Neural Substrates of Language Acquisition

Patricia Kuhl and Maritza Rivera-Gaxiola

Neuroimaging is being used with infants. "The goal on experiments on infants has been to determine whether the initial state and the learning mechanisms are speech specific and species specific."

"The combination of computational and social abilities may be exclusive to humans (Kuhl 2007)."

"The studies suggest that exposure to language in the first year of life begins to set the neural architecture in a way that vaults the infant forward in the acquisition of language."

EEG/ERP is inexpensive, noiseless, "excellent temporal resolution"

At birth infants are capable of distinguishing between all phonemes that are possible in human languages, but with exposure to the native language, infants learn to "tune out" the phonemes that are not distinguished in their native language.

"By the end of the first year, the infant brain is no longer universally prepared for all languages, but instead primed to acquire the language(s) to which the infant brain has been exposed."

So adults truly cannot learn a second language the same way infants learn their first.

In a study (Kuhl et al 2003), a group of infants exposed to a live person speaking Mandarin were able to distinguish Mandarin sounds, while groups exposed to television and audio Mandarin tested the same as groups that were never exposed to Mandarin at all. So social interaction is very important to infant language acquisition. This could be due to the ways humans learned language for centuries--we could be evolutionary accustomed NOT to learn language from computers.

"A sudden increase in vocabulary typically occurs between 18 and 24 months of age--a "vocabulary explosion" (Granger & Brent 2004, Fernald et al. 2006)--but word learning begins much earlier. Infants show recognition of their own name at four and a half months (Mandel et al. 1995)." p. 520

"new words may be encoded in the same neural regions as previously learned words."

""Newborns tested by Imada et ak showed no activation in motor speech areas for any signals, whereas auditory areas responded robustly to all signals, suggesting that perception-action linkages for speech develop by three months of age as infants produce vowel-like sounds."

"Studies across languages showed that by one year of age infants do not accept mispronunciations of common words, words in stressed syllables, or monosyllabic words, indicating that their representations of these words are well-specified by that age." references omitted

Researchers are "strongly interested" in which comes first: phonemes or words, or if learning is bidirectional.

Data on bilingual infants has conflicting results.

Some believe that it may take bilingual infants longer to accumulate a normal vocabulary in their native languages because of decreased input. There have not been many studies on bilingual infants.

Subject and Topic: A New Typology of Language

charles n. li and sandra a. thompson
1975

subject-prominent lanaguges: English
Topic-Prominent Languages: Chinese
Subject-Prominent and Topic-Prominent Langauges: Japanese, Korean
Neither Subject nor Topic Prominent Languages: Tagalog

Mandarin (TP):
"here very hot
'It is hot in here.'

possible this-CLASS war will soon end aspect
'It is possible that this war will soon end.'"

double subject constructions:
Japanese:
"fish-TOPIC red snapper SUBJ delicious
'Fish (topic), red snapper is delicious

Korean:
airplane-top. 747-subj. big-stative
'Airplanes (topic), the 747 is big.'

Mandarin:
that tree leaves big
'That tree (topic) the leaves are big.'"

'He cut-break the rope': encoding and categorizing cutting and breaking events in mandarin

Jidong Chen

"Mandarin categorizes cutting and breaking events on the basis of fine semantic distinctions in the causal action and the caused result."

"3 semantically different types of predicates can be identified: verbs denoting the C&B action subevent, verbs encoding the C&B result subevent, and resultative verb compounds (RVC) that encode both the action and the result subevents."

qie1-duan4
cut.eith.single.blade-be.broken

"The 1st verb (V1) of the RVC, qie1, encodes only the sub-event of the cutting action while the second verb (V@), duan4, encodes the state change of being broken that results from the cutting action."

"I show that Mandarin supports the proposed universal distinction between C&B verbs . . ., but Mandarin C&B verbs differ crucially from the English cut and break in their semantics and argument structure, and the two-way distinction of C&B verbs should be broadened to include a third type of C&B verb, the RVC."

"C&B RVC verbs are semantically compositional. As a whole, they entail a state change, which is indefeasible..."

"...the verb expressing the causal act always precedes the verb expressing the state change . . . no NP can intervene between the constituent verbs..." ungrammatical: qie1 (rope) duan4

"So the state change that is entailed by a monomorphemic verb like cut or break in ENglish is defeasible for Mandarin C&B verbs, as in (5): I qie1 the rope, but rope not duan4 (canceling the action)

"Mandarin C&B verbs encode the causal sub-event and leave lexicalization of the resulting event to an additional verb."

"In contrast, ENglish monomorphemic C&B verbs lexicalize--hence, entail--the resulting event."

"Both English C&B verbs and Mandarin C&B RVCs are telic. But the former are compatible with the progressive aspect, the use of which switches the focus to the process leading up to the state change. In contrast, Mandarin C&B RVCs are incompatible with progressive aspect: they focus only on the endpoint of the event (Tai 1984) and they present the event as a non-decomposable whole (Li and SHirai 2000; Smith 1997)."

"Correspondence analysis revealed 5 distinct clusters of C&B causal events and the Mandarin verns used to describe them:
1. Cutting with scissor(-like) (2-bladed) instrument (e.g. nail clippers, pliers)...thsi distinction is not universal...
2. Cutting with single-blade(-like) instrument...
3. Breaking with hammer(-like) instrument...
4. Pulling on flexible 2-D object (e.g. cloth, paper) with hands or a hand(-like) instrument...'tear, rip'.
5. Bending or pulling on a linear (usually rigid) object (e.g. stick, carrot) with hands or a handlike instrument."

Mandarin C&B action verbs can be distinguished by the instrument and the manner.


"Mandarin does not have an overarching generic verb like cut that can be used across events like slicing, hacking, chopping, trimming, and sawing. Rather, a specific C&B verb must be selected on the basis of the manner or instrumet involved."

'These result verbs can be seen to differentiate C&B events along the following semantic distinctions:
-feature os the affected object (linear objects vs. others).
state or degree of being broken (e.g. sui4 'be in pieces' vs. po4 'be broken')
direction of seperation: cross wise vs. in some other direction (e.g., duan4 'be broken (of long objects broken crosswise').

7b apple qie1 le
The apple cut

"Sentences like 7b resemble the inchoative construction superficially, since the affected object appears in the subject position and the cause is left unexpressed. But such sentences are not real inchoatives."

"Result verbs resemble English breka verbs to some degree: both specify a state or state change but leave the causing subevent open; both can appear in the inchoative construction. But Mandarin result verbs differ crucially from English break verbs in that they cannot be directly used as causatives."