Formulaic sequences (FSs), be they multiword or multi-morphemic, are understood to be processed as entities when learning and using languages. Previous research on FSs has largely concentrated on English. This chapter discusses vocabulary production, especially production of FSs, by L2 learners of three typologically different languages: English, Swedish, and Finnish. Finnish as an agglutinative language has to an extent different types of sequences from more analytic/synthetic English and Swedish. The data, which was gathered through a keystroke logging programme, show how the structures in learners’ mother tongues affect the number and nature of vocabulary items produced as clusters, challenging the prevailing idea of the dominance of FSs in language production. The data also give further evidence to there being a difference between learner-internal and learner-external sequences.