From Perception to Production: Kuhns NLM and NLM-e - Coursework Example

From Perception to Production: Kuhn’s NLM and NLM-e Inserts His/Her Inserts Inserts Introduction Children are able to learn and speak their native languages rapidly and effortlessly, usually within the first three years. The process with which children learn their native languages follows similar development patterns irrespective of culture. Over the past few decades, many studies as well as arguments have taken place on the subject of human speech development. These arguments began with Chomsky’s (1957) nativist criticism of Skinner’s (1957) article, “verbal behavior”. According to Chomsky (1957), the universal rules that cover phonologies and grammar were genetically specified in infants. As such, the input of languages acted as a trigger for the appropriate innate rules. He held that language development was a biological function just as the development of the body. Skinner (1957) on the other hand, held that language developed as a result of learning. He argued that the development of language in a child was through feedbacks and instituted control of reinforcement. Kuhl (2000a) argues that both opinions made several assumptions. The first assumption that they made was on the biological preparation that infants make in order to learn the language. Secondly, the assumption of the type of language inputs that infants are exposed to. The third assumption made was in the manner in which changes in development occur. In a later article, Kuhn (2000b) argued that both these views failed to encompass the process involved in language development and that infants utilized a new technique of learning in which the input of language is mapped by the brain. Over the years, various studies have tried to offer a clear understanding of this process and several models of language development have been raised. The arguments raised by Chomsky and Skinner led to the development of a new interactionist view that introduced the aspects of development and learning. Development as defined by Kuhn (2000b) denotes those changes based on internal factors or maturation that take place in an organism over time, resulting in the manifestation of information that is defined in the genome. Learning, on the other hand, defines those processes that are dependent on explicit experiences and yield permanent changes. Developmental Changes in Speech While previous theories endeavored to explain the manner with which language develops, explaining the orderly transition witnessed in all infants irrespective of culture is still a challenge. Eimas, Miller and Jusczyk (1987) noted that one of the most important transitions during development occurs in speech perception. Their study found out that at birth infants display the ability to differentiate all the phonetic units utilized by languages in the world. They found out that all infants displayed these collective skills, irrespective of the environment in which they grow up in. Werker and Tees (1984) noted that these universal perceptual skills begin to abate and at the end of the first year, infant are unable to discern foreign language contrast that they once could easily differentiate. Kuhl et al. (2006) notes that the perception of native languages experiences drastic improvement between the ages of 6 and 12. A study carried out In Japan revealed that Japanese infants are able to perceive /r-l/ distinction (important in English but carries the same meaning in Japanese). At 12 months old, Japanese children are unable to perceive this distinction while American infants are more capable of identifying differences between the two. This change is also noticed in speech production. There is a universal evolution in speech development that is witnessed in infants all over the world. This evolution contains five stages (Ferguson, Menn and Stoel-Gammon 1992): (1) cooing which occurs during the first 4 months whereby infants utter sounds that bear a resemblance to vowels; (2) canonical babbling that occurs between 5 to 10 months whereby infants utter constant-vowel strings (“baba-baba”). (3) At 10 to 15 months, the infants enters the first words phase where they utilize a stable phonetic form to describe an object; (4) at 18 to 24 months, infants engage in two word utterances, in which infants are able to associate two words in a meaningful manner; (5) the final phase is meaningful speech which occurs after 15 months, in which infants utilized babbling and meaningful speech to create long intoned utterances. The universal ability of infants to produce utterance that cannot be distinguished soon abates through divergence that reflects influences from the native language. At the end of the first yea, the utterance of infants raised in different cultures start to be distinguishable. During this time, infants also display clear speech production patterns both in phonetics and in prosodic patterns. The transition from perception to production has elicited numerous studies in which Kuhn has proposed two models: The Native language Magnet model (NLM) and the Native Language Magnet Expanded model (NLM-e). Perceptual magnet At the heart of Kuhn’s model is the perceptual magnet effect. According to this, children as well as adults experience alterations in their language perception abilities as a result of exposure to language. The theory behind this is that experiences in the ambient language results in a “mapping” that alters perception. The perceptual magnet effect was tested using phonetic prototypes defines by Rosch (1975) as good instances that are representative of a given category a whole (a dove, as opposed to ostrich in the category bird). While prototypes do not have the same meaning, they are easier to remember and have shorter reaction times after identification. Studies carried out by Kuhn and her associates on phonetic prototypes made to important discoveries (Grieser and Kuhl 1989; Kuhl 1991 and Miller 1994). First, adult subjects of a selected language were proficient at recognizing phonetic prototypes in their native language. The second discovery was that phonetic prototypes acted as perceptual magnets for other sounds in a specific category. These studies revealed that when individuals listen to a prototype of a phonetic category and are requested to relate it with sounds that encompass it. It was noted that the prototype perceptually attracts other elements of the category towards itself, giving rise to the magnet metaphor. In the perpetual magnet effect, elements in the same space as the prototype are perceptually attracted to the prototype and in the process decreasing the distance between the prototype and members belonging to its category (Kuhl 1993). Native Language Magnet model The Native Language Model tries to explain the initial stages of infant language development as well as changes that occur as a result of language experience. The model shows how the development of native language, speech in infants might affect both the perception and production of speech. The main premise behind this model is that once the native-language patterns are coded, interferences in learning of new patterns take place as they do not conform to the mental filters that have already been established. The Native Language Magnet model explains the transition from perception to production in three phases. The first phase defines the initial abilities of infants. At birth, infants have the ability to discern the phonetic differences of all of the world’s languages. This is demonstrated by a hypothetical first format (F1) /second format (F2) coordinate space that is divided into categories. These divisions perceptually divide the vowels of all languages. According to the model, the ability of infants to hear all the important divergences at this phase is not dependent on specific language experiences. The boundaries that develop in phase 1 initially arrange perception in a phonetically significant manner. The predilections noticed at birth are unlikely to be as a result of a “language module”. This idea is supported by the fact that other animals also display this phenomenon (Dooling, Best and Brown 1995). This phenomenon illustrates that the evolution of speech takes advantage of distinction of sounds that are separated by the auditory system. The second phase describes the vowel space of 6 months old infants raised in different language and cultural settings. At this time, infants display the ability to discern all phonetic categories. The distributional of vowels which infants perceive show differences according to different languages. Native Language Magnet holds that infants epitomize this ambient language data, and that these mental representations create magnetic effects specific to that particular language. This means that at 6 months old, the formation of language specific perceptual maps begin. The third phase illustrates how magnet effects change the initial status of speech perception, and shape the handling of stimuli from foreign languages. The magnetic effect causes the reduction of perceptual distortions in those stimuli that are close to the magnetic attractors while those near the boundaries are maximized. This results in the disappearance of some boundaries which initially defined the perpetual space as it is reconfigured to fit in the particular magnet of a given language. The magnetic effects wipe out those boundaries connected to foreign but not native languages. In the third phase, a perceptual space that was once described by perceptual boundaries (auditory cuts) that separate all speech categories has been supplanted by a radically warped space controlled by magnetic effects that fully reconfigure the space. It is in this phase that infants are unable to discern foreign-language divergences that were previously discriminable. It is at this stage that a language-specific listener first arises as a result of mapping of received speech that lead to the alteration of stimulus differences that infants respond to. If an infant is exposed to two different languages, they develop magnetic effects for both the languages. Native Language Magnets theory presents explanations for the speech perception developmental changes witnessed in infants. It holds that a developing magnet attracts sounds that were previously differentiable toward it, causing them to be less differentiable. The magnetic effects appear before changes in the ability of infants to perceive foreign-language contrasts. Kuhn explains that NLM can explain the difficulty encountered by Japanese listeners on the American /l/ and /r/. As the Japanese /r/ is does not conform to the American /r/ and /l/ prototypes, its magnetic effect will attract both the /r/ and /l/, leading native Japanese speakers unable to differentiate between the two sounds. Native Language theory holds that early language experience forms an intricate perceptual network that allows language to be received’. NLM holds that the native language of an individual (whether an adult or an infant), will govern the manner in which other languages are perceived. From Perception to Production (See below) Figure 1: Retrieved from Kuhn (2004) Using NLM, Kuhn (2004) tried to explain how infants progress from perception to speech. This process comprises of several stages that occur during development. The first stage that she identified is computation strategies. Between 6 and 12 months, infants begin to assess the distribution of sound patterns. While adults hear ‘l’ and ‘r’ as either separate (English) or similar (Japanese), those who speak these two languages utter highly differentiable sounds. Japanese infants are exposed to both English ‘l’ and ‘r’ sounds. It has been noted that the manner in which sounds are distributed offer important clues on the phonemic configuration of any given language. Infants are very aware of the distributional frequencies of language prototypes, and react to all cases near a prototype modal value by grouping them thus enhancing category learning. NLM proposes that infants will exhibit perceptual magnet effect for sounds in their native language, as the prototypes that have been collected and grouped will act as magnets for other sounds that belong to the category of the prototype. The second stage is discovering words. After children have collected prototypes and are able to group words in categories, the begin identifying spoken words. It was noted that by the time children reach 18 months, 75% of them are able to identify 150 words and can speak about 50 words. The segmentation of words by infants is also improved by their computational skills. Saffran et al (1996) revealed that infants utilized sequential probabilities in order to identify words. Sequential probability is the chance that ‘ty’ will follow ‘pre’ rather that ‘bay’ coming after ‘ty’. Prosodic cues also help infants detect prospective words and are prominent in natural speech. It was noted that over 90% of multisyllabic English words used in conversations start with a linguistic stress on the initial syllables. As such, infants have the ability to group words from speech that replicate the strong-weak pattern. The third stage is social influences in learning languages. Computational learning holds that infants are able to perceive sounds by being exposed to the correct form of auditory input. I however noted that social interaction was very important in the learning of language. In both speech perception and production, the existence of another human being interacting with the infant has a tremendous impact on language learning. Studies carried out on infants in social isolation have shown that isolation poses severe problems to language learning. Modern theory has proposed that the learning of language in infants is based on the infant’s appreciation of how others, communicate their awareness of visual attention and their drive to imitate. The final stage is native language neural commitment. Dehaene-Lambertz and Gliga (2004) carried out an MRI study on the brain of adults and infants and found out that they share identical active brain regions when they are exposed to normal speech but however differ when they are exposed to backward speech. According to the native language neural commitment theory (NLNC), the learning of language causes the development of dedicated neural networks that encrypt the native-language patterns of speech. According to the theory, this neural commitment to prosodic and statistical divergences in native language enhances the manner in which individuals will utilize these learned patterns in the future when dealing with higher-order computations of the native language. Infants who are good at detecting the patterns in their native language proceed faster towards complex structures of the language in question. Native Language Magnet Theory, Expanded (NLM-e) In 2008, Kuhn and her associates extended the findings made in the Native language Magnet model. Unlike the three step NLM model, the NLM-e model has four steps. In the NLM-e model, Kuhn et al. (2008) tries to better explain the perception-production link in infants. The NLM-e has five general principles, improvements to the previous model. The first principle holds that speech directed to infants and distribution patterns are the main means with which change from the universal perception of phonetic patterns of a language specific perception occurs. Distribution patterns were first explained in the NLM model. The second agent of change is infant directed speech (mothers). According to NLM-e, mothers usually exaggerate phonetic dissimilarities when addressing infants. They stretch the auditory clues of phonetic components, embellishing their differences thus making them easily distinguishable. Englund (2005) held that by enhancing the dissimilarities between phonetic components, their discernible characteristics are easy to learn; hence the language learning process of infants is enhanced. The second principle is that exposure to language creates neural commitments that determines how the infants learns in the future. This principle was first introduced in the first model when Kuhn (2000) introduced the concept of native language neural commitment (NLNC). Under this concept, the manner in which language is initially coded I the brain influences future abilities to understand the phonetic structure of a new language. Native language Neural Commitment describes the manner in which early exposure to language results in physical changes in the neural circuitry as well as tissues reflecting the perceptual and statistical characteristics of the language input. The networks developed in the brain become attuned to the native language. This commitment leads to reinforcement of the ability to identify higher order patterns of speech (words, morphemes) that take advantage of stored phonetic patterns. At the same time, this commitment leads to the decrease of sensitivity to other phonetic structures. Kuhl (2004) thus argues that language development needs the commitment of the neural network to native language patterns. The third principle is that early language is affected at the phonetic level by social interaction. Studies carried out revealed that infants possess computational skills that aid the acquisition of language. These computational skills are, however greatly influenced by social interaction and natural speech (as opposed to recorded or mechanical speech). According to NLM-e, complex language learning may require social interaction as clues from the speaker has the ability to point out important information to the young infant (helps identify important phonetic units). The fourth principle is that the link between perception and production is created developmentally. NLM-e forecasts the formation of strong ties between perception representations created from language experience and vocal imitation. The connection between the two is formed through development of the infant. This model holds that infants create a connection between perception and production as a result of experience and mapping between production and perception as a result of learning. According to the model, infants first go through is sensory learning as a result of language experience after which the development of motor patterns occurs. The vocal play in which infants engage in enables them to associate the auditory products of their own vocalization to the motor patterns that produced them, and thus a learned mapping between the two aspects is created. Infants endeavor to copy the sounds they are exposed to and are directed by the level of match between the sounds stored in memory and those that they produce. The final principle is that the growth of language is predicted by early speech perception. The model proposes a link between early perceptions of native-language phonetic elements and the future development of the language. Kuhn et al. (2008) carried out studies on the relationship between language learning and early phonetic perception. They found out that at six months the perception of native language predicted outcomes of language learning for the next 18months of the infant’s life. The phases of NLM-e are represented in figure 2 below. Figure 2: The NLM-e model Retrieved from Kuhn et al. (2008) Conclusion Kuhn’s NLM and NLM-e both try to explain language learning in infants. Both of these theories are based on perceptual magnets and native language neural commitment. These two models hold that at birth, infants have the ability to identify the phonetic predictors of any language in the world. As they develop, exposure to their native language forms neural networks in the brain that commit to phonetic units of their native language. After this, learning of the native language improves while the ability to identify predictors of the other languages declines. At this point, a link between perception and production occurs both as a result of language learning and mapping between production and perception. Mapping arises from vocal play where infants learn to compare the sounds they produce to those already stored in the neural networks. The models, thus offer us a means through which we can learn the development of language in infants from 6 months to 3 years. References Chomsky, N. 1959. A Review of B. F. Skinners Verbal Behavior. Language, 35, 26-58. Dehaene-Lambertz, G. & Gliga, T. 2004 Common neural basis for phoneme processing in infants and adults. Journal of Cognitive Neuroscience, 16, 1375–1387. Dooling, R. J., Best, C. T. & Brown, S. D. 1995. Discrimination of synthetic full-formant and sinewave /ra-la/ continua by budgerigars (Melopsittacus undulatus) and zebra finches (Taeniopygia guttata). Journal of the Acoustical Society of America, 97, pp.1839–1846 Eimas, P. D., Miller, J. L., & Jusczyk, P. W., 1987. On infant speech perception and the acquisition of language. In S. Harnad, ed. 1987. Categorical perception. New York: Cambridge University Press, pp. 161-195 Englund, K. T. 2005. Voice onset time in infant directed speech over the first six months. First Language, 25, pp.219–234. Ferguson, C., Menn, L. & Stoel-Gammon, C. eds., 1992. Phonological development: models, research, implications. Timonium, MD: York Press. Grieser, D. & Kuhl, P., 1989. Categorization of Speech by Infants: Support for Speech-Sound Prototypes. Developmental Psychology, 25, pp. 577-588 Kuhl, P. K. 1991. Human adults and human infants show a “perceptual magnet effect” for the prototypes of speech categories, monkeys do not. Percept. Psychophys, 50, 93–107. Kuhl, P. K. 1993 Early linguistic experience and phonetic perception: implications for theories of developmental speech perception. J. Phonet, 21, 125–139. Kuhl P. 2000a. Language, mind and brain: Experience alters perception. In Gazzaniga M. (ed) The New Cognitive Neurosciences 2nd edition. Cambridge MA: The MIT Press Kuhl P. 2000b. A new view of language acquisition. Paper presented at the National Academy of Sciences colloquium Irvine CA Kuhl P. (2004) Early language acquisition: Cracking the speech code. Nature Reviews Neuroscience, 5, pp.831-843 Kuhl, P. K., Stevens, E., Hayashi, A., Deguchi, T., Kiritani, S. & Iverson, P. 2006 Infants show a facilitation effect for native language phonetic perception between 6 and 12 months. Developmental Sciences, 9, pp. F13–F21 Kuhl P., B. Conby, S. Coffey-Corina, D. Padden, M. Rivera-Gaxiola & T. Nelson (2008) Phonetic learning as a pathway to language: new data and native language magnet theory expanded (NLM-e). Philosophical Transactions of the Royal Society, 363, pp. 979-1000 Miller, J. L. 1994. On the internal structure of phonetic categories: A progress report. Cognition, 50, pp. 211-285 Rosch, E. 1975. Cognitive reference points. Cognitive Psychology, 7, pp. 532–547 Saffran J., R. Aslin & E. Newport. 1996. Statistical learning by 8-month old infants. Science 274, pp.1926-1928 Skinner, B. F. 1957. Verbal behavior. New York, NY: Appleton-Century-Crofts, Inc. Werker, J. F. & Tees, R. C. 1984a Cross-language speech perception: evidence for perceptual reorganization during the first year of life. Infant Behavior Development, 7, pp.49–63. Read More