Abstract

Second/foreign language writing in European languages is increasingly done with the help of digital online tools (AI powered translators and/or dictionaries). In the research presented here, the impact of such tool use on the writing process and on the output produced by adolescent and young adult learners of English as a foreign language is investigated. Texts produced by a total of 274 learners whose first or dominant languages are either French or German are analyzed. 

In a first series of analyses, I compare to what extent writing with such tools affects metrics of lexical and syntactic complexity, of accuracy, and of fluency (number of tokens produces per time unit). Moreover, I investigate the impact of a short instruction on the characteristics and features of online tools on their use in the writing process.

In a second step, I demonstrate to what extent the texts produced can be automatically categorized (via machine-learning algorithms based on the texts' unigrams and bigrams) as to whether they were produced with/without tools and as to whether they were written by learners who are dominant in French or German. The lexical and collocational units most strongly associated with the two languages and the two experimental conditions are discussed.

I will end the talk thinking possible implications of the results for second/foreign language teaching and learning.