2. Methods
2.1 Participants
Two groups of participants were recruited for the purpose of this study: one group consisting of parents and their children who provided the AoA data of the early vocabulary for our study, and another group consisting of college students who provided rating data on imageability of words.
Parents and children: 928 Mandarin-speaking children aged between 1;0 and 2;6 were recruited from Beijing 301 Hospital, Beijing Haidian Hospital, and Tieying Hospital, where they received regular physical check-ups. Each age group had at least 40 children (range, 40~67). Parents (71% are mothers on average) were asked to indicate whether their child spoke the word on the designated questionnaire (the checklist, see below), depending on the age of their children, and to provide basic demographic information of their educational level, occupation, and income (see Table 1 for more details about the children and their family information for each age group). Most families (about 82%) belong to the middle class as assessed by our questionnaire based on annual family income. Over 80% mothers in this study had graduated from colleges or vocational schools and were white-collar workers, such as editors, teachers, accountants, engineers, lawyers, salesmen, or technicians in factories, etc. A trained nurse in the hospital helped to collect the questionnaires. Among the 928 parents, 278 of them had children aged between 1;0 to 1;4 and completed the Infant Checklist, and 650 had children aged between 1;5 and 2;6 and completed the Toddler Checklist (for more information see Hao et al., 2008). A child was not counted in table 1, and child’s data were excluded from our analysis, if the child had any the following conditions: premature birth, exposure to a second language, or insufficient birth information. Parents provided written consent to the data-collection procedure, understanding that they were free to withdraw from the process at any time during the completion of the inventory forms.
Table 1. Child participants’ demographic information for each age group
Notes. a: groups according to children’s age in month; b: proportion of mother interviewee; c: proportion of mothers’ highest education above a college level; d: proportion of sum of father’s and mother’s income more than 4000 CNY per month; e: the number (44) is less than the total number (45) of this group because some data are not available. The same situation also applies elsewhere in the table.
College students: 40 undergraduates from Beijing Language and Culture University participated in the rating study. They were all native speakers of Mandarin Chinese. They were asked to rate words from the Toddler Checklist with respect to the imageability of words (see details below). They received small gifts(e.g. pens) for their participation.
2.2 Materials and procedure
The instrument used in the present study was the Early Vocabulary Inventory for Mandarin Chinese, a checklist that followed the format of the original CDI for English (see Hao et al., 2008, for a complete list of the words from the two checklists, downloadable at <http://blclab.org/early-vocabulary-inventory-formandarin-chinese/>). The construction of the checklist considered problems associated with previous checklists for Mandarin Chinese and reflected languagespecific properties in capturing early vocabulary development. A number of steps were taken to ensure the high validity of the checklist for Chinese children. First, among all the possible Chinese equivalents of the lexical items that appeared in the English CDI, we excluded items such as lawn mower, which are uncommon in Chinese daily life, but included items that are more culturally frequent, such as xiongmao (‘panda’) and xifan (‘porridge’). Second, we also asked nine mothers whose children aged were between 1;5 and 2;6 to revise and refine the checklist based on their experience with their own children (see Hao et al., 2008, for a more detailed description of the construction of the instrument).
The final checklist includes two parts, following the example of the original English CDI. Part 1 is the Infant Checklist, used for infants aged 1;0 to 1;4, which contains 232 words. It consists of the following 14 categories: action words, animals, body parts, clothing, descriptive words, food and drink, furniture and rooms, games and routines, outside things and places to go, people, pronouns, small household items, toys, and vehicles. The Infant Checklist has 160 items (69%) in common with the English CDI Infant Checklist; these items are direct translation equivalents in the two languages. Part 2 is the Toddler Checklist, used for children aged 1;5 to 2;6, which contains 710 words and is thus a more comprehensive list of words than the Infant Checklist. It consists of the following 20 categories: action words, animals, body parts, clothing, connecting words, descriptive words, food and drink, furniture and rooms, games and routines, helping verbs, outside things, people, places to go, pronouns, quantifiers and articles, question words, small household items, toys, vehicles, and words about time. The Toddler Checklist has 407 items (57%) in common with the English CDI Toddler Checklist.
The 710 words from the Early Vocabulary Inventory for Mandarin Chinese were split into two sets of 355 words for student ratings. Each college student rated one set of words on imageability on a 7-point scale. Imageability was rated with regard to how easily each word elicited a mental image, where 1 = least imageable, such as idea and 7 = most imageable, such as apple (see also Barca, Burani & Arduino, 2002; Liu, Shu & Li, 2007).
To identify the role of input on AoA of words, we used word frequency estimates from the SUBTLEX-CH of Cai & Brysbaert (2010), a Chinese word frequency database based on film subtitles, as approximate estimates of children’s language input. The word frequency norms based on the subtitles of movies and TV series were recently developed in several languages including English, Dutch, German, Spanish, Greek, and Chinese. In several studies they were shown to consistently outperform the traditional norms based on written texts such as books, newspapers or internet pages during word recognition tasks, and were thus considered to provide good estimates of daily language exposure (Brysbaert, Buchmeier, Conrad, Jacobs, B?lte & B?hl, 2011; Cai & Brysbaert, 2010; Cuetos, Glez-Nosti, Barbon & Brysbaert, 2011; Keuleers, Brysbaert & New, 2010; New, Brysbaert, Veronis & Pallier, 2007).
2.3 Data analyses
Two steps of data analysis procedures were applied: first, the percentage of each category of vocabulary was reported to describe the trajectories of vocabulary development for children, as has been done in previous studies (Bates et al., 1994; Tardif, 1996). Second, in order to explore which factors could be the predictors of the AoA for a specific word, a set of other analyses were conducted:(1) we calculated the Pearson correlations between imageability, word frequency, word length, grammatical category membership (noun or verb), and the AoA of words; and (2) we ran multiple regression analyses treating AoA as the dependent variable, and the imageability of words, word frequency, word length and word garmmatical category as independent variables.
Because of our focus on nouns and verbs in this study, for the word grammatical category variable we only included the noun-verb contrast. This decision was also based on the consideration that nouns and verbs made up the majority of the early vocabulary, while other lexical categories (e.g. adjectives and closed-class words) were significantly fewer in number. The word length variable here is mainly focused on the difference between monosyllables vs. multisyllables. Chinese words consist of one or more syllables, each of which is represented by a character in written script. Among the 450 words we analyzed in the second part, 136 were monosyllables, 276 disyllables, and only 38 tri-syllables.
When calculating correlations and conducting regression analysis, the averaged value of imageability over raters was calculated for each word and entered for analysis, the word frequency data was log10 transformed, and the variables of word length and word grammatical category were dummy coded (word length: 1 = monosyllabic, 0 = multisyllabic; grammatical category: 1 = noun, 0 = verb).
The final words used in the second part of our data analyses (i.e. correlation analysis and multiple regression analysis) only included 348 common nouns and 102 action verbs (i.e. 450 out of the 710 words). The other 260 words were not included in this part of the analyses because they were: (i) words (64 items; 9.01%) that are not acquired until age 2;6 by the 50% criteria of Goodman et al. (2008), for example, cha 1 zuo 4 ‘plug socket’; (ii) other types of word like social words (57 items; 8.02%), such as yeye ‘grandpa’, descriptive words (69 items; 9.72%), such as kongde ‘empty’, closed-class words (38 items; 5.35%) and others (10 items; 1.4%); and (iii) words (5 common nouns and 17 action verbs; 3.1%) that cannot be found in the SUBTLEX-CH database, for example, gonggongqiche ‘bus’.