Understanding the writing process and the production of sequences in a language provides us with information on how language users and learners see structures and patterns in a given language. Realising individual differences (learner-internal sequences) and acknowledging potential deviations from what is in standard language considered as sequences (learner-external sequences) enhances teaching and learning these patterns (Mitchell &Myles 2017).

In this study, we discuss the possibility of identifying formulaic sequences in the texts written by L2 learners of Swedish, English and Finnish with the help of

keystroke logging. With a formulaic sequence, we refer to a sequence of two or more lexical items which the learner acquires through linguistic input as an unanalysed whole (Wray 2002). Previous research on formulaic language has often been corpus-driven, frequency-based and has focused on English as L2 (Jaworska, Krummes & Ensslin 2015). This study aims to explore in greater detail what kind of lexical elements might be produced as entities when writing, considering the different typological nature of the three languages.

The participants were university students (N=20) learning Swedish, English or Finnish in Finland. In data gathering, we used keystroke logging which is a method for recording keyboard activities during typing (Strömqvist & al. 2006). The writing can be replayed in real time, and pausing and revisions studied in detail. In addition, we ran stimulated recall interviews with some participants to understand the events behind their writing process. The results show that learner-internal formulaic sequences can be identified by studying keystroke logging data, which reveal interesting differences between the three languages as well as between learner-internal and learner-external sequences.

We discuss how keystroke logging as a method, and these findings open doors for a new kind of research in the field of formulaic sequences, and in understanding and teaching typologically different languages.