Frequency does not predict the processing speed of multi-morpheme sequences in Japanese

Abstract This paper reports on a psycholinguistic experiment that examines frequency effects for multi-morpheme sequences in Japanese. While frequency effects for multi-word sequences are fairly well documented, crosslinguistic coverage on such effects remains scarce. Investigations into agglutinative languages can provide insights into whether frequency effects manifest across structural levels (morphological or syntactic). Drawing on the methodology of Arnon and Snider, this study investigates whether highly frequent, compositional multi-morpheme sequences in Japanese are processed faster than their non-frequent counterparts. The results fail to demonstrate significant effects of whole-string frequency on the processing of multi-morpheme sequences. The lack of frequency effects is discussed in light of (i) methodological changes from previous research to adapt the experiment to Japanese and (ii) structural properties of multi-morpheme sequences, which cut across the traditional notion of words.