Individual Differences in the Tendency to Use Multiword Information in Natural and Artificial Languages

Work in the last decades showed that learning from multiword units is often beneficial for language learning, impacting mastery of arbitrary linguistic relations and predicting efficient language processing. Much of this work has looked at differences between first (L1) and second language (L2) learning, documenting differences in how children and adults approach language learning, with only a few studies looking at individual differences in reliance on multiword units. Here, we ask whether adults differ in their tendency to draw on multiword units when learning a new language, and if so, whether such differences are related to learning outcomes and to language processing. We used an artificial language with grammatical gender to measure participants' tendency to treat article-noun sequences as one unit during language learning, and a multiword recall task to measure their tendency to benefit from multiword units in their native language (L1). Our findings show that individuals differ in their tendency to chunk information into multiword units when learning an artificial language, with most participants falling into one of two distinct groups, showing a steady pattern of preferences to treat article-noun sequences as either one or more than one unit throughout the task. This tendency was found to be numerically, albeit not significantly, related to how much individuals benefit from multiword information in their L1. These findings document a novel dimension of individual differences – the tendency of a learner to rely on multiword units, which may be related to different aspects of language learning and processing.