Cross-language studies of lexical retrieval: Tip-of-the-tongue states in first and foreign languages

Abstract

This dissertation investigates "tip-of-the-tongue" states (TOTs) in native speakers of English, Russian, and Spanish, studying foreign languages, and in fluent Spanish-English bilinguals. Study (1) explored retrospective reports of subjects' every-day experiences with TOTs. Study (2) investigated TOTs (fragmentary information, associated words, resolution type) that were recorded in structured diaries over a four-week period. Experimental study (3) examined TOTs elicited through definition and translation primes in Spanish-English bilinguals in the U.S., and speakers of Spanish in Mexico. Studies (1) and (2) found that English, Russian, and Spanish TOTs display similar characteristics, but also differences concerning reported letters, syllable numbers, and associates. Foreign language TOTs also displayed differences compared to first language TOTs (different target word types, more phonologically related associates, 24% interlingual associates, extensive reference use). Bilingual TOTs involved 22% interlingual associates and above-average resolutions through reference use. Most of the TOT targets across all groups were nouns; proper names occurred relatively infrequently. Subjects' access to gender in Russian and Spanish noun TOTs, strong syntactic constraints on word associates, and the similarity of most target-associate pairs in either meaning or form support two-stage models of lexical production: Word meaning and syntax is processed at a first stage, dissociated from a second stage at which sound structure is accessed. Study (3) elicited high TOT rates for targets from the diaries supporting the respresentativeness of the diary data. Bilinguals were found more susceptible to TOTs (32%) compared to the control group (14%). Translation proved to be a useful TOT elicitation technique reducing ambiguity compared to definition primes. A comparison of targets of different cognate status found increased recall for cognates compared to non-cognates but no reduction in TOT elicitation. Concerning TOT causation and development, it is argued that neither the incomplete activation hypothesis nor the blocking hypothesis can completely account for this data corpus. Various TOT types were suggested: incomplete activation (with or without non-blocking or facilitating associates), incomplete activation with late blocking associates, and early blocking. Whereas most TOTs appeared to be the product of incomplete target activation, some TOTs occurred as a consequence of word substitution errors.