Frequency effects in the processing of morphologically complex Turkish words

This is an empirical study that examines how morphologically simple and complex words in Turkish are represented in the brains of native speakers. Two experiments are reported that use various "frequency of occurrence" metrics as independent variables. The secondary findings of the study are that (a) frequency is an extremely complex concept, especially in the case of an agglutinating language like Turkish; (b) different frequency measures are highly correlated; (c) frequency distributions are uneven at several levels; (d) the overwhelming majority of grammatically possible forms are never used even in a large corpus; (e) in an agglutinating language like Turkish, morphology has a deep impact even at sub-lexical levels such as the distribution of letter-ngrams; (f) conducting psycholinguistics experiments online rather than in a laboratory environment is a feasible option; (g) letter shape does not have an effect on word recognition accuracy; (h) morphologically complex Turkish words are processed two times more slowly than simple words, suggesting that suffix sequences add a significant workload to the recognition process. The three main findings of the experiments, on the other hand, are that (a) more frequent simple words are processed faster than less frequent simple words, thus replicating a well-established finding in a typologically different language; (b) complex words are probably processed from left to right, and, most importantly, (c) the human brain can use suffix sequences to recognize complex words, thus suggesting that there exist mental representations for frequently occurring suffix sequences, probably in addition to mental representations for individual suffixes.