <td colspan = "3" style="font-weight: bold;">Rearticulated</td> <td colspan = "3" style="font-weight: bold;">Checked</td> </tr> <tr> <td style="font-weight: bold;">Low</td> <td>/ʂnːà/</td> <td>〈xna〉</td> <td>"red"</td> <td>/ʂnːàˀà/</td> <td>〈xna'a〉</td> <td>"my mother"</td> <td>/nàˀ/</td> <td>〈nha〉</td> <td>"there"</td> </tr> <tr> <td style="font-weight: bold;">High</td> <td>/lːá/</td> <td>〈la〉</td> <td>"hot"</td> <td>/zíˀí/</td> <td>〈zi'i〉</td> <td>"heavy"</td> <td>/ʐáˀ/</td> <td>〈xha〉</td> <td>"clothes"</td> </tr> <tr> <td style="font-weight: bold;">Rising</td> <td>/gǎ/</td> <td>〈ga〉</td> <td>"nine"</td> <td>/wèˀé/</td> <td>〈we'e〉</td> <td>"get drunk"</td> <td>/wěˀ/</td> <td>〈we'〉</td> <td>"wound"</td> </tr> <tr> <td style="font-weight: bold;">Falling</td> <td>/jô/</td> <td>〈yo〉</td> <td>"dirt"</td> <td>/jóˀò/</td> <td>〈yo'o〉</td> <td>"house"</td> <td>/kâˀ/</td> <td>〈ka'〉</td> <td>"like that"</td> </tr> </tbody> </table> <audio controls><source src="audio/10_red.wav" type="audio/wav"></audio><audio controls><source src="audio/11_mother.wav" type="audio/wav"></audio><audio controls><source src="audio/12_there.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/13_hot.wav" type="audio/wav"></audio><audio controls><source src="audio/14_heavy.wav" type="audio/wav"></audio><audio controls><source src="audio/15_clothes.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/16_nine.wav" type="audio/wav"></audio><audio controls><source src="audio/17_drunk.wav" type="audio/wav"></audio><audio controls><source src="audio/18_wound.wav" type="audio/wav"></audio><br> <audio controls><source src="audio/19_dirt.wav" type="audio/wav"></audio><audio controls><source src="audio/20_house.wav" type="audio/wav"></audio><audio controls><source src="audio/21_like_that.wav" type="audio/wav"></audio> - Stimuli: 49 monosyllabic words were elicited in isolation from a male speaker of Yateé Zapotec. Each word was repeated three times. --- Acoustic measures: F0: Higher the F0, higher the pitch <img width = 550px src="image/f0_100_200.png" /><br> <audio controls><source src="audio/F0_100hz.wav" type="audio/wav"></audio><audio controls><source src="audio/F0_200hz.wav" type="audio/wav"></audio> --- Acoustic measures: Residual H1* **(Chai & Garellek, 2022)**: The lower the energy of H1, the more glottal constriction <img src="image/h1_illustration.png" /> <audio controls><source src="audio/spectral_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/spectral_creaky.wav" type="audio/wav"></audio> --- Acoustic measures: Harmonic-to-noise ratio (HNR): The lower the HNR, the noisier <img width = 650px src="image/hnr_illustration_modal.png"><img width = 650px src="image/hnr_illustration_noisy.png"> <audio controls><source src="audio/no_noise_02082023.wav" type="audio/wav"></audio><audio controls><source src="audio/all_noise_02082023.wav" type="audio/wav"></audio> <p style="font-size:23px">(Generated using UCLA Voice Synthesizer; Kreiman et al., 2006)</p> --- Acoustic measures: Amplitude of voicing - Measured by Strength of Excitation (SoE): The amplitude of voicing is usually low during glottalization. <img width = 650px src="image/soe_wave_illustration.svg"> <audio controls><source src="audio/a_modal.wav" type="audio/wav"></audio><audio controls><source src="audio/a_creaky.wav" type="audio/wav"></audio> --- Acoustic measures: <p class="subpoint">F0</p> <p class="subpoint">Duration</p> <p class="subpoint">Amplitude of fundamental frequency (Residual H1*);</p> <p class="subpoint">Harmonic-to-noise ratio (HNR);</p> <p class="subpoint">Amplitude of voicing (SoE)</p> Calculated using VoiceSauce (Shue et al., 2011) **Question**: Which parameter(s) most effectively distinguish <span style="color: #dfc27d">checked</span> phonation from <span style="color: #018571">modal</span> and <span style="color: #80cdc1">rearticulated</span> phonation types? --- Results of Linear Discriminant Analaysis (LDA) <table style="width: 100%;"> <tbody style = "font-size: 25px"> <tr> <td style="vertical-align: middle;text-align:right; width:400px">26.79%<br> <span style="color: #dfc27d">Checked</span> vs. <span style="color: #018571">Modal</span><br> Residual H1* (final)<br>Amplitude of voicing (final) </td> <td style="vertical-align: middle;text-align:center;display: inline-block;"><img width = "700px" src="image/yz_lda.svg"></td> </tr> <tr> <td></td> <td style="vertical-align: top; text-align:center;display: inline-block;">73.21%<br> <span style="color: #dfc27d">Checked</span> vs. <span style="color: #80cdc1">Rearticulated</span><br> Residual H1* (medial, initial), Amplitude of voicing (medial) </td> <td></td> </tr> </tbody> </table> --- <img width = 650 src="image/yz_h1cr_modal_check_annotate.png" /><img width = 650 src="image/place_holder_yz_h1cr.png" /> --- <img width = 650 src="image/yz_h1cr_modal_check_annotate.png" /><img width = 650 src="image/yz_h1cr_reart_check_annotate.png" /> --- <img width = 650 src="image/yz_soe_modal_check_annotate.png" /><img width = 650 src="image/place_holder_yz_h1cr.png" /> --- <img width = 650 src="image/yz_soe_modal_check_annotate.png" /><img width = 650 src="image/yz_soe_reart_check_annotate.png" /> --- <span style="color: #dfc27d">Checked</span> vs. <span style="color: #80cdc1">Rearticulated</span>: Rearticulated vowels have more glottalization <strong>in the middle</strong>. - <span style="color: #dfc27d">Checked</span> vs. <span style="color: #018571">Modal</span>: Checked vowels have more glottalization <strong>in the end</strong>. - The **late-phased glottalization** most effectively distinguish checked phonation from modal (no gl) and rearticulated (mid-phased) phonation types in Yateé Zapotec. ---