In lieu of an abstract, here is a brief excerpt of the content:

4฀ Emergent฀Semantic฀Structur e฀an d฀ Language฀Acquisition:฀A฀Dynamic฀ Perspective฀ Ping฀Li฀ The representatio n o f languag e ha s bee n traditionall y considere d a s a construction ou t of basic structural building blocks in the form o f symbol s and rules . Thi s approac h tend s t o loo k a t linguisti c representation s statically. A contrasting approach , i n the spirit o f recent development s i n connectionist network s an d statistica l learning , attempt s t o captur e linguistic representation s dynamically . I t consider s linguisti c representations a s emergent propertie s tha t evolv e ou t o f a continuousl y developing an d adaptin g system . A n eas y wa y t o understan d thi s latte r approach i s t o conside r ho w a hexagona l shap e emerge s fro m th e honeycomb: ever y honeybe e pack s a smal l amoun t o f hone y t o th e honeycomb fro m multipl e angles , but n o honeybe e ha s a grand pla n (o r a genetically determine d rule ) for makin g the hexagonal structur e (Bates , 1984). This chapte r wil l provid e a dynami c perspectiv e fro m connectionis t and statistica l approache s t o examin e languag e acquisition , i n particular , the acquisition of semantic structure. I will first discuss empirical problems, and the n theoretica l perspectives . 80฀PIN G฀LI฀£? Th e Acquisitio n o f Tense-Aspect Morpholog y The las t twent y year s hav e witnesse d a larg e amoun t o f researc h o n th e acquisition o f tense-aspect morphology in connection with the acquisitio n of semanti c categorie s (Slobin , 1985 , 1992) . Thes e studie s hav e investigated childre n learnin g Chines e (Erbaugh , 1978 ; Li, 1990, 1993a) , English (Bloom , Lifter, 6c Hafitz , 1980 ; Brown , 1973 ; Harner , 1981) , French (Bronckar t 6c Sinclair , 1973) , Germa n (Behrens , 1993) , Italia n (Antinucci 6c Miller, 1976) , Japanese (Shirai , 1993), Turkish (Aksu, 1978) , among othe r languages . One robust findin g fro m thes e studies is that young childre n see m t o associate tense-aspec t marker s initiall y wit h a se t o f verb s tha t encod e particular lexica l aspec t properties . Thi s i s typicall y characterize d b y children's early use of past tense (e.g., the English -ed) or perfective aspec t (e.g., the Chinese -le) to comment on an immediately completed event tha t results i n a visibl e chang e o f stat e (Brown , 1973 ; Slobin , 1985) . Fo r example, i n English , Bloo m e t al . (1980 ) foun d tha t th e distributio n o f children's us e o f tens e an d aspec t marker s i s correlate d wit h differen t semantic categories . I n particular , betwee n th e age s o f 1 yea r an d 1 0 months, an d 2 year s an d 4 months , -ed an d irregula r pas t tens e form s occur overwhelmingl y wit h verb s tha t nam e non-durative , completiv e events, i.e. , event s 'wit h a relativel y clea r result ' suc h a s find, fall, an d break. I n contrast t o th e pas t o r perfective markers , progressive marker s (e.g., -ing) occur in children's utterances almos t exclusively with 'process oriented ' actio n verbs that nam e durative, non-completive event s such a s play, ride, an d write. I n Chinese , L i (1990 , 1993a ) conducte d thre e experiments t o tes t th e significanc e o f thi s 'result-process ' distinctio n i n children's acquisitio n o f aspec t marker s betwee n age s 3 an d 6 . Th e comprehension experimen t reveals that children understand th e perfectiv e marker -le better wit h resultativ e verb s an d th e imperfectiv e marke r zai better with process verbs. The production experimen t show s that ther e is a stron g associatio n betwee n perfectiv e aspec t...

Share