Research Opportunities in Linguistics

From lingwiki

Research Exchange

This page serves as a research exchange for faculty and graduate students in linguistics who are seeking research assistance and undergraduates who are looking for opportunities to take part in research.

Posts concerning research assistance should include:

• Title of the project
• Specific tasks/duties you would like done
• Any specific skills or abilities that are required
• Time commitment expected (ex. 3 hours per week, a total of 40 hours over the course of the summer, up to 20 hours total, etc.)
• Start date
• Compensation information (Uncompensated, experience only? UROP? Course credit? Work-study? Hourly rate?)
• Contact information
• Status of the project as "open" or "closed"

Once the project is finished or you have enough help, please mark the position as closed or delete the posting.

Open Research Positions

Semantics & Baseball (OPEN, Prof. Ezra Keshet): Looking for students to help transcribe radio broadcasts of baseball games in order to examine how this language encodes the action going on in the game. Time commitment can vary from 1 game (4-5 hours' work) up to as many games as you want, starting in early May. Work is uncompensated, but credit opportunities are possible. Email me at ekeshet@umich.edu if you are interested.

Database of Phonetic Symbols (SOMEWHAT OPEN, Terry Szymanski): I'm looking for volunteers to enter data into an electronic database of phonetic symbols in current or past use by linguists. This database will have applications in language-processing software and could help improve optical character recognition of scanned linguistic texts. Work as much or as little as you like. No compensation: email tdszyman@umich.edu

Discourse Prosody (OPEN, Joseph Tyler): Looking for a volunteer to help measure prosodic features (e.g. amplitude, speech rate) in spoken recordings. You would be working with Praat and a Praat script I have written to automatically collect new prosodic measures to correlate with the discourse structure of a news article. The goal of the project is to see what prosodic measures speakers use that indicate the large-scale structure of spoken language. Experience with Praat preferred. Time commitment is flexible, although initial tasks may take 5 hours. No compensation. Email me at jctyler@umich.edu if you are interested.

Multi-Lingual Grammar Annotation (OPEN, Terry Szymanski): I am assembling a corpus of electronic texts of language description materials (i.e. grammars and lexicons) for as many languages as possible. I am looking for volunteer (unpaid, non-credit) help in building this corpus. Duties include:

• Search the internet to find relevant electronic texts and add them to the corpus
• Annotate texts using the GATE annotation software. Annotation includes labeling foreign words in the text and identifying glosses of these words within the text.
• Work on any language or language family that interests you
• Get experience using NLP tools (GATE: the General Architecture for Text Engineering) and concepts in language documentation, language engineering, information extraction, and machine learning

Contact Terry Szymanski tdszyman@umich.edu if you are interested.

Phonological simplification processes: corpus searches (OPEN, Andries Coetzee): I'm looking for a student (or students?) to help me search through large speech corpora for specific examples of phonological simplification processes. In order to do this, you need to have some experience in scripting/programming in Perl, Python or some similar computer language. I have some knowledge, but not enough to pull this off on my own. I am hoping that we can put our skills together, and figure out how to do this together. In addition to the research experience that you will gain, I can also offer you compensation of $10 per hour for up to 20 hours of work. Email me at coetzee@umich.edu.

An Inventory of English Verb Usage (OPEN, Nick Ellis & Matt O’Donnell): A GRADUATE student is required to work up to 10 hours a week through Fall and Winter on a project building a linguistic database. This is constructed by defining search parameters for a part-of-speech tagged and dependency parsed corpus that match the descriptions given in the Cobuild project. The descriptions are linguistic and stated in clause structure and word class terms and need to be translated into search syntax and program code. Once in code they will be submitted to the search program which collects matching instances and writes them to the database. The RA will need to become familiar with and contribute to this software and database architecture and be able to run new patterns through the system and diagnose problems. Knowledge of information retrieval in general and working with linguistic databases and natural language processing (NLP) tools (preferably in Python and or Perl) and XML processing will be an advantage. Further desirable skills include knowledge of web application frameworks (particularly Django and Python) and some course background in network science and network analysis. Compensation $10-14 per hour, depending on relevant experience. Email ncellis@umich.edu or mbod@umich.edu if you are interested.

Checking the Inventory of English Verb Usage (OPEN, Nick Ellis & Matt O’Donnell): An undergraduate student is required to work up to 10 hours a week through Fall and Winter on a project assessing the accuracy of a linguistic database of verb argument constructions (VACs). This is based on the Grammar Patterns from the Cobuild project. These patterns are stated in terms of clause structure components and word class restrictions. A linguistics student with a grounding in grammar and syntactic theory is required to work through the patterns in the volume and turn them into a template that can be used in the search software. The template and software is already under development and the RA will be trained in its usage. Once the results have been collected into the database for a pattern, the RA will use the web interface to examine a sample of the retrieved sentences to check that they are genuine instances of the VAC pattern. The software has a mechanism for reporting incorrect matches and carrying out the necessary precision and recall analysis. Background in programming and computational linguistics is not necessary but would be helpful. Compensation $10 per hour. Email ncellis@umich.edu or mbod@umich.edu if you are interested.