Children’s vocabulary growth

Cumulative vocabulary of each child and the mean cumulative vocabulary size in each age stage reveal a vocabulary growth, from 290 word types to 882 types. The vocabulary growth before 42 months is a steep distribution while the growth after 42 months is a flat distribution It indicates that the rate of vocabulary growth is higher in stages before 42 months. However, the rate of vocabulary growth slows down in 43-48 months. In terms of mean frequency in each age stage, the mean type frequency reveals that children produce more word types as they grow up, from a mean of 89.3 to 209.6 types. Mean token frequencies are 297.7 in 19-24 months, 636.3 in 25-30 months, 595.3 in 31-36 months, 808.0 in 37-42 months, and 732.6 in 43-48 months. The M-shaped distribution of mean token frequency seems to imply that children do produce more words in 25-30 months and 37-42 months, due to language development rather than

and begin to use them frequently. The mean cumulative type frequency in this study is similar to but higher than that in Hsu’s results (1996). Children in this study had 290 words before 2 years old, 741 words before 3 years old, and 882 words before 4 years old. Children in Hsu’s study had 260 words before 2 years old, 634 words before 3 years old, and 771 words before 4 years old. The two study reach the same result in general.

POS frequency and proportion

In early stage, 19-24 months, nouns and verbs account for approximately 70% in children’s vocabulary. After children’s 30 months, the proportion of nouns and verbs will decrease. The proportion of 8 POS (ADV, CL, QN, ADJ, CONJ, PREP, WH, and MOD) will increase after 24 months. The distribution indicates that children acquire content words first and frequently use them in early ages, whereas they acquire function words later, and increase the use of function words in later stages. In terms of type frequency, children produced averagely more than 30 nouns and 30 verbs in 19-24 months, and in later stages, about 50 to 60 nouns and 65 to 75 verbs. The frequency distributions of nouns, verbs, and adjectives are an M-shaped distribution. The frequency rises in 25-30 months and 37-42 months, whereas it falls in 31-36 months and 43-48 months. On the contrary, the distributions of other parts-of-speech keep rising before 42 months and it drops after 42 months. As for token frequency, children produced more nouns and verbs than other POS in all stages. They produced about a mean of 93 nouns and 97 verbs in 19-24 months, but they produced about more than 100 nouns and verbs in later stages. The frequency distributions of nouns, verbs, adjectives, and modals are also an M-shaped distribution. The declined token frequency in 31-36 months and 43-48 months may imply that children learn to use other methods

to express their ideas, such as using more vocabulary of other lexical categories. This seems to be the case that the number of word tokens keeps increasing in ADV, CL, QN, CONJ, PREP, WH, and others until at the 37-42 months. Moreover, since word classes can be further grouped into open class and closed class, the result of POS in this study has shown different performance in open class and closed class. Open class or content words are produced earlier and have a higher proportion in children’s speech than closed class or function words. The findings in this study support the idea that children’s first words tend to be content words, while grammatical function words appear in children’s speech later (Fernald & Marchman, 2006).

Examining the average number of types a child may produce, two quantitative changes and two qualitative changes were observed. A quantitative change occurred in 25-30 months when children have acquired and produced more different words in all word classes than in 19-24 months old. The mean total type frequency in 31-36 months (168.8) does not differ much from that in 25-30 months (166.4), so there may be nothing changed or may be a qualitative change. However, the type frequency of each POS in 31-36 months reveals that there does something changed. Children acquired and tended to use more words of all most lexical categories, expect for nouns and adjectives.

There seems to be a qualitative change in this stage; children may learn new nouns of different categories and use a different strategy to construct their utterances, such as using more function words in their speech. In 37-42 months, a quantitative change occurred again, which seems to imply that children get familiar with what they learned in 31-36 months and use with a higher frequency. In 43-48 months, children produced fewer different word types in almost all word classes. This change may be caused by a more precise word use, such as using fewer words to express their ideas, or using other

TTR

The type-token ratio of POS in each age stage provides some findings. Nouns, verbs, and POS classified as others have a flat distribution over five ages, which represents that TTR do not change much in different age stages. Previous section reports that children acquire more types of nouns and verbs, and children frequently use them as time goes by. It is also proven that the TTRs of nouns and verbs have a flat trajectory.

The declining TTR of CL and MOD imply that children use some classifiers and modals more frequently in later stages. It supports that children acquire function words later and increase the use of function words in later stages.

Noun bias or Verb bias

Investigating the development of nouns and verbs leads to a question that whether there is a universal bias or not. Is there a nouns bias or verb bias in vocabulary of Mandarin-speaking children? This study tries to measure NVR by including broad and strict nouns and verbs in order to figure out the bias of Mandarin-speaking children in Taiwan. The results show that four types of NVRs have a similar development trend.

NVR1 and NVR4 including broad or strict of nouns and verbs do not have much difference, a weak noun bias in 19-24 months and a verb bias in later stages. NVR2 which includes broad nouns and strict verbs shows a noun bias throughout all stages. On the contrary, NVR3 which includes strict nouns and broad verbs reveals a verb bias throughout all stages.

All types of NVRs can provide a trajectory of children’s development of using nouns and verb. The NVRs are the highest in the early stages. The stage of 31-36 months is the turning point. NVRs decrease sharply before 31-36 months, and the ratios rise in later stages. However, the inclusion of broad or strict nouns and verbs leads to

different conclusion about noun bias or verb bias. Thus, researchers studying noun bias or verb bias should provide clear definitions of nouns and verbs.

In general, noun bias was observed in children’s early vocabulary in this study.

Nouns are first introduced into children’s mental lexicon and children produce more nouns in age before 2 years old. The result seems to support Haryu and colleagues' explanation (2005) that the lack of verb morphology and the argument-dropping property of Mandarin makes verb learning even more difficult than nouns, and more difficult in Mandarin than other languages. On the contrary, children’s vocabulary after 24 months show a verb bias tendency, and this may reflect the linguistic characteristic of Mandarin and also imply that children know arguments can be dropped and begin to drop arguments as in adult’s speech.

Lexical diversity calculated from D values

Lexical diversity was measured by D. A higher value of D represents a higher diversity. The results show that D values of all vocabulary, of nouns, of broad verbs, and of strict verbs all have a similar developmental trend: D value keeps increasing until 37-42 months and it decline in 43-48 months. A statistically significant age effect on the D values, except for D of nouns, is found in 19-24 months and 25-30 months. The developmental trend indicates that children’s speech become more and more diverse as they grow up, which supports the finding in Liu’s study (Liu et al., 2008) that an increasing diversity and complexity was found in children’s lexicon as time goes by. A similar lexical diversity development of Mandarin-speaking children is found in this study (age range = 19-48 months, M = 44.8, SD = 14.1) and in Liu’s study (age range = 13-60 months, M = 44.21, SD = 19.11). The developmental trends observed in this

typically-developing Mandarin-speaking children age before 4 years old. In addition, a higher D value of nouns than strict verbs before 30 months seems to reveal that children’s lexical diversity in early stages comes from the diversity of nouns.

Meanwhile, D values of nouns overlaps D values of strict verb after 31-36 months, indicating a similar lexical diversity.

5.1.2. Children’s vocabulary organization