Monolingual and bilingual language processing: A study of Korean, English, and Korean-English speech errors
MetadataShow full item record
This study is a comparison of monolingual speech errors made by Korean and English monolingual speakers, monolingual speech errors made by Korean-English bilingual speakers, and bilingual speech errors made by Korean-English bilingual speakers; it also incorporates a corpus of bilingual errors involving Spanish, French, Ewe, Hebrew, Romanian, Italian, and Japanese. The various data sets were compared along 14 parameters derived from previous research on Germanic speech error data, which make predictions about universal speech production planning mechanisms. Also the study compares data collected from three different methodologies: naturalistic, film-narration, and a word game; the two experiments were designed to put speakers in bilingual mode and thus induce bilingual speech errors. I found many similarities among the data sets: errors in all data sets tended to preserve the syntagmatic structure of the planned utterance (e.g. syllable structure, allophones and allomorphs, number of words, sequence of lexical categories, etc.); lexical errors tended to preserve the original planned meaning (i.e. lexical substitutions and blends most often involve two semantically related words from the same lexical category). These similarities suggest some universals of speech production planning mechanisms. There were also differences, attributed to: (1) differences in language structures; (2) data collection methodologies; (3) the speaker being monolingual or bilingual; and (4) whether the speech errors were monolingual or bilingual. Specifically, I found that bilingual speech errors differ from monolingual in the following ways; (1) the greatest difference was the proportion of errors in the three main types of errors: lexical errors were the most frequent in bilingual speech errors, whereas phonological errors were the most frequent in monolingual speech errors; (2) the majority of bilingual phonological errors were paradigmatic non-contextual errors, whereas most phonological monolingual errors were syntagmatic and contextual; (3) a higher percentage of [+sem] word pairs were found in bilingual lexical errors than in monolingual lexical errors, due to a large number of translation equivalents being involved. I used these findings to propose a speech production planning model which can account for both monolingual and bilingual processing.