About Me

TL;DR

I am a Linguistics PhD student at USC. My research focuses on the perception and production in speech comunication, using tools including statistical/computational models and behavioural experiments.


Research

phonetics & phonology | computational linguistics | natural language processing | syntax

I am currently a PhD student at USC. My current projects focus on the functional-cognitive factors that may condition tonal coarticulation, beyound the scopes of phonetics-phonology interface.
I obtained my MA degree from the Graduate Institute of Linguistics, National Taiwan University, Taipei, Taiwan, where I worked with Prof. Chenhao Chiu in the Speech Behaviour and Science Lab. One main focus of my research has been phonetics and phonology and their intersection. My thesis discusses the phonetic and phonological status of tonal coarticulation and its production and perception for tone language speakers. I am also interested in the intersections of phonetics/phonology with computational linguistics, especially the language modeling of speech. One main project I have been working on is the simulation of tone communication under tonal coarticulation based on speaker/listener neural agents. I have also been investigating the learning of Squliq Atayal pre-uvular vowel lowering based on generative adversarial networks.
Beyond the study of sounds, my broader interest is how language as a unique device of humankind stands out from other cognitive modalities, as well as what language has in common with them. As such, I take delight in the research of a large variety of linguistics subfields, including syntax, language acquisition, and psycholinguistics.
Aside from theoretical research, my research also spans to the field of natural language processing. I worked with Prof. Hsuan-Lei Shao on legal judgment text similarity, domain-specific collocation extraction, and word network properties. I also integrate my expertise in phonetics with speech recognition. I am currently working on an automated speech correctness system aimed for L2 learners and speech therapy patients. I also worked as a full-time RA for Dr. Lun-Wei Ku in the NLPSA lab, Institue of Information Science, Academia Sinica, Taipei, Taiwan. My principle research interests centered around large language models (LLMs), including vision-enhanced LLMs, the improvemnt of LLMs' abstract reasoning ability and the investigation of the relationship between cognition and language through LLMs.

Languages

黃柏瑄

Resumé

Education

2024–
University of Southern California

PhD | Department of Linguistics

2020–2023
National Taiwan University

MA | Graduate Institute of Linguistics

Thesis: Production and perception of tonal coarticulation in Taiwan Mandarin and Taiwan Southern Min

2017–2020
National Chengchi University

BA | Department of English

2015–2020
National Chengchi University

BA | Department of European Languages and Cultures

Research Experience

February 2024–June 2024
Institute of Information Science, Academia Sinica

Full-time Research Assistant

July 2021–February 2024
Graduate Institute of Linguistics, National Taiwan University

Research Assistant

August 2023–December 2023
Graduate Institute of Political Science, National Taiwan Normal University

Research Assistant

July 2021–October 2022
Graduate Program of Teaching Chinese as a Second Language, National Taiwan University

Research Assistant

Teaching Experience

Spring semester 2023
Graduate Institute of Brain and Mind Sciences, National Taiwan University

Teaching Assistant

Technology of Brain Science

Lecturers: Prof. Joshua G.O. Soo & Prof. Chia-Lin Lee

Spring semester 2023 & Spring semester 2022
Graduate Institute of Linguistics, National Taiwan University

Teaching Assistant

Discovering Languages

Lecturer: Prof. Chenhao Chiu

Spring semester 2021 & Fall semester 2020
Department of Foreign Languages and Literatures, National Taiwan University

Teaching Assistant

Freshman English

Lecturer: Prof. Wen-Hsien Hsu

Invited Talks

March 2024

Electroglottography in linguistics research with Python

Institute of Linguistics, Academia Sinica, Taipei

October 2023

Face extraction with MediaPipe in linguistics research

Speech Behaviours and Speech Sciences In-class Talk, Taipei

November 2022

Face extraction with MediaPipe in linguistics research

MultiMoco NTU Workshop, Taipei

November 2021

Using PCA and LDA for ultrasound image classification

The 1st NTU GIL Ultrasound Workshop, Taipei

Publication


Publications

Huang, Po-Hsuan and Her, One-Soon. 2024. On a dichotomy of question types: The case of Paiwan. Concentric: Studies in Linguistics, 50(1), 20–56. doi: 10.1075/consl.00033.her

Chiu, Chenhao, Huang, Jianzhi and Huang, Po-Hsuan. 2023, December 4–8. Perceptual identification of high vowels in Taiwan Mandarin. The Journal of the Acoustical Society of America, 154, A35.

Huang, Po-Hsuan and Chiu, Chenhao. 2023. Perception of coarticulated tones in Taiwan Mandarin and Taiwan Southern Min. In: Radek Skarnitzl & Jan Volín (Eds.), Proceedings of the 20th International Congress of Phonetic Sciences, 446–450. Guarant International.

Chiu, Chenhao and Huang, Po-Hsuan. 2023. Lip postures of high vowels in Taiwan Mandarin. In: Radek Skarnitzl & Jan Volín (Eds.), Proceedings of the 20th International Congress of Phonetic Sciences, 1052–1056. Guarant International.


Conference Articles

Chiu, Chenhao, Huang, Po-Hsuan and Huang, Jian-Zhi. 2024, June 24–25. Coordination of tongue and lips in Taiwan Mandarin high rounded vowels. The 13th Ultrafest X: Conference on Ultrasound Imaging for Speech and Language Research. Aizu, Japan.

Huang, Po-Hsuan. 2024, May 13–17. Production and perception of tonal coarticulation through computational communication simulation. The 13th International Seminar on Speech Production (ISSP 2024). Autrans, France.

Chiu, Chenhao, Chang, Cheng-Hsiang, Huang, Jian-Zhi and Huang, Po-Hsuan. 2024, May 13–17. Enhancing lip contrasts between /u/ and /y/ in Taiwan Mandarin. The 13th International Seminar on Speech Production (ISSP 2024). Autrans, France.

Huang, Po-Hsuan and Shao, Hsuan-Lei. 2024, March 11–15. Iterated domain-specific collocation extraction through mutual information in Mandarin. The 30th Conference on Natural Language Processing (NLP 2024). Kobe, Japan.

Huang, Po-Hsuan and Liu, Yi-Chen. 2023, October 21. Ultrasonic-and-optical-imaging-assisted automated speech correctness judgment model. Tunghai International Conference on Second Language Teaching and Research (SLTR.THU). Taichung, Taiwan.

Huang, Po-Hsuan and Chiu, Chenhao. 2023, May 26–27. Tonal Coarticulation in Taiwan Mandarin and Taiwan Southern Min. Hanyang International Symposium on Phonetics and Cognitive Sciences of Language (HisPhonCog). Soeul, Korea.

Huang, Po-Hsuan and Chiu, Chenhao. 2022, November 3–4. Articulation and syllabic affiliation of prenuclear glide in Taiwan Mandarin. Ultrafest X: Conference on Ultrasound Imaging for Speech and Language Research. Manchester, United Kingdom.

Huang, Po-Hsuan, Lai, Jing-Hong and Liu, Te-Hsin. 2022, September 23–25. On the Interaction between L1 Transfer and Universal Constraints – Evidence from the Acquisition of Mandarin Tones by French Speakers. The 34th North American Conference on Chinese Linguistics (NACCL). Bloomington, IN, United States.

Huang, Po-Hsuan and Chiu, Chenhao. 2022, January 6–9. Vowel-glide distinction in high vocoid diphthong structures in Squliq Atayal: An ultrasound and acoustic study. The 96th Annual Meeting of the Linguistic Society of America (AML). Washington, DC, United States.

Huang, Po-Hsuan. 2021, October, 23. Vowel–glide distinction in high vocoid diphthong structures in Squliq Atayal: An ultrasound and acoustic study. The 22nd National Conference on Linguistics (NCL). Taipei, Taiwan.

Huang, Po-Hsuan and Her, One-Soon. 2021, October 15–16. On a two-way distinction of questions in Paiwan. The 7th Young Scholars’ Symposium on Taiwan Languages (YSSTL). Online.

Packages & Softwares

  • Praat
  • Pandas
  • PyTorch
  • TensorFlow
  • PsychoPy/PsychoJS
  • E-Prime
  • Matlab
  • NumPy
  • Plotly
  • MediaPipe
  • Seaborn
  • Glob
  • Github
  • Pyplot
  • Tkinter
  • lme4
  • PyPI
  • scikit-learn
  • Deepedge

Portfolio

Chinese Political Speech Word Net

Chinese Political Speech Word Net

Text Mining

PTT Crawler

Text Mining
pymi demo

pymi

Text Mining

Ultrasonic-and-optical-imaging-assisted speech correctness judgment model

Ultrasound

UltraSelectArea

Ultrasound

PyTiers

Acoustics

PlotPolarGAMM

Ultrasound

input_detect

Acoustics

Contact

Los Angeles CA 90089

3601 Watt Way Grace Ford Salvatori 301

benson32169@gmail.com