4.2 Construction-specific Factors: Quantifier Types and Parallelism
4.2.1 Donkey Sentences: Quantifier Types
As described in Section 4.1.1, mean scores were counted with the coding of the readings elicited by quantifier types for donkey sentences. Mei ‘every’ and youxie ‘some’ were coded according to Kanazawa (1994) with a universal reading and an existential reading, respectively.
For the quantifier bushi meige ‘not every,’ Krifka (1996) and Geurts (2002) argue that it elicits a universal reading, yet examining the adults’ answers to this type of quantifier, we found that it was not such a case in Mandarin Chinese; hence, not following what literature has argued, an existential reading was assigned to sentences with bushi meige ‘not every.’
An overall comparison between quantifier types in all the four groups was examined, and
as illustrated in Figure 4-2, the subjects interpreted donkey sentences with the highest mean
scores on mei ‘every,’ the second highest on bushi meige ‘not every,’ and the lowest on youxie
‘some.’
Figure 4-2. Subjects’ Overall Interpretations of Quantifier Types in Donkey Sentences
The overall between-type differences within the four groups were found (χ2(2)=116.78,
p<.001), indicating the subjects identified the differences of quantifier types and found mei
‘every’ the easiest to interpret. To see more specifically the mean scores of quantifier types in
each group, let’s take a look at Table 4-3, where significant differences between quantifier
types were found (KS: χ2(2)=62.01, p<.001; Grade 2: χ2(2)=52.14, p<.001; Grade 4:
χ
2(2)=41.52, p<.001; Adults: χ2(2)=26.43, p<.001). In other words, each group interpreted thethree quantifiers with existential and universal readings in different degrees.
mei 'every' youxie 'some' bushi meige 'not every'
M 0.93 0.50 0.74
SD 0.14 0.35 0.38
0 0.2 0.4 0.6 0.8 1
Table 4-3. Each Group’s Interpretations of the Three Quantifier Types
received the highest mean score, followed by Grade 2, Grade 4, and KS (Adults: M=0.95 >Grade 2: M=0.94 > Grade 4: M=0.93 > KS: M=0.89). For youxie ‘some,’ only the adults received a high mean score (M=0.78) while the three child groups obtained low mean scores (KS: M=0.24; Grade 2: M=0.42; Grade 4: M=0.57), and for bushi meige ‘not every,’ Grade 2, Grade 4 and the adults all interpreted it with high mean scores (Grade 2: M=0.91; Grade 4:
M=0.91; Adults: M=0.97), yet the mean score of KS was quite low (KS: M=0.17).
However, with a closer look at the means in each quantifier in the adults, in contrast to sentences with mei ‘every’ and bushi meige ‘not every’ (M=0.95, 0.97, respectively), a lower mean score in sentences with youxie ‘some’ was found (M=0.78). As a result, a second look at this type is necessary.
Since youxie ‘some’ is an existential quantifier, supposedly it should elicit an existential reading as what literature has argued (Kanazawa 1994, Krifka 1996, Geurts 2002). For the four
task items designed for youxie ‘some,’ 78-100% of the subjects interpreted three of them with existential readings; nevertheless, Question 5 was answered with an existential reading by roughly only two thirds of the adults. It was this task item that lowered the mean scores of the
youxie ‘some’ type. Question 5 was “Youxie you maomi de nuren hen chongai maomi ‘Some
women who have a cat pamper it,’” and its pictures are shown in Figure 4-3.
Figure 4-3. Question 5 in the DSII for Youxie ‘Some’
Picture A, which displayed an existential reading, was the correct answer to Q5, but only about 63% of the adults chose it correctly. This result might be due to pictorial ambiguity that caused such a low mean score. Picture B, which indicated a universal reading, was chosen where all crosses and hearts in the two pictures could be confusing and misleading as in Figure 4-3. It might be inferred that to the adults, only two people in Picture B could also indicate the idea of youxie ‘some,’ and the cats not hugged could also denote the idea of not being loved. As a result, in order to find out the differences in quantifier types with less flaw, the scores of this item were deleted in the four groups. The revised mean scores of each quantifier type and the
within-group comparisons in each group are presented in Table 4-4.
Table 4-4. A Revision of Each Group’s Interpretations of the Three Quantifier Types
Quantifier
The revised mean scores of the youxie ‘some’ type increased in Grade 2, Grade 4 and the adults (Grade 2: M=0.48, Grade 4: M=0.67, Adults: M=0.88). Moreover, the revised scores for quantifier types were still found significantly different in each group (KS: χ2(2)=61.16, p<.001;
Grade 2: χ2(2)=40.26, p<.001; Grade 4: χ2(2)=20.66, p<.001; Adults: χ2(2)=8.67, p<.05).
Pairwise comparisons between quantifier types are illustrated in Table 4-5.
Table 4-5. The between-type Differences in the Four Groups Quantifier
Grade 4
0.0001672 0.6518 0.0001992Adults
0.05522 0.2016 0.0073Note: The adjustment of p-values for three paired comparisons was .022.
2 Three pairwise comparisons – mei ‘every’ vs. youxie ‘some,’ youxie ‘some’ vs. bushi meige ‘not every,’ and mei ‘every’ vs. bushi meige ‘not every,’ – were needed; therefore, the adjusted p-value was .05/3 = .0167, and rounding it to the nearest hundredth was .02. Hence, if the result was a p-value of .04, it was counted as
The two-sample Wilcoxon tests indicated that KS, Grade 2, and Grade 4 interpreted mei
‘every’ and youxie ‘some’ significantly differently (p<.02). In addition, except KS, where a
significant difference was also found in the comparison between mei ‘every’ and bushi meige
‘not every’ (p<.02), the other three groups showed significant differences between youxie
‘some’ and bushi meige ‘not every’ (p<.02).
Concerning the developmental pattern for each quantifier, between-group comparisons were investigated with the use of the Kruskal-Wallis rank sum test. First, as seen in Table 4-6, it was found that there was no significant difference in sentences with the quantifier mei ‘every’
(χ2(3)=2.72, p=.44) for the subjects, who all interpreted donkey sentences with mei ‘every’ with universal readings.
Table 4-6. The between-group Differences in the Three Quantifier Types
Quantifiermei ‘every’
Note: The adjustment of p-values for six paired comparisons was .01.
insignificant.
For the quantifier type youxie ‘some,’ the overall between-group differences were significant (χ2(3)=48.84, p<.001), where comparisons between KS and Grade 2, KS and Grade 4, KS and the adults, Grade 2 and the adults, and Grade 4 and the adults were found significant as well (p<.01). The only insignificant difference was the comparison between Grade 2 and Grade 4 (p=.04), but the three experimental groups all showed significant distinctions from the adults, indicating that the three child groups did not interpret youxie ‘some’ with an existential reading in an adult-like way.
Last but not least, an overall between-group comparison for bushi meige ‘not every’
revealed a significant difference (χ2(3)=93.62, p<.001). Between-group comparisons showed that Grade 2, Grade 4 and the adults were all significantly different from KS (p<.01), and that the three older groups did not show significant differences (Grade Grade 4: p=.44; Grade 2-Adults: p=.04; Grade 4-2-Adults: p=.18). Hence, only KS did not assign existential readings to