Two types of words• Content words:– nous, verbs, adjectifs, adverbs– open class– most are low-frequency words• Function words: – articles, pronouns, conjunctions, auxiliaries, etc…– closed class– tend to be highly frequentAcquisition of function words• Children typically omit function words(telegraphic style)– ex: ‘gros camion’, ‘Papa donne’• Hypothesis: young children do not processfunction words– they are not perceptually salient (monosyllabic, reduced vowels)– they have no observable meaning• Experimental tests of this hypothesisProduction: repetition taskcontent word functor (function stringword/morpheme)English English Pete pushes the dogEnglish Nonsense Pete pusho na dogNonsense English Pete bases the depNonsense Nonsense Pete baso na depSubjects: 2-year-old English-speaking toddlersProduction: repetition task• Hypotheses and predictions– children fail to encode weakly stressed syllables in the input content words (English or nonsense): preservedfunctors (English or nonsense): omitted – children selectively attend to words with familiar referents English content words: preservedall else: omitted– children's functor omissions are due in part to their recognition of these elements as separable morphemes content words (English or nonsense): preservedfunctors: English ones omitted more than nonsense onesResults (low MLU subgroup)content functor504030• more omissions of functors than 20of content words• content words: no ...
Content words: nous, verbs, adjectifs, adverbs open class most are low-frequency words
Function words: articles, pronouns, conjunctions, auxiliaries, etc closed class tend to be highly frequent
•
•
•
Acquisition of function words
Children typically omit function words (telegraphic style) ex: ‘gros camion’, ‘Papa donne’
Hypothesis: young children do not process function words they are not perceptually salient (monosyllabic, reduced vowels) they have no observable meaning
Experimental tests of this hypothesis
Production: repetition task
content word
English
English
Nonsense
Nonsense
functor (function word/morpheme) English
Nonsense
English
Nonsense
string
Pete pushes the dog
Pete pusho nadog
Petebases thedep
Petebaso nadep
Subjects: 2-year-old English-speaking toddlers
•
Production: repetition task
Hypotheses and predictions children fail to encode weakly stressed syllables in the input content words (English or nonsense): preserved functors (English or nonsense): omitted children selectively attend to words with familiar referents English content words: preserved all else: omitted children's functor omissions are due in part to their recognition of these elements as separable morphemes content words (English or nonsense): preserved functors: English ones omitted more than nonsense ones
50
40
30
20
10
0
Results(low MLU subgroup)
content functor
Gerken, Landau & Remez (1990)
• more omissions of functors than of content words •contentwords: no difference between English and nonsense •functors: more omissions of English than of nonsense ones
Functor
Results
omission as a function of knowledge of content words
Gerken, Landau & Remez
(1990)
•
•
•
Comprehension: picture identification task
Picture book with 4 pictures per page, one depicting the target word
Conditions:
Findthebird for me
Find * bird for me
Findwasbird for me
Findgubbird for me
Synthetic speech
90 80 70 60 50 40 30 20 10 0
Results (2-year olds)
Voice 1
Gerken & McIntosh (1993)
Voice 2
the * was gub
•
•
Conclusion
Two-year old children typically do not produce function words
However, they process these words
when they repeat a sentence, they omit real function words more often than nonsense function words
they better understand phrases containing correct function words than phrases containing no, incorrect, or nonsense function words
•
•
Acquisition paradox
Knowing function words can help you learn content words: determiners co-occur with nouns (le toucale,ta jaurime) pronouns co-occur with verbs (elle toucale,tu jaurimes)
Knowing content words can help you learn function words: nouns co-occur with determiners (x biberon) verbs co-occur with pronouns (x mange)
stem?htsicriucalrysotnikaerbstnanfidowHo
•
•
Acquisition paradox
Semantic bootstrapping(Pinker, 1984): infants learn a few nouns that refer to concrete objects and a few verbs that refer to concrete actions based on the syntactic distribution of these items, they can then learn other nouns and verbs (as well as the co-occurring function words)
Phonological bootstrapping(Shi et al., 1998): infants learn to distinguish function words and content words on the basis of phonological (and distributional) differences between the two categories note that they then need to sort function words and content words into two categories (nominal/nouns and verbal/verbs)