Psycholinguistics James Myers May 7, 2004 Language development I: Preliminaries and phonology OVERVIEW: 1. Introduction to development 2. Prelinguistic communication 3. Infant speech perception 4. Speech segmentation 5. Child phonology ============================================================= 1. Introduction to development 1.1 Overview of the next three weeks: Week 1: The beginnings of language development; also speech perception and phonology. Week 2: Word learning, morphology, syntax, discourse. Week 3: Theoretical issues: nativism and modularity. 1.2 Remember that linguists focus on representations, while psychologists focus on processes: Linguists say "language acquisition": getting. Psychologists say "language development: changing. Language development does not stop in adulthood (especially in the lexicon and related aspects like morphology). 1.2.1 Some practical "dangers" in studying language development: Behavior vs. knowledge (babies may be smarter than they act or sound) The Clever Hans effect: (i.e. babies may be stupider, too!) Clever Hans was a horse (1904, Germany) who seemed to be able to do math and understand complex sentences. In 1904, Oskar Pfungst studied Clever Hans and found that the horse was merely responding to involuntary movements made by the people watching him. To avoid this problem when studying babies, the adults interacting with the babies in experiments should not know what stimuli the baby is receiving or what response is expected. 1.2.2 Some issues in language development: Does language development/acquisition occur through regular old learning? Or are innate factors crucial? Empiricism: emphasizes role of experience. Nativism: emphasizes innate factors. Is language development primarily dependent on general cognitive development, or does language develop at its own rate? (i.e. the modularity issue). 2. Prelinguistic communication 2.1 Where do babies start? Innate cognitive abilities (more on these in Week 3) Adult language (the "raw data" babies use) 2.2 Child-directed speech ("baby talk", "motherese") Higher and more variable in pitch, more exaggerated in intonational contours: infants do prefer listening to it Culture plays a role in the nature of child-directed speech. In some cultures (e.g. middle-to-upper class White Americans), parents "communicate" even with infants who are obviously too young to do so [Carroll p. 250] In other cultures, communication with infants may be much more restricted. Heath (1983) found that in African American working class community in the South, adults don't direct speech very much AT babies, and adults respond to baby sounds as mere "noise". 2.3 Language as a tool: communicative gestures. 2.3.1 Infants live in a world of communication before they have fully mastered adult language; speech is just one of many ways to communicate (also crying, smiling, pointing, etc). This world of meaning is necessary in order to learn the mechanics of language (especially syntax): Macnamara (1972, p. 1): "Infants learn their language by first determining, independent of language, the meaning which a speaker intends to convey to them, and by then working out the relationship between meaning and language." See also semantic bootstrapping in Week 2. 2.3.2 Two of the (socio-)cognitive requirements for communication: Intention: The producer of a gesture must do it on purpose in order to achieve some goal. Babies seem to produce communicative gestures long before 8 months: e.g. smiles and cries. But at this age, they are not goal-directed: even if crying doesn't get a response, just keep crying! Around 8 months, gestures like pointing and showing, which do seem to be intentional and goal-directed: Infants wait for response Persistence (e.g. keep pointing to hold attention) Development of alternative plans Theory of mind: The producer of a gesture must believe that other people have minds: communication involves changing minds, not merely causing behavior. Infants must learn what it means when others point; they don't learn it consistently until 12-16 months. Chimpanzees never learn this at all! (Povinelli et al. 1997) 2.4 Note regarding modularity: The mechanics of language (especially speech perception) begin to develop LONG before "intention" and the "theory of mind" are developed; thus language development does not require the ability to communicate. Another note: The drive to communicate does not itself cause language to develop. Children with certain language disorders (including "Genie", who will appear in two weeks) have a strong urge to communicate, but their grammars are still disturbed. 3. Infant speech perception 3.1 When do babies begin to learn their language? In the womb! DeCasper and Spence (1986): Mothers-to-be read story aloud during last 6 weeks of pregnancy. When babies born, half of the babies were read the same story, while the other half were read a new story. Measure: Babies sucked on a pacifier connected to a computer that recorded the sucking rate. Results: The babies who heard the same story modified their sucking rate when they heard it again (relative to the other group). How? Sound can filter into the womb, especially prosody: rhythm, loudness, vowel duration, pitch. 3.2 Categorical perception in infants: Eimas, Siqueland, Jusczyk, and Vigorito (1971): Issue: Do infants perceive speech sounds categorically? Subjects: 4-month-old babies acquiring English Stimuli: tokens on the /ba/-/pa/ continuum: VOT (msec)= 0 20 40 60 80 Adult perception:[------ba-------][-------pa------] Babies were sucking on pacifiers connected to a computer. First they were played one syllable over and over. Then the experimenters changed the syllable. Results: 0-20: no change in sucking rate 60-80: no change in sucking rate 20-40: an increase in sucking rate (Same categorical perception as adults!) Lasky, Syrdal-Lasky, and Klein (1975) Issue: Are these perceptual speech categories unlearned? Subjects: 4-6.5-mo.-old babies acquiring Spanish Stimuli: tokens on the /ba/-/pa/ continuum: VOT= -60 -20 20 60 Spanish adults: [-------ba-------][-------pa] English adults: [----------ba-------------][--pha--] Thai adults: [--bba--][-------ba-------][--pha--] Mandarin: [---pa--][--pha--] ^ ^ Note: Spanish is unusual in dividing up the VOT continuum between -20 and 20 msec. Most languages divide it up at one or both of the points indicated by "^". Results: 20/60: change in sucking rate -60/-20: change -20/20: NO change Interpretation: young infants begin with innate phonetic categories. However, recall that chinchillas also perceive voicing categorically; thus phonetic categories may be innate, but not specific to language processing (i.e. not modular). 3.3 The effect of language experience on speech perception: Werker and Tees (1984): 6-8 month-old American infants can discriminate between many phonemic contrasts that are not made in English (e.g. retroflex sounds in Hindi) 8-10 months: decline in discrimination ability 10-12 months: apparent lost of this ability. This is not a loss of a discrimination ability but rather a focus in attention to just the most relevant cues: Sounds that are quite different from all of the adult's categories are still discriminated quite well (e.g. Zulu clicks; Best, McRoberts, and Sithole 1988). Japanese adults can learn to discriminate English /r/ and /l/ through training. 4. Speech segmentation (Segmentation: Separating out the words from fluent speech.) 4.1 Without finding the words, children cannot learn morphology, syntax, semantics, even phonology. But segmentation is very hard: even computers cannot do it reliably. What clues might infants be using? 4.1.1 Familiarity with words learned in isolation? Probably not, since even familiar words don't sound exactly the same in isolation as in fluent speech (think of the Mandarin third-tone rule, for example!) 4.1.2 Prosody: Possibly, since we know that infants learn about prosody very early (even in the womb), and also that prosody provides useful information about syntactic constituents: After the boy sneezed, the doctor prescribed a course of injections. Most words in English start with stressed syllables, so maybe kids start by assuming that every stressed syllable is the beginning of a new word: MAYbe KIDsa SSUMethat SENtences LOOK LIKE THIS 4.1.3 Phonemic regularities: Certain phonemes RARELY occur next to each other within a word; other phonemes OFTEN do: staMPing *staNPing teN Pigs *teNPigs Thus the sequence /np/ implies a word boundary. In Mandarin, two adjacent Tone 3 syllables implies a word boundary! 4.1.4 Repeated phoneme sequences: If a sequence keeps reappearing in the input the a child, maybe they'll eventually guess that it's a word. Brent (1999) describes a computational model founded on this approach. There's an on-line demo of this at lsrg.cs.wustl.edu. 4.2 An experiment on infant segmentation: Myers, Jusczyk, Kemler Nelson, Charles-Luce, Woodward and Hirsh-Pasek (1996): Issues: When do infants begin to segment words, and do they use prosodic cues to do it? Task: Head-Turn Preference Procedure: There are two loudspeakers next to infant, one on the left and one on the right Two types of stimuli: A: adults consider more "natural" B: adults consider less "natural" Measure: duration of head turns towards speakers playing A or B. How to interpret the results: If infants look longer at A, then they "prefer" A. Stimuli: Stories were told in "motherese", edited to insert pauses either between words ("coincident" version) or within words ("noncoincident" version) Prediction: If infants recognize word units, then they will prefer the coincident version, just as adults do. Results and interpretation: Words are apparently not segmented until 11 months. This is true even for unfamiliar words, which shows that infants are using a general strategy based purely on sound. They are NOT, however, using stress (i.e. it didn't matter how the words are stressed; the kids still responded to the real word boundaries). Thus perhaps they are using phonemic regularities...? 4.3 However, Jusczyk later conducted many other studies with other methods, finding that stress also can play a role in speech perception by infants as well. Actually, they seem to use many cues, with perhaps stress cues being used earlier than cues using segments (Jusczyk 1999). For example, Jusczyk, Houston, & Newsome (1999) found that 7.5 month old infants tended to assume a stressed syllable was a word onset, since they perceived "guiTAR is" as if it contained the "word" "TARis". Thus the lack of evidence for stress cues in Myers et al. (1996) might be a problem of the particular method used. 5. Child phonology The above changes in speech perception are paralleled in speech production: 5.1 By end of 2nd month, a lot of "cooing" (vowel-like sounds) 5.1.1 By 6-7 months: Babbling: syllable-like units produced (usually CV's) At first, simply reduplicated: bababababa... By 11-12 months: variegated babbling (e.g. for an American baby: bigodabu...) Note: babbling is NOT communicative; in fact, infants babble MORE when adults are NOT around. Thus it looks more like "practice"! So babbling is likely to be governed by innate linguistic factors, rather than a general "drive to communicate" 5.1.2 By 11-12 months, children's production very clearly conforms to the parent's phonemic system. Note 1: this is the same age when infants "lose" the ability to discriminate non-native phonemic constrasts. Note 2: Lexical tone is easy for infants, much easier than segmental phonology (voicing, etc). 5.1.3 Around 12 months infants produce first words: Note: this is AFTER infants learn to segment words out of adults' fluent speech (perception is easier than production) 5.1.4 Around 16-18 months there is a sudden explosion in vocabulary, with some estimates saying infants learn 8 words a day up through 6 years (14,000 words) Possible explanation (e.g. Jaeger 1992): The vocabulary explosion is assisted by the child's "discovery" of Duality of Patterning, which aids in memory storage and access...? 5.2 Children often use their own private phonological rules (see e.g. Macken 1995). Here's how a child acquiring Mandarin sang a famous children's song (from Prof. Tsay's project). Can you guess the song, and figure out what the rules are? h = aspiration, N = velar nasal, ~ = vowel nasalization, @ = schwa, I = high central vowel, x = voiceless velar fricative ka khiaN ka khiaN Ni~ k@ pi kI k@ m@~ Na~ mo~ khaN ma~ ma~ khuo pi kI khaN khai xI phiau Nia~ This may suggest that they are developing their own grammar, not merely imitating adults. But it could also be due partly (or mostly) to the development of articulation, i.e. merely motor control. This issue is still quite controversial! REFERENCES Best, C. T., McRoberts, G. W., & Sithole, N. M. (1988). Examination of perceptional reorganization for nonnative speech contrasts: Zulu click discrimination by English-speaking adults and infants. Journal of Experimental Psychology: Human Perception and Performance, 14, 345-360. Brent, M. R. (1999). Speech segmentation and word discovery: A computational perspective. Trends in Cognitive Science, 3, 294-301. DeCasper, A. J., & Spence, M. J. (1986). Prenatal maternal speech influences newborns' perception of speech sounds. Infant Behavior and Development, 9, 133-150. Eimas, P. D., Siqueland, E. R., Jusczyk, P., & Vigorito, J. (1971). Speech perception in infants. Science, 171, 303-306. Heath, S. B. (1983). Ways with words: Language, life, and work in communities and classrooms. Cambridge University Press. Jaeger, J. J. (1992). Phonetic features in young children's slips of the tongue. Language and Speech, 35, 189-205. Jusczyk, P. W. (1999) How infants begin to extract words from speech. Trends in Cognitive Science, 3, 323-328. Jusczyk, P. W., Houston, D. M., & Newsome M. (1999). The beginnings of word segmentation in English-learning infants. Cognitive Psychology, 39 (3/4), 159-207. Lasky, R. E., Syrdal-Lasky, A., & Klein, R. E. (1975). VOT discrimination by four to six and a half month old infants from Spanish environments. Journal of Experimental Child Psychology, 20, 215-225. Macken, M. A. (1995). Phonological acquisition. In J. A. Goldsmith (Ed.) The Handbook of phonological theory (pp. 671-696). Blackwell. Macnamara, J. (1972). Cognitive basis of language learning in infants. Psychological Review, 79, 1-13. Myers, J., Jusczyk, P. W., Kemler Nelson, D. G., Charles-Luce, J., Woodward, A. L., & Hirsh-Pasek, K. (1996). Infants' sensitivity to word boundaries in fluent speech. Journal of Child Language, 23, 1-30. Povinelli, D. J., Reaux, J. E., Bierschwale, D. T., Allain, A. D., & Simon, B.B. (1997). Exploitation of pointing as a referential gesture in young children, but not adolescent chimpanzees. Cognitive Development, 12(4), 327-365. Werker, J. F., & Tees, R. C. (1984). Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, 7, 49-63.