Using Grammars in On-line Education: Automatic error correction to improve writing skills

Add to your list(s) Download to your calendar using vCal

• Dan Flickinger, CSLI, Stanford University
• Monday 17 June 2013, 12:00-13:00
• FW11, Computer Laboratory.

If you have a question about this talk, please contact Ekaterina Kochmar.

Computational linguists who develop grammar implementations often begin with the motivation to encode their hypotheses about the particular structures and the general principles which illuminate the analysis of a given language. Often these hypotheses are tested either on naturally occurring text corpora, or on systematically constructed test suites illustrating the range of linguistic phenomena under study, including both well-formed and ill-formed example sentences. Since many applications that make use of grammar implementations emphasize robustness of analysis over precision, it is nice for the grammarian to encounter an application where precision is demanded. One such application can be found in online education courses designed to teach basic writing skills, where students are given exercises in which they construct sentences whose grammaticality is then judged by the system.

In this talk I draw on current work incorporating the English Resource Grammar (developed at CSLI ) into an existing online course for teaching sentence composition to elementary school students. Adaptation of this linguistically informed grammar has involved both extensions via so-called mal-rules, as well as reductions in the grammar’s coverage via masking to avoid unwanted ambiguity, given the restricted vocabulary made available to the students for each exercise. While the implementation currently focuses on judging syntactic well-formedness, we have also added some support for identifying semantic errors, by testing the equivalence of the semantic representation that the grammar assigns to the student’s sentence with that of a set of correct answers supplied for that exercise. I will conclude with a report on the efficacy of this approach based on improvements in student scores year over year on state-administered end-of-year exams in one large urban school district where the EPGY writing course has been in use for several years.

This talk is part of the NLIP Seminar Series series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Cambridge Centre for Data-Driven Discovery (C2D3)
• Cambridge Forum of Science and Humanities
• Cambridge Language Sciences
• Cambridge talks
• Chris Davis' list
• Department of Computer Science and Technology talks and seminars
• FW11, Computer Laboratory
• Graduate-Seminars
• Guy Emerson's list
• Interested Talks
• Language Sciences for Graduate Students
• NLIP Seminar Series
• PMRFPS's
• Simon Baker's List
• Trust & Technology Initiative - interesting events
• bld31
• ndk22's list
• ob366-ai4er
• rp587

Note that ex-directory lists are not shown.