O n f e r e n c e

Robert Reynolds 
Universitetet i Tromsø, Tromsø, Norway 
[e-mail] 
Automatic Evaluation of Potential Targets for Textual Enhancement: 
Identifying Optimal Sentences for Learner Uptake 
Bio data 
Robert Reynolds
has taught Russian for 11 years at institutions of higher education. He 
received an MA in Slavic Linguistics at The Ohio State University in 2009, where he studied 
morphological theory and computational linguistics. Robert is currently a PhD Candidate at 
the University of Tromsø where he researches intelligent computer-assisted language 
learning. 
Abstract 
Drawing on input processing research in second language acquisition theory, I explore 
potential means for automatically rating individual sentences according to the likelihood 
that they will lead to learner uptake of a given target language pattern. 
Natural Language Processing (NLP) technology has been implemented in CALL 
environments to help learners to study authentic texts (language produced by native 
speakers for native speakers). For example, Meurers et al. 2010 developed an application 
that allows a learner to automatically embed grammar exercises within texts found on the 
Internet (English, German, or Spanish). Assuming that a given text has many examples of 
the target language pattern, the application would ideally be able to automatically 
determine which tokens in a text are most optimal for learner uptake. Until now, this 
question has been avoided by highlighting either all target structures (which can be 
overwhelming), or only a randomly selected set. The proposed measures would help such 
applications to intelligently select optimal tokens. 
The solution developed in this study is based on VanPatten (1996, etc.), who posits ten 
principles of input processing. These principles represent strategies that L2 learners rely 
on to find meaning in a text. I explain how these principles can be operationalized. Future 
work will test these measures against language teacher intuitions and language learner 
performance.