Flashcards in Language development Deck (23):
explain the rules of language
Phonemes are the basic units of sound in a language. Phonemes are then combined to make morphemes, which are the smallest units of meaning in a language. However, language requires more than phonemes and morphemes. It also requires syntax, the rules of grammar that determine how words are ordered within sentences and phrases to form meaningful expressions. And semantics, which are the set of rules governing the meaning of words (nouns, verbs, adjectives). Thus, from a small number of individually meaningless sounds, children can come to generate thousands of meaningful auditory patterns, that combine according to a set of grammatical rules which can then produce an infinite number of messages.
behavioural view children learn language the same way that learn any other kind of behaviour – through operant and classical conditioning and through imitation. From this view, children acquire language as they imitate others’ speech and are reinforced for grammatically correct utterances. Through classical conditioning if a word is reliably paired with the object (e.g. the mother says bottle every time she presents the toddler with the bottle) then the child will begin to understand that the word bottle represented the three dimensional object. However, although this approach explains how children come to understand language, it does not account for how the child learns to produce the sounds of language. Operant conditioning has had more support in explaining speech production. Children are thought to emit a variety of sounds and through the process of reinforcement (e.g. parent praise) will result in the sound being emitted more frequently. Evidence for this can be found in language studies that demonstrate that in early infancy, children across cultures emit similar sounds. However, over the course of the first two years of life, certain sounds gain prominence and others disappear such that the child begins to produce the language of their culture.
the nativist perspective sees language acquisition as natural and almost automatic. According to Chomsky the development of language is very different to the development of any other human behaviour. He regarded language ability as a uniquely human biological predisposition and accordingly, the ability to comprehend and produce language is innate. He referred to studies of the brain as examples of evidence for this theory, which found the location of language interpretation (wernickes) and production (broca’s). However, the discovery of language areas in the brain does not clarify exactly how much of language ability is innate. According to Chomsky our innate language competence consists of a language acquisition device (LAD). This is equipped with the rules of universal grammar, which children simply map to their native language. Once they have learnt enough words, they spontaneously use the rules of universal grammar to understand the meaning of the speech they hear and to combine words into their own grammatically consistent sentences. However, linguistic analysis shows a wide variety in the grammatical rules of different languages and it seems impossible to imagine a universal grammar that could be applied to all of them.
The interactionist perspective of language development emphasised the interaction between innate predispositions and environmental factors in language development. The interactionist approach argues that rather than the similar age of language development being down to innate factors, it is actually because they are all members of the same species who share many common experiences. What is innate, is not any specialised linguistic knowledge but a sophisticated brain that matures slowly and predisposes children to develop similar ideas at about the same age. Thus, the theory that language is acquired from an interaction of a human's innate biological capabilities to acquire language with exposure to language in the environment in which the child is developing.
decasper and fifer:
conducted one of the classic HAS studies and tested infants 24 hours after birth. By sucking on a nonnutritive nipple in different ways, a newborn human could produce either its mother's voice or the voice of another female. Infants learned how to produce the mother's voice and produced it more often than the other voice. This suggested that not only had they learned cause and effect – e.g. they learned the contingency between sucking and producing the mothers voice but also that they could distinguish between mother’s voice and another mother’s voice after less than 12 hours of postnatal experience. The neonate's preference for the maternal voice suggests that the period shortly after birth may be important for initiating infant bonding to the mother and also suggests that prenatal auditory experience may help to shape voice preferences and parent-infant interactions after birth.
decasper and spence:
DeCasper & Spence: In a follow-up study, DeCasper and Spence had 16 pregnant mothers read the Dr. Seuss book The Cat in the Hat to their fetuses twice a day for the last 6.5 weeks of pregnancy. Again they used the nonnutritive nipple to measure the babies' responses. This time, the babies could suck to hear a tape recording of their mothers reading The Cat in the Hat or to hear the mothers reading another children's book, The King, the Mice, and the Cheese, which is also a poem but which has a very different meter. The babies sucked to hear The Cat in the Hat. They also found that if the story was read by their own mother or another mother, it had no effect, the babies would still work to produce the familiar story. The control group however, who had no prenatal stories had no preference between stories but did prefer the voice of their own mother. Again, this is an example of prenatal learning and supports a nativist approach/interactionist approach.
The timing of these preferences, occurring shortly after birth, lead Kisilevsky et al., to develop the hypothesis that fetuses are indeed able to remember and recognize human voices that they are exposed to in utero (the in- utero-speech-experience hypothesis). Support for this hypothesis would not only provide evidence of the early onset of speech perception and memory, but would also have implications for theories of the early development of social and cognitive abilities (e.g., language acquisition and attachment). Fetuses were exposed to tapes of own mother and another mother (95db) reading an adult poem, they found that the fetuses who had the mother reading the poem had an increased heart rate. The results can be seen as supporting those theoretical models of speech perception which assume that the foundation for speech perception and language acquisition is laid before birth – thus a nativist approach. Second, the precocious language-processing abilities observed in newborns and young infants may not be due to a hard- wired speech-processing module in the brain, as assumed by Chomsky (1980) rather, in utero experience may play a critical role. Finally, at least in the domain of speech processing, there may be signs of perception, memory, and attention suggesting some involvement of higher brain structures in the prenatal period.
werker and kuhl:
Werker and Kuhl tested the ability of infants of different ages to discriminate speech sounds. The infants were all from English speaking homes and were tested with speech contrasts that are not used in English but are important in two other languages. The babies learnt that whenever they heard a change in the series of sounds they were listening to they could see an interesting sight by turning their head to one side. Thus, discrimination between the speech sounds was inferred if the infants quickly turned their heads in the correct direction following a sound change. By 6-8 months of age infants readily discriminated between the sounds that they heard. By 10-12 months however, the infants no longer perceived the differences they detected a few months before. Thus, infants are born with the ability to discriminate between speech sounds in any language, but they gradually begin to specialise, retaining their sensitivity to sounds in their native language. Also demonstrates the use-it-or-lose-it phenomenon. However, although infants can learn to categorise based on the distribution of phonemes in their environment, English adults with practice can learn to discriminate Hindu phonemes and Japanese with extensive training can learn to discriminate between r and l. So the loss is not complete but does demonstrate a critical period.
Furthermore, through the process of habituation: Saffran et al., suggested that in order to learn language infants need to learn a lot of things, and one of the things they have to learn is to segregate the speech stream. This a problem faced by all language learners and is the segmentation of fluent speech into words. This process is particularly difficult as word boundaries in fluent speech are marked inconsistently with pauses, suggesting that infants must learn which, if any, acoustic cues correlate with word boundaries in their native language. The boundaries could also be marked by prosody (changes in pitch). However, one important source of information that can in principle, define word boundaries in any language is the statistical information contained in the sequences of sounds. Within a language, the transitional probability from one sound to the next will generally be highest when the two sounds follow eachother within a word whereas transitional properties spanning a word boundary will be relatively low. For example, given the sound sequence pretty-baby, the transitional probability from pre to try is greater than the transitional probability of ty to ba. Saffran et al., was interested in discovering whether 8-month old babies can extract information about word boundaries solely on the basis of sequential statistics and whether or not recognising these statistical regularities could be a mechanism for infant leaning. The infants were familiarized to 2mins of speech stream with 4 repeating nonsense words and used the head turning procedure to judge their responses: bidakupadotigolabubidaku.. the word was filtered so it had no prosodies and no pauses or breaks.
The part words were created by joining the final syllable of a word to the first two syllables of another word. These part-words could only be judged as novel if the infants had learned the words with sufficient specificity and completeness that sequences crossing boundaries were relatively unfamiliar. So, although the syllables followed each other in the sequence already, the only difference was that the words had high statistical probabilities and the part-words had low statistical probabilities. Infants preferred the part words which suggested that the infants succeeded in remembering particular groupings of three-syllable strings containing higher transitional probabilities surrounded by lower transitional probabilities, they therefore used correlations from the environment as a mechanism for learning, showing that they are active learners. These results raise the intriguing possibility that infants possess experience-dependent mechanisms that may be powerful enough to support word segmentation as well as other aspects of language acquisition. Therefore, language development may be best characterised as resulting from innately biased statistical learning mechanisms rather than innate knowledge. If this is the case, then the massive amount of experience gathered in the first postnatal year may play a far greater role in development than has previously been thought.
whole object bias
this bias leads the child to assume that a novel word refers to a whole object and not to its parts or to the material it is made of. Therefore, when children hear the word cat, possibly in the presence of a cat, they are biased to conjecture that it refers to the whole cat and not the cat’s whiskers, paws or tail. Some researchers have suggested that this bias emerges out of infants sensorimotor experiences with objects moving as wholes.
principle of conventionality
suggests that when a child hears a novel word the meaning of the word is also in fact novel, because if a word with a known meaning could have been used the speaker would have used that word. Recently Henderson and Graham tested this account with two-year-old children. Children first watched as an experimental announced her intention to find a mido and appeared to find one among three other novel objects. Then, children were asked by either the same experimenter or a second experimenter to choose the mido from a set of novel objects. Children were successful in both conditions at choosing the target object (61%) indicating that they can fast map and that they can understand the meanings of words across individuals. Additionally, children in both of these groups were significantly more accurate than children in control groups where no novel names were used. Overall, these data suggest that 2-year old children do understand the principle of conventionality and can use it to map a novel name to a novel object.
principles of contrast
The principle of contrast, which works in conjunction with the principle of conventionality, also predicts that children will avoid choosing a familiar object in response to a novel name. If children understand this principle, then they understand that names contrast in meaning, that is, that different forms reflect different meanings (e.g. dog and collie have different meanings).
The mutual exclusivity assumption is similar to the principles of conventionality however it is the assumption that different words refer to different kinds of things. So, for example. Members of the category labeled dog do not overlap with member of the category labeled cow. This assumption also provides a basis for overriding the whole object assumption, which children must do in order to learn terms for parts and properties of objects. So, if a child knows the word cup and his or her mother says “this is the handle”, the child wont take handle to be a synonym for cup but will look for something else to be the referent of the new term.
Markman and Wachtel (1988) demonstrated that 3- and 4-year-old children can use mutual exclusivity to learn terms for parts and for substances. When a novel label was mentioned in the presence of an object with a known label, children rejected the term as a second label for the object and interpreted it instead as a label for a part of the object or its substance
One of these biases is the shape bias this is when young children learning English are biased to attend to the shape of solid rigid objects when learning novel names. The shape bias refers to the robust finding that young children shown a novel solid exemplar object and told a novel name (e.g. “See this? This is a zup!”) are biased to generalize the new name to test objects of the same shape, rather than test objects that match in color or material. A young child shown a novel solid object and told a novel name (e.g. ‘this is a dax’) will most likely say that only other objects that share the same shape as the exemplar can be called by the same name as the exemplar. Thus, young children are said to show a ‘shape bias’ when generalizing novel names for solid objects. There is some debate in the literature concerning the nature and origin of the shape bias. Nevertheless, studies suggest that children’s biased attention to shape in noun generalization tasks emerges over the course of early vocabulary development, and that development of the shape bias aids early noun learning. In particular, children who learn to attend to shape when naming novel objects subsequently show accelerated vocabulary development.
However, with non-solid substances the opposite happens, when the exemplar object is made from a non-solid substance such as hair gel or face cream, older children are more likely to generalize the novel name to test objects made from the same material as the exemplar and so you get a “material bias”.
samuelson and smith
Samuelson and Smith (1999) compared novel noun generalizations with solid and non-solid stimuli in children with a range of vocabulary sizes. Samuelson and Smith found that the children with the fewest nouns in their productive vocabulary did not distinguish between the solid and non-solid stimuli—they were equally likely to generalize by shape and material with both kinds of stimuli. Only children who had at least 150 nouns in their productive vocabulary reliably generalized novel names for solid objects to other solid objects by shape. It appears, then, that the shape-bias emerges only after children have already learned many names for solid objects in shape-based categories.
Samuelson (2002) therefore hypothesised that perhaps, if word learning biases are learned by learning words then we should be able to change these biases by changing the words that children know. To test this she taught 15-21 month olds different types of nouns, there were two groups – a shape group and a material group. The infants came in once a week for 9 weeks and were taught 12 real nouns that they are not typically taught until they are older. The results show that in the shape group they learned a shape bias but they also over extended this shape bias to the material objects! The data for the material condition showed the results to be at chance and so they did not learn their bias. The interesting finding is that at the start of the study the two groups were matched for vocabularies and were both just above the 50 word mark. However, during testing the shape group started to learn a few more words than the material group, after testing however, the shape group had a dramatic increase compared to the material group so it appears that learning the shape bias early facilitates word learning. Thus demonstrating that the shape bias is really a product of word learning and not an innate bias.
Children’s ability to readily map words to referents in the world and retain these mappings over time, with only minimal exposure, is commonly called fast mapping. For example, in Carey’s study, preschool-aged children were able to select an olive green tray when their school teacher gestured to two trays, one blue and one olive green, and asked them to get “the chromium tray, not the blue one, the chromium one.” (13/14 children got the chromium tray). After 1 week, children were provided with a brief reminder of the object- label mapping and then given a comprehension test. The majority of children at the comprehension test retained the association of “chromium” to the color of olive green. In sum, fast mapping behavior includes both the ability to quickly map words to referents in the world and the ability to retain, and build upon, these mappings over time. However, very little is known about the mechanisms underlying children’s retention of word mappings
Over extension errors however, are when a child incorrectly uses a word to describe a wider set of objects or actions than it is intended to. E.g. he may use the word fly to refer to anything small, including specks of dirt, crumbs, small sweets. These errors demonstrate how the child is mapping words onto categories of things.
Under extension errors occur when a child incorrectly uses a word to describe a narrower set of objects or actions than it is meant to e.g. a child may only use the word dog to refer to only Twiggy, the family dog.
Nelson: Functional hypothesis: An alternative functional hypothesis suggests that children first learn to use words that describe important functions or purposes. For example, lamps give light, blankets keep us warm. According to this view, overextension errors result from functional confusions. A dog and a cat both do similar things and serve the same purposes as pets. So a child is likely to confuse them. So, things that get used the same get named the same. Futhermore, questions like “what does it do” and “what is it for” seem more important to young children than “what does it look like”. Therefore, children are more likely to include in their initial word meanings information about the actions and relationships an object engages in.
Semantic Features theory
-Clark: Semantic Feature Theory: Clark proposed on the basis of child observations of child language over extension, that children acquire word meanings ‘bit by bit’, through the gradual addition of semantic features. The sematic feature hypothesis supposes that perceptual properties of objects are mapped directly onto word meaning, bit by bit. Thus the word ‘dog’ might initially be specified only in terms of the feature ‘four legged’ and thus the child will refer to all four legged animals as a ‘dog’. Thus, childrens early word learning and early errors are perceptually based, so things that look the same get named the same.