Every day we all use thousands of words; each word selected, in a fraction of a second, from the tens of thousands stored in our “mental dictionary.” How do we do that? How are the words in that “dictionary” related to one another? For example, when we want to talk about an event in the past do we find the right verb and then add “ED” as a past tense marker? Or do we just “look up” the past tense form and use it?
For decades students of language have debated whether morphologically complex words (i.e. consisting of more than one meaning unit, like WALK+ED) were listed as separate items in the “mental lexicon” or whether only one lexical entry was needed for the root word, with inflectional variants being constructed (or deconstructed) by “rule”. For example, was WALKED stored as a separate lexical item, or was it constructed from its root WALK , with the ED ending added via grammatical rule? While most linguists argued that a morphologically complex word would always be constructed by rule if it could be, psycholinguists tried (unsuccessfully) to chose experimentally between decompositional and wholistic models of the access of morphological complex words.
As is often the case when competing theoretical models reach an impasse, researchers may have been looking at the wrong variables. Linguist Joan Bybee (Morphology, 1983) proposed, based on evidence from language change, that the type of storage and access depends crucially on the relationship between the frequency of a stem and the frequency of all its inflected relatives. The result would be that a high frequency word like PLAYED would be accessed wholistically while a low frequency word like SULKED would be accessed componentially (SULK +ED) because the frequently occurring ED affix would have a stronger representation than either the stem (SULK) or the inflected form (SULKED).
The current psycholinguistic study used acoustic duration as a measure of lexical processing of morphologically complex words in a production task. It found evidence that processing for morphologically complex lexical items componential or wholistic depends crucially on the frequency of the lexical item.