I. De la tensión global a la transformación local
I.3. Cuestión social y las nuevas clases al debate reformista
The selective trust paradigm consisted of three types of questions and the questions involved different functions for selective trust. Thus, performance was analysed by each question. As selective trust indicates children’s ability to trust selectively tracking speakers’ past reliability, performance on the total scores was also examined to confirm children’s development of selective trust for each EM and VT task.
The EM task
The Explicit Judgment questions were given in order to assess whether children evaluated a speaker’s inaccuracy in providing labels for the familiar objects. One point was given if the child pointed to, or verbally indicated a speaker who was previously incorrect with –napo (It seems), for the question, “Which one was not very good at naming?” The maximum score was 2. For the Ask questions, when the child regarded the previously accurate speaker with –te (I saw) as a better source of information, he or she received one point. Similarly, for the Endorse questions, when the child labelled the names of the novel objects based on the previously correct speaker, he or she received one point. For both Ask and Endorse trials, the maximum scores were 3. Performance on each Ask and Endorse question is shown in Figure A.11 in the Appendix. The distributions of Explicit Judgment, Ask and Endorse questions of the EM task are shown in Table 5.4. The data are presented as a function of age in order to understand at what age the children become able to judge accuracy and learn selectively. Over the two questions of the Explicit Judgment trials the expected frequencies would be a pattern of 1-2-1 (.25-.50-.25 for the proportion), while a pattern of 1- 3-3-1 (.125-.375-.375-.125 for the proportion) would be applied for both Ask and Endorse questions over the three trials. These patterns were applied to the VT task.
In the present study, performance on the maximum scores (either two out of two or three out of three trials) was reported on each question because it would indicate whether the children of each age successfully judged others’ testimony. First, it appears that judgment on the speaker’s inaccuracy might be relatively easier among three types of the questions. Asterisks in Table 5.4 show binomial probabilities against the number by simply guessing in each question. According to analyses by the binomial tests, the children of older age correctly judged who was inaccurate in providing the labels of the familiar objects, p < .001 (two-tailed and hereafter) for both 5- and 6-year-olds. On the other hand, the younger children did not
distinguish the speakers’ in/accuracy, p = 1.0 for the 3-year-olds and p = .08 for the 4-year- olds. For the proportion, 27.3% of the 3- year-olds, 47.1% of the 4- year-olds, 61.9% of the 5- year-olds and 80% of the 6- year-olds correctly responded on both Explicit Judgment
questions; that is, the children’s precise appraisal of others’ reliability significantly increased with age as also shown above.
Similarly, the younger children did not correctly predict who would provide better information on the Ask questions, p = .82, and p = .31 for the 3- and 4-year olds, respectively. In contrast to the successful judgment of the older children, the 5-year-olds showed random choices, p = .53. On the other hand, the 6-year-olds’ performance was at significance, p = .05. Apart from the 6- year-olds, the children showed no preference in asking for information considering the differences between the speakers. On the Endorse questions, it appears that the children across all age group had difficulties in labelling the novel objects taking account of the past history of reliability of the speakers, p = .30, p = .72, p = .08, p = .24 for the 3-, 4-, 5-, and 6- year-olds, respectively. Although the 6-year-olds did not exercise selective learning on the Endorse questions, it seems that only they showed selective trust considering the performance as a whole including the three types of questions. In order to confirm this possible interpretation, one-sample t-tests on the total score were carried out for each age group. The results confirmed that only 6-year-olds performed at a level of above-chance, t(9) = 3.67, p = .005. The findings seemed to suggest that the 6-year-olds, but not the younger children, were able to judge the speakers’ credibility and learn tracking their past in/accuracy, when information was identified by evidential markers.
Table 5.4
Frequencies on three types of questions on the EM task according to age groups
Explicit Judgment Ask Endorse
Age groups 0 1 2 0 1 2 3 0 1 2 3
3-years (n = 11) 3 5 3 0 3 6 2 0 4 4 3
4-years (n = 17) 2 7 8† 3 4 6 4 3 4 7 3
5-years (n = 21) 3 5 13*** 2 10 5 4 2 6 7 6†
6-years (n= 10) 0 2 8*** 0 2 4 4† 0 4 4 2
Note. EM: Evidential markers. †p < .10. *** p < .001.
Taken together, it is somewhat surprising that the children by the age of six were able to perform selective trust, which was inconsistent with the previous findings from Chapter Four showing 4-year-olds’ and onwards’ preference for the previously accurate speaker. Accordingly, the findings seem to support the alternative hypothesis assuming older preschoolers’ understanding of selectivity.
The VT task
In the VT task, the speakers conveyed information using the mental-state verbs: I know vs. I think. One speaker with I know was always correct, while the other with I think
always provided incorrect information. The scores on the each question of the VT task were given as mentioned above. The children’s performance on each question of the Ask and Endorse trials are shown in Figure A.12 in the Appendix. The children’s responses on the three types of questions across all children in the VT task are shown in Table 5.5. Inspections
of Table 5.5 indicated several distinctive findings from those of evidential markers, according to the responses of the maximum score of each question. It appears that the children’s
relatively early understanding of selective judgment was shown when information was conveyed by the mental-state verbs. First, on the Explicit Judgment, binomial probability tests against the expected frequency of the maximum score (i.e., 2) found that the children of all age groups apart from the 6- year-olds responded correctly on the question of who was being inaccurate naming the objects, p = .07, p < .001, p < .001, and p = .16 for the 3-, 4-, 5-, and 6- year-olds, respectively. Compared to performance on the EM task, the 3-year-old children’s correct responses on both questions increased by 100%; that is, 54.5% of them judged correctly. For other groups, 70.6%, 61.9% and 50% of the 4-, 5-, 6- year-olds, respectively had both questions correct.
Table 5.5
Frequencies on three types of questions on the VT task according to age groups
Explicit Judgment Ask Endorse
Age groups 0 1 2 0 1 2 3 0 1 2 3
3-years (n = 11) 4 1 6† 2 6 1 2 0 3 6 2
4-years (n = 17) 0 5 12*** 1 4 5 7** 0 4 6 7** 5-years (n = 21) 1 7 13*** 1 10 3 7* 4 3 9 5
6-years (n= 10) 1 4 5 1 2 2 5** 1 5 1 3
Secondly, the 4-year-olds chose to ask the previously accurate speaker for new information in the Ask question, p = .007 (for three out of three trials), but the 3-year-olds’ performance did not significantly differ from guessing, p = .82. The 5-, and 6-year-olds also preferentially selected the accurate speaker, p = .02, and p = .009, respectively. Selective learning appeared to be more difficult than predicting who would be a more reliable source. Only 4-year-olds supplied the names of the novel objects by the previously accurate speaker at a level of above chance, p = .006. On the other hand, the children of the other three groups did not differ from chance, p = .82, p = .22, and p = .24 for the 3-, 5-, and 6- year-olds. Even though the 3-year-olds tended to show judgment of others’ in/accuracy, their performance on the subsequent test trials remained at chance levels. Although the 6-year-olds’ performance was at chance, it does not seem to indicate that they were not able to judge in/accuracy. The 6-year-olds’ performance will be reviewed in the Discussion section of this chapter. In
general, the findings seem to indicate that the children by the age of four became aware of the speakers’ trustworthiness and this led to their selective learning when the mental-state verbs were the means of communication. In order to support these findings, further analyses by one-sample t-tests, as done for the EM task, were carried out for each age group. As would be expected, the 4-, and 5-year-olds performed significantly differently from chance levels, t(16) = 5.00, p < .001, and t(20) = 2.36, p = .03, respectively, indicating their understanding of selective trust. On the other hand, the 3- and 6-year-olds did not differ from chance performance, t(10) = .69, p = .51, and t(9) = 1.34, p = .21, respectively.
In summary, the findings seemed to suggest that the children’s access to others’ epistemic states might be influenced by the means of how information was conveyed. It appears that the children’s ability to exercise selective trust emerged by the age of six from the evidential markers and by the age of four from the mental-state verbs. However, as earlier analyses by the repeated measures ANOVA tests showed the differences in the scores of the
questions were not significantly different from each linguistic reference. Hence, it seems that the linguistic differences between the grammatically embedded evidential markers and explicit mental verb terms might be very subtle.
5.4. Discussion
The aim of the present study was to examine whether young children’s ability to show selective trust was affected by verbal input with different linguistic markers. For the differences in linguistic access, evidential markers (i.e., -te (I saw) vs. -napo (It seems/I think) and mental state verbs (i.e., I know vs. I think) were used in order to differentiate two
speakers’ status of knowledge. The findings were consistent with the previous research
showing young children’s preference for a speaker with a past history of being accurate. What was distinctive from those findings is that the children’s judgment on reliability and selective learning is dependent here on how the speakers express their in/accuracy. When the speakers’ in/accuracy was stated by the mental-state verbs the children by the age of four performed selective trust, whereas the children at 6-years tended to understand differential credibility. That is, the findings indicate that they also consider the speaker’s attitudes along with knowledge status. Importantly, it appears that the means by which the attitudes were conveyed within a language system is also critical, along with differential accuracy. In this section, I will discuss performance of selective trust according to each linguistic marker (i.e., belief verbs and evidentials) in order to review how or why children performed selective trust at different ages in the following two sections. I will then discuss comparison between
selective trust tasks by the two types of linguistic expressions. Finally, I will review a methodological issue raised in the previous chapter.