Prepared by Diana Wang and Brian kay (additions and comments by Frawley)
A. There are different facts that are involved in language acquisition:
1. Language development is species-specific and modular.
2. Linguistic development cannot be traced to something like general intelligence: this is seen in the linguistic ability that is found in even those with disorders who have low IQs (William's Syndrome, Turner's Syndrome)
B. There are learning/maturation trade-offs
1. Input can be varied, impoverished, and ostensibly lack structure: despite this, children still seem to learn; caretaker speech seems to have no correlation to children learning language
2. The input that comes through is positive: there is a correction of truth, but not form
3. In the area of maturation, there seems to be something that must be minimally builtin since: (a) all children go through the same phases of language instances; (b) there are abstract specifications that are "initially hard-wired", such as sonorance and closure; (c) there are regular phases of linguistic development that are tied to critical periods.
C. Acquisition is structure-dependent, not traceable to statistical properties of the input: e.g., there are patterns in phonology and morphology
D. Two problems the study of language acquisition:
1. Children produce less than they comprehend
2. What you know is quite different from how you use it (competence vs. performance)
The tasks facing language learners:
1. Segmentation, or the discovery of units in the speech stream.
a. The units seem obvious to those competent in a language.
b. But the unitization of speech is not a physical phenomenon: a sound spectrogram shows that there are no breaks between words or even sentences.
c. It was previously thought that children break down a speech stream from its smallest units to its largest. That is, they identify individual words in the stream, and then go on to the larger ones. This is known as a bottom-up approach.
d. Now, however, research has shown that children can go top-down. They can identify the larger units in the stream such as clauses or phrases by means of prosody, information about the stresses used in the speech stream. Research by Jusczyk at Hopkins has shown that children can identify their own name in a speech stream at the age of 4.5 months. Possibly, this can help them to discover other units.
2. Interpreting the flux and flow of the world through language
a. If the child is going to use language to describe the world, he/she has to have the same knowledge of the world that we do. This is because our (human) language is fundamentally designed to describe the world that we (humans) sense.
b. Concepts such as the unity of objects are not as obvious as they seem to humans. For example, if someone points at an object on a table and says "umbrella," they may be referring to the object, the table, or a combination of the object and the table, etc. Some evidence for (b):Renee Baillargeon has conducted an experiment that shows that infants have the same intuitions of object permanence that we do. She presents the 4-month olds with an object behind a screen. Then, through use of a hidden trap door, she shows by moving the screen that the object has disappeared. The infants are surprised by this, as we would expect if they believed like we do that objects are permanent. In a control experiment where the object does not disappear, the infants fail to display interest.
But others have had different ideas of how object permanence (the fact that we know an object we have previously sensed still exists) is known by humans: a) Bishop Barkeley thought this was a proof for a God which sees everything at all times. b) Jean Piaget believed that infants had to work out this concept, to search and find that objects they were no longer sensing still existed. Baillargeon's research appears to prove Piaget wrong.
3. Making a mapping between events in the world and elements in the speech you hear:
One problem with understanding that a word equals a thing in the world is the variability of things in the world. For example, the concept signified by the word "run" looks different depending on who does it (and a host of other factors). The concept of colors as we have in our society is difficult for children to pick up. Variable wavelengths are all assigned the signifier "red," for example. 4. Research problems in language acquisition: How do you figure out what a child knows about language before he/she can describe what they know?
Jusczyk used the auditory preference paradigm to figure out that children at 4.5 months can identify their names in the speech stream. The advantage of this paradigm is that the child does not have to respond in any way except with a look. Studies forcing the child to respond in any other way might be biased because of the child's reluctance to respond to the researchers' game.
In this paradigm, a mother and child sit on a chair facing three walls with lights and speakers. In order to keep the mother from giving the child signals as to what to do, she listens to complex music on earphones. There is a green light on the wall directly facing the child. On the walls to his/her left and right, there are red lights and speakers. The green light flashes (to equalize thechild's choices after each trial) and then a name comes from one of the wall speakers. Then a similarly stressed name comes from the other side's speaker. The dependent variable in the experiment is the visual fixation time, how long the child looks at each side. It was found that the child looked longer at the side where his/her name came from the speaker.
A similar paradigm was used for the experiments by Hirsh-Pasek and Golinkoff to determine the child's knowledge of the verb end morpheme /ing/. This morpheme is generally produced reliably at the age of 24 months.
In this method, a child sits in front of two television sets (with a light in the middle to equalize the child's attention after each trial). On the TVs, characters from Sesame street are shown in activities. A voice asks the child "Where is" a certain activity requiring an end morpheme of /ing/ (for present tense).
The children were aged 18-20 months. The research used three endings for the verbs that were voiced to the child. The first was /ing/, the second the adverb ending /ly/, and the third was the made-up /lu/. The time the child looked at the correct side (for the activity being described )with the correct ending was recorded.
If the child were looking for an equal amount of time at all the verb forms, they would show no comprehension of the end morpheme. If they were just looking at the /ing/ and /ly/ trials more, they would be showing a preference for English end morphemes.
The results showed that all of the children looked at the /ing/ side for longer. Girls would watch the side with no dancing if "dancelu" was heard. Presumably, they were wondering if it was a new word, not just an inflected form of "dance." The boys were 50% watching the /lu/ side and 50% watching the non-/lu/ side. After many trials, both sexes started to look at the /ly/ side for longer, as if this was simply a new use of the morpheme that they were not sure of at first.