Abstract
The present thesis investigates if recurrent multi-word sequences, which are pervasive in language use, are probable storage and processing units in language users' minds. The material is extracted from the Norwegian oral corpus NoTa-Oslo, and consists of frequent five-word sequences that traditionally have been assumed to be online computations, not represented as units in the mental lexicon.
In line with usage-based theory, I hypothesize that high frequency units of any size are entrenched activation patterns in mind, constituting storage and processing units, even if the units may be products of regular computations.
A psycholinguistic experiment has been conducted to measure the perception of frequent multi-word sequences compared to infrequent multi-word sequences. The results suggest that frequency affects the mental representations, even for units beyond the word. This fact challenges the traditional approach to language patterning, or formulaicity, which does not consider frequency to be a decisive factor in holistic storage of complex units.
A usage-based model of language storage and processing is contrasted with a dual-mechanism model in their ability to predict and explain the storage and processing properties of the recurrent multi-word sequences. Based on the findings, I suggest that the usage-based model is better suited to account for the recurrent use of multi-word sequences.