Keelan's Home Page
This is me rounding Devil's Elbow in the 2003 Mt. Diablo Challenge. It
is a bicycle race near the Bay Area, and I was one of 9 unicyclists in
the pack that year. The total distance was 10.8 miles and the gain in
elevation was 3,249 feet (my time 2:10).
I'm a Research Scientist at Educational Testing Service, where I conduct research on automated scoring of non-native spoken responses to assessments such as the TOEFL iBT and TOEIC. I received my PhD in 2009 from the University of Pennsylvania in the Linguistics Department. My thesis advisor was Bill Labov.
I have two blogs:
- The Evanini Family: A blog mostly dedicated to providing updates about our family to out-of-town family members.
- SpeechTechie: A repository for posting some coding / application details that I have found useful for my research.
- 2013. Keelan Evanini, Shasha Xie, and Klaus Zechner. Prompt-based Content Scoring for Automated Spoken Language Assessment. Proceedings of the Eighth Workshop on Innovative Use of NLP for Building Educational Applications, pages 157-162, Atlanta, Georgia, June 13 2013. [pdf]
- 2013. Xinhao Wang, Keelan Evanini, and Klaus Zechner. Coherence Modeling for the Automated Assessment of Spontaneous Spoken Responses. Proceedings of NAACL-HLT, pages 814-819, Atlanta, Georgia, 9-14 June 2013. [pdf]
- 2013. Keelan Evanini and Becky Huang. Production of English vowels by speakers of Mandarin Chinese with prolonged exposure to English. Proceedings of Meetings on Acoustics, volume 18. [pdf]
- 2012. K. Zechner, K. Evanini, and C. Laitusis. Using Automatic Speech Recognition to Assess the Reading Proficiency of a Diverse Sample of Middle School Students. Proceedings of the InterSpeech Workshop on Child, Computer, and Interaction. [pdf]
- 2012. K. Evanini and B. Huang. Automatic detection of [th] pronunciation errors for Chinese learners of English. Proceedings of the International Symposium on Automatic Detection of Errors in Pronunciation Training, Stockholm. [pdf]
- 2012. S. Xie, K. Evanini, and K. Zechner. Exploring content features for automated speech scoring. Proceedings of the 2012 Meeting of the North American Association for Computational Linguistics: Human Language Technologies (NAACL-HLT). [pdf]
- 2011. K. Evanini and K. Zechner. Using crowdsourcing to provide prosodic annotations for non-native speech. Proc. Interspeech. [pdf]
- 2011. S.-Y. Youn, K. Evanini, and K. Zechner. Non-scorable response detection for automated speaking proficiency assessment. Proc. ACL-HLT, The 6th Workshop on Innovative Use of NLP for Building Educational Applications. [pdf]
- 2011. X. Sun and K. Evanini. Gaussian mixture modeling of vowel durations for automated assessment of non-native speech. Proc. ICASSP. [pdf]
- 2011. K. Evanini, C. Lai, and K. Zechner. The importance of optimal parameter setting for pitch extraction. Proceedings of Meetings on Acoustics. [pdf]
- 2010. L. Chen, K. Evanini, and X. Sun. Assessment of non-native speech using vowel space characteristics. Proc. SLT.
- 2010. A. Dinkin and K. Evanini. An elementary linguistic definition of Upstate New York. Selected Papers from NWAV 38, University of Pennsylvania Working Papers in Linguistics, v. 16.2. [pdf]
- 2010. D. Suendermann, J. Liscombe, R. Pieraccini, and K. Evanini. "How am I Doing?": A new framework to effectively measure the performance of automated customer care contact centers. In Advances in Speech Recognition: Mobile Environments, Call Centers, and Clinics, ed. A. Neustein., p. 155 - 180. Springer.
- 2010. K. Evanini, D. Higgins, and K. Zechner. Using Amazon Mechanical Turk for transcription of non-native speech. Proc. NAACL HLT, Workshop on Creating Speech and Language Data With Amazon's Mechanical Turk. [pdf]
- 2009. J. Tauberer and K. Evanini. Intrinsic vowel duration and the post-vocalic voicing effect: Some evidence from dialects of North American English. Proc. Interspeech. [pdf | slides]
- 2009. K. Evanini, S. Isard, and M. Liberman. Automatic formant extraction for sociolinguistic analysis of large corpora. Proc. Interspeech. [pdf]
- 2009. D. Suendermann, K. Evanini, J. Liscombe, P. Hunter, K. Dayanidhi, and R. Pieraccini. From rule-based to statistical grammars: Continuous improvement of large-scale spoken dialog systems. Proc. ICASSP. [pdf]
- 2008. D. Suendermann, J. Liscombe, K. Evanini, K. Dayanidhi, R. Pieraccini. C^5. Proc. SLT. [pdf]
- 2008. K. Evanini, P. Hunter, J. Liscombe, D. Suendermann, K. Dayanidhi, R. Pieraccini. Caller Experience: A method for evaluating dialog systems and its automated prediction. Proc. SLT. [pdf]
- 2008. A shift of allegiance: The case of Erie and the North / Midland dialect boundary. Selected Papers from NWAV 36, University of Pennsylvania Working Papers in Linguistics, v. 14.2. [pdf]
- 2008. Classifying and clustering dialects of North American English. Proc. NESCAI. [pdf]
- 2007. K. Evanini, D. Suendermann, R. Pieraccini. Call classification for automated troubleshooting on large corpora. Proc. ASRU. [pdf]
- 2007. How to formalize variation: Stochastic OT models and /s/ deletion in Spanish. Proceedings of LACUS 33, Linguistic Association of Canada and the US. [pdf]
- 2007. Book note for American English: dialects and variation. Language in Society 36(1):7-8.
- 2007. The phonetic realization of pitch accent in Huave. Proceedings of the 33rd Annual Berkeley Linguistics Society. [pdf]
- 2006. T. Wolford and K. Evanini. Features of AAVE as Features of PRE: A Study of Adolescents in Philadelphia. Selected Papers from NWAV 34, University of Pennsylvania Working Papers in Linguistics, v. 12.2. [pdf]
- 2005. K. Evanini and G. Nguyen. Papers from NWAVE 32, University of Pennsylvania Working Papers in Linguistics, v. 10.2.
The title of my dissertation is The permeability of dialect boundaries: A case study of the region surrounding Erie, Pennsylvania. The entire dissertation can be downloaded from Penn's electronic disseration repository. Individual chapters are available here:
- Front Matter: title page, acknowledgements, abstract, table of contents, list of figures, list of tables
- Chapter 1: introduction to the empirical, methodological, and theoretical aims of the dissertation
- Chapter 2: a review of the evidence from earlier dialect geography studies (LAMSAS, DARE) that show Erie's phonology and lexicon to be Northern
- Chapter 3: a description of my data collection procedure, including how I contacted speakers for inclusion in my corpus, a description of the interview materials, and how the interviews were processed
- Chapter 4: a description of the procedure I used for automatic vowel analysis including forced alignment, measurement point selection, and formant prediction
- Chapter 5: natural break maps for 17 vowel classes
- Chapter 6: a study of the merger of /o/ and /oh/ in the Erie region
- Chapter 7: a study of other vowels that pattern differently in the Midland and the North (/ow/, /uw/, /ae/, /o/)
- Chapter 8: an investigation into the use of Midland lexical and morphosyntactic features in Erie
- Chapter 9: the settlement history of Erie and how it helps explain the course of linguistic change in the region
- Chapter 10: some brief concluding remarks
- Appendices: materials used in the interview, including the word list, minimal pairs, sentences for acceptability judgments, and DARE's reading passage
These are programs that I have written and found useful for my research. The quality of documentation, portability, usability, etc. varies widely among them. Please feel free to contact me if you have any questions about them, or if you'd like to suggest improvements.
- quickTrans: a collection of shell, python, and Praat scripts that I use for transcribing interviews [download | README]
- extractFormants: a program for automatically obtaining vowel formant measurements; takes as input a speech audio file and a word- and phone-level transcription produced by forced alignment, outputs a set of formant measurements for each vowel in the audio file [download | README]
- vowel.r: some R functions I use for plotting vowel systems [download | example data file]