CPSC05 Language Center
Team Members: Chris Harvey, Brendan Kennedy, Ryan Menezes,
Brian Ramaswami, Kevin Wisdom
Faculty Advisor: Dr. James Hunter
Sponsor: English Language Center
Liaison: Chris Sharman
Project background or overview:For the past 15 years, Gonzaga's English Language Center has been using a custom-built (MS Access)database for storing, analyzing, and tracking English language learners' spoken language errors. Primarily this is used for providing corrective feedback to our ESL students, but a no less important function is to assemble a data base of linguistic data for research into language acquisition. The database currently contains approximately 500k words, in the form of about 50k utterances, of which about 10k have been analyzed for syntactic and semantic content.
While the application has served our needs very well, it has considerable limitations. First, it is not accessible to students, so the corrective feedback functions are only possible via other platforms such as Blackboard, audio recording, pdf outputs, and so on. Second, this rich data is only available to ELC faculty, which is a loss to researchers in NLP (natural language processing) and to GU, since this is the only project of its kind to date.
In order to make the application more practically useful and scalable, it needs to be migrated to a SQL server/web-based application. For this to happen, we would need to translate existing functionality to web-based technologies and create additional applications so that a) any instructors anywhere can participate in the data collection and corrective feedback cycles; b) language researchers would have access to a searchable database of learner language; c) students could access their own language samples and record their own reformulations (corrections) for assessment purposes. Ideally, voice recognition would provide text versions as well; d) instructors could provide assessment of these reformulations (correct/incorrect); e) analysts could enter syntactic/semantic analysis of the original errors and reformulations (the taxonomy for this analysis has already been elaborated, but the application must be flexible enough to allow for further elaboration without the need for additional coding). Additionally, existing NLP technologies from the field of computational linguistics could be harnessed to facilitate the automatic analysis (POS tagging, parsing, automatic error detection, etc.,) to automate several functions and move towards machine learning, rendering the applications both more functional and creating the first corrective feedback "machine" of its kind.The proposal here is to divide the above into discrete modules which could be tackled as a series of projects for computer science senior projects. Evidently, some are considerably more challenging than others, but each of them has not only a real-world application and an immediate demand, but also involves skillsets (database applications, AI, voice recognition, etc.) which are presumably highly transferable.
Statement of Project goal:
In this phase, the goal is to migrate the existing data to a SQL, web-based platform and to add/adapt functionality, including audio recording capability.
Project Deliverables:
1. Web interface with secure login (for instructors, researchers, and students) – design and look of web interface distinct for three user groups
2. Functionality for each user group:
a. Instructors: set up courses, add students, input student language production data, record reformulations, etc.
b. Students: access own language production data, record reformulations, test self, complete periodic timed reformulation tests
c. Researchers: access anonymous language production data, categorized by proficiency level and first language
Project Management:
1. WBS and project schedule to be determined.
2. Initial meeting to explain vision for the project and discuss some ideas on how to best accomplish our goal.
3. Bi-weekly meetings with sponsor, maybe more often to keep the project on track.
4. Weekly status reports (Phone call or email).
Project Research:
1. Understanding the current Access application.
2. Creating a secure web application that users can interact with.
3. Creating a web application that is visually appealing to users.