Psycholinguistics James Myers June 3, 2003 Bilingualism OVERVIEW: 1. Early bilingualism 2. Later bilingualism 3. Bilingual language processing ============================================================= 1. Early bilingualism 1.1 Bilingualism (or multilingualism) is the ability to use more than one language. 1.1.1 Bilingualism is very common: There are over 5000 languages spoken in only 200 countries, most of which have only one "official" language: speakers of "unofficial" languages must be bilingual. Smaller countries must also learn to speak the language of their powerful neighbors (especially English, of course). Thus even in so-called "monolingual" countries like Japan there is actually much bilingualism (everyone studies English in school, there are many native speakers of Chinese, Korean, Ainu, etc, and even Japanese dialects can differ a lot, e.g. Okinawan. 1.1.2 Bilingualism is also ancient: In traditional societies (e.g. hunter-gathering societies), groups are small, and dialects easily become distinct languages. Thus travel and trade require bilingualism. 1.1.3 So is bilingualism a weird trick, "boring" compared to the more important phenomenon of first-language acquisition and processing, or is it, instead, exactly what our brains were built to do? 1.2 First, the facts: the course of development in a child acquiring two or more languages simultaneously (summarized from Taylor & Taylor, 1990). 1.2.1 One-word stage: Child uses words from both languages, but rarely uses exact equivalents (e.g. an English-Mandarin bilingual child would tend to use "dog" or ª¯, rather than both). 1.2.2 Beginnings of sentence production: language mixing occurs. (Unless otherwise noted, examples come from Taylor & Taylor, 1990). Lexicon: German-English bilingual: "EIN big cow, from up in HIMMEL" (ein = "a", Himmel = "heaven") Chinese-English bilingual (Jane Tsay's niece): "OPEN the light" Morphology: German-English bilingual: "PHEIFTing" (pfeif = "whistle") Syntax: French-English bilingual: "a house pink" (English words, French syntax) 1.2.3 By four years, bilingual children have developed some metalinguistic awareness and make a clear social distinction between the languages: Swedish-German bilingual (translated here): "Why is it 'mir' and 'mich' in German and only 'mig' in Swedish?" (Oksaar 1981) Lisa has a German mother and an Italian father (Volterra & Taeschner, 1978): Italian father tries to speak to Lisa in German: Lisa: No, non puoi. (No, you can't) Father: Ich auch spreche Deutsch. (I also speak German [note: ungrammatical]) Lisa: No, tu non puoi! (No you can't!) 1.3 Models of bilingual language development (summary after Satterfield, 1999) 1.3.1 The Unitary Language Hypothesis postulates that the bilingual child does not distinguish grammars in the initial stages of language acquisition but rather has a single "fused" system. This seems to be supported by data, but it fails to provide an explanation for how the languages eventually split. Moreover, it ignores the "naturalness" of bilingualism, rather than recognizing it as a built-in ability of the human brain. 1.3.2 The Separate Development Hypothesis claims that from the earliest instances, the child's multiple grammars develop independently of each other. However, evidence for separate languages at the very beginning is unclear, and in any case, how does child know from the very beginning that there is more than one language in the environment? 1.3.3 Satterfield's own model (simulated computationally) is able to treat bilingualism as a natural subcase of monolingual language acquisition. Satterfield assumes that language acquisition develops in accordance with innate Universal Grammar. Thus language acquisition involves hypothesizing possible grammars and figuring out which hypothesis is correct. This hypothesis-testing is done automatically through parameter-setting. Satterfield then assumes that parameter-setting is done through a genetic algorithm. That is, a grammar is like the genotype, with language being like the phenotype. In biological evolution, change occurs when some random variations in genotypes produce more successful phenotypes and come to dominate in the population. Perhaps language acquisition occurs the same way, with the child's hypothesis space being represented as a population of different hypothesis grammars. Some grammars produce languages that are closer to the adult language, and thus eventually come to dominate in the child's mind. Thus in bilingual children, more than one grammar is able to survive, since more than one grammar fits the data. In biology, species in a single area can coexist only if they operate in separate ecological niches (i.e. eat different foods, etc). Similarly, in language acquisition, a child will only maintain separate grammars if they are needed to deal with distinct kinds of language data. Thus at first, the grouping of hypothesis grammars overlap, which simulates the claims of the Unitary Language Hypothesis (although actually at this stage the child has hundreds of overlapping grammars). As more data is input, the child gradually clusters similar hypotheses together into stable subgroups which eventually become the winning grammars of the mature bilingual system. The advantage of this approach is that it describes developmental change from "fused grammar" to "separate grammars", and also treats bilingualism as natural: it uses the same machinery as are presumably used by monolinguals as well. 1.4 People that acquire multiple languages as children seem to be native speakers of more than one language. But are they truly "perfectly balanced bilinguals"? Maybe not. Cutler, Mehler, Norris, & Segui (1992) found that different languages give rise to different strategies in speech perception, even among supposedly "balanced" bilinguals. Task: syllable-detection: Is the given CV (visual) found in the given word (auditory)? BO bow "yes" BO boat "yes" BO beat "no" French monolinguals were faster to respond if the CV matched a real syllable in the word, implying that French speakers break words into syllables during speech perception: BA balance [ba.lans] FASTER BA balcon [bal.kon] SLOWER By contrast, English monolinguals did not show such a syllable bias, which is apparently related to the fact that syllable boundaries are influenced by stress in English: BA balance [ba.l@ns] SAME BA balcony [bal.k@ni] SAME What about "perfectly balanced" French-English bilinguals? They acquired both French and English as children and sounded like perfect native speakers in both languages. However, they were divided into two groups by how they answered this question: "Which language would you miss more if you lost it?" Those that answered "French" behaved like the French monolinguals, and those that answered "English" behaved like the English monolinguals! 2. Later bilingualism 2.1 Languages learned in childhood can all be called L1 (first language). L2 is a language you learn in adulthood (second language). 2.1.1 Unlike L1 learning, L2 learning is not so easy and automatic: WHY? Nativist answer: Your brain has a biological clock that just shuts it down after a certain age. Thus we acquire L1 using innate mechanisms, but after these stop working, L2 must be learned mechanically the way any new skill must be learned. Empiricist answer (e.g. Elman et al., 1996): Your brain gets "full" with the connections built learning L1, so there is no more room for L2. Any "biological clock" effect is not restricted to language, but physical and chemical changes in the brain cause the ability to learn anything complex to decline with age and increasing experience. 2.1.2 How to learn L2: scientific research (e.g. Taylor & Taylor, 1990) has revealed what you already know from experience: Be motivated (helps focus energies on the problems, softens frustration, etc) Practice a lot (forms new habits in lexical access, sentence production and comprehension, motor control, and speech perception). 2.2 Which aspects of L2 are the hardest? 2.2.1 Phonology: New sounds require changes in motor control habits (e.g. unaspirated affricates in Mandarin for English speakers). Categorical perception also forms strong habits that are hard to overcome (e.g. Japanese listeners' difficulties in distinguishing English /r/ vs. /l/; Strange & Dittmann, 1984). 2.2.2 Morphology: Inflectional morphology is hard (e.g. English number agreement and tense for native Chinese speakers, Chinese aspect for native English speakers). Meier (1991) also notes that adult learners of American Sign Language (ASL) never master the inflectional morphology. 2.2.3 Syntax: Easy: basic word order. Meier (1991) found that adult learners of ASL were almost error-free in using basic ASL word order. Hard: syntactic rules that change word order. Singleton and Newport (in press) found that adult learners of ASL have trouble with topicalization: Target: My friend, he thought my second child was deaf. Actual: My friend thought, my second child, he thought he was deaf. 3. Bilingual language processing 3.1 Code-switching: (the following is just copied from the "Language and social cognition" handout) CODE-SWITCHING: on-line switching between two or more grammars during a conversation (vs. "code-mixing", which might involve mere borrowing of words or phrases into the "main" on-line language of the conversation) 3.1.1 In spite of its surprising nature to monolingual speakers, naturally occurring code-switching is actually not at all difficult to process: Chan, Chau and Hoosain (1983): Subjects: Cantonese-English bilinguals in Hong Kong. Materials: passages written in Chinese with occasional switches into English, either naturally occurring passages (from Hong Kong publications) or artificially created passages Results: the natural passages did not slow down reading time, but the artificially created passages did. 3.1.2 Code-switching is a sign of fluency, not of linguistic incompetence: Poplack (1980): Some Spanish-English bilinguals in Los Angeles often code-switch within a sentence, while others mostly code-switch between sentences. This difference correlates with their fluency as bilinguals, as measured by self-report. 3.1.3 Code-switching obeys strict linguistic rules: Poplack (1980) Free morpheme constraint: Code-switches only happen between free morphemes (i.e. between words) *eat-IENDO "eating" (Spanish inflectional suffix) ES POSIBLE QUE TE MOGUEEN "They might mug you." (This is borrowing, not code-switching.) Equivalence constraint: Code-switches only happen at points where the syntactic rules of both languages can be maintained at the same time. I told him that ... pa' que la trajera ligero. (so that he would bring it fast) Thus the function words in a constituent tend to belong to the language which provides the constituent's syntax. 3.2 The bilingual lexicon(s): The words from different languages seem to form partially separate networks: 3.2.1 Associative links are stronger between words of the same language than between words of different languages Kirsner, Smith, Lockhart, King, & Jain (1984): cross-language priming (e.g. French "pomme" and English "apple") much weaker than within-language priming (e.g. "apple"-"apple"). [comment: doesn't there seem to be some serious confounds here?] Cross-language associative links are stronger for cognates (historically related words): Cristoffanini, Kirsner, & Milech (1986); Taylor (1976) These experiments seem to show that links are stronger between words that are similiar not only in meaning, but also in phonological form; being historically related is not relevant. Would the same occur for the following English and Chinese words? »¡ "say" ³Q "by" (as in "I was hit BY him") 3.2.2 Association across the networks is mediated by concepts: Potter, So, Von Eckardt, & Feldman (1984). Subjects: fluent Chinese-English bilinguals, nonfluent English-French bilinguals Tasks: A. picture-naming in L2 B. translating words from L1 to L2 Results: Both groups showed the same pattern: Faster in task A than task B. Implications: translation requires an extra step: L1 into concept, then concept into L2 3.2.3 There are stronger cross-language associative effects (e.g. in free word-association tests or in priming) for concrete words (e.g. "table") than for abstract words (e.g. "freedom"): Kolers (1963), Saegert & Young (1975). Implications: concrete words allow mediation through visual imagery, whereas the meanings of abstract words are not encoded so vividly. Note: This supplies more evidence against semantic feature theory, which would seem to predict the opposite results since abstract words are easier to express with features than concrete words: freedom = quality[not[must]] table = object+flat-surface+...? REFERENCES Chan, M-C., Chau, H. L. H., & Hoosain, R. (1983). Input/output switch in bilingual code switching. Journal of Psycholinguistic Research, 12, 407-416. Cristoffanini, P., Kirsner, K., & Milech, D. (1986). Bilingual lexical representation: the status of Spanish-English cognates. Quarterly Journal of Experimental Psychology, 38A, 367-393. Cutler, A., Mehler, J., Norris, D. G., & Segui, J. (1992). The monolingual nature of speech segmentation by bilinguals. Cognitive Psychology, 24, 381-410. Elman, J. L., Bates, E. A., Johnson, M. H., Karmiloff-Smith, A., Parisi, D., & Plunkett, K. (1996). Rethinking innateness: A connectionist perspective on development. Cambridge, MA: MIT Press. Kirsner, K., Smith, M., Lockhart, R. S., King, M. L., & Jain, M. (1984). The bilingual lexicon: language specific units in an integrated network. Journal of Verbal Learning and Verbal Behavior, 23, 519-539. Kolers, P. A. (1963). Interlingual word associations. Journal of Verbal Learning and Verbal Behavior, 2, 291-300. Meier, Richard P. (1991). Language acquisition by deaf children. American Scientist, 79, 60-70. Oksaar, E. (1981). Linguistic and pragmatic awareness of monolingual and multilingual children. In P. S. Dale & D. Ingram (Eds.) Child language: An international perspective. University Park Press. Poplack, S. (1980). Sometimes I'll start a sentence in Spanish Y TERMINO EN ESPANOL: toward a typology of code-switching. Linguistics, 18, 581-618. Potter, M. C., So, K.-F., Von Eckardt, B., & Feldman, L. B. (1984). Lexical and conceptual representation in beginning and proficient bilinguals. Journal of Verbal Learning and Verbal Behavior, 23, 23-38. Saegert, J., & Young, R. K. (1975). Translation errors for abstract and concrete responses in a bilingual paired-associate task. Bulletin of the Psychonomic Society, 6, 429. Satterfield, T. (1999). Bilingual selection of syntactic knowledge: Extending the Principles and Parameters approach. Kluwer. Singleton, J. L., & Newport, E. L. (in press). When learners surpass their models: The acquisition of American Sign Language from inconsistent input. Cognitive Psychology. Strange, W.S., & Dittmann, S. (1984). Effects of discrimination training on the perception of /r/-/l/ by Japanese adults learning English. Perception & Psychophysics, 36, 131-145. Taylor, I. (1976). Similarity between French and English words: A factor to be considered in bilingual behavior? Journal of Psycholinguistic Research, 15, 85-94. Taylor, I., & Taylor, M. M. (1990). Psycholinguistics: Learning and using language. Prentice Hall: New Jersey. Volterra, V., & Taeschner, T. (1978). The acquisition and development of language by bilingual children. Journal of Child Language, 5, 311-326.