Cross-linguistic Comparison Of Children’s Early Development Of Category Level Words

Categorization is one of the most important abilities possessed by human to know theworld and a fundamental concept in cognitive linguistics. There are two major theoriesconcerned with categorization:“Prototype Theory” and “Decategorization Theory”. Thecategory taxonomy can be divided into three different levels: basic-level category,superordinate level and subordinate level. As for children’s developmental order of categories,there are mainly two different views: one states that basic-level categories develop first inchildren’s category development, followed by subordinate categories, and superordinatecategories acquired at last. While the other states that superordinate categories develop first inchildren’s early categorization. However, these two controversial views can be attributed tothe different methodologies adopted. Research holding that basic-level categories are namedand learned first usually selected subjects older than2years old, and these subjects’production was the focus of their studies, while research considering superordinate categoriesdevelop first mainly focused on subjects’ category comprehension, they chose subjects asyoung as3-to-18months old.The study reported in this thesis aims to investigate children’s early development ofcategory levels from a cross-linguistic approach. The main research questions are that:(1)Which is the dominant category level in children’s early lexicon development in bothMandarin Chinese and French?(2) What is the developmental order of superordinate,basic-level, subordinate categories in children’s early lexicon development?(3) What are thesimilarities or differences in early lexical development between Mandarin-speaking andFrench-speaking children?This study first used corpus data to explore the development of children’s early categoriesin these two languages. The age of the subjects was set from15months to23months duringwhich the human infants already have the capacity to form categorical representations. Datafrom two different corpora: CELA (Mandarin Chinese) and CHIDIES database (French) wereanalyzed, and the number of the subjects was five for each language. Then in order to furtherconfirm the results of the corpora analyses, control experiments, which applied thepsychological software E-Prime to record the reaction time and correct frequency, were alsoconducted. Participants were60in total, and they were classed into two age groups: school-agedchildren and adults. For each age group, there were30subjects, among which15were femaleand15were male subjects. Stimuli were240colored photographs. Among the240pictures,120are for the same group, and the other120were for the different group. As for the same group, there were four pictures in each trial, and all of them belonged to the same category.While in the different group, the forth picture did not belong to the same category with the firstthree. For each category level, there were10groups of pictures, i.e.40pictures for eachcategory level.The results of the corpora analyses showed that (1) children speaking these twolanguages could produce basic-level categories as early as15months, followed bysubordinate categories;(2) as for the superordinate categories, for each language, there was atleast one subject who had no production; and for those who had production, the number wasvery limited;(3) with the increase of the subjects’age, the production of basic-level categoriesgrew obviously, both for Mandarin Chinese and French. As to the control experiments, thetwo different age groups also showed the same basic-level advantage. Taking both thereaction time and correct frequency into consideration, the basic-level category took less timewhile enjoyed higher accuracy. From the cross-linguistic perspective, the above results canreveal that basic-level category is dominant in children’s category development, followed bysubordinate category and superordinate category, which corresponds to Rosch’s prototypetheory.