5.Discussion

Bilingualism is often reported to be accompanied by advantages.For example，it has been claimed that bilinguals’ combined L1 and L2 vocabulary exceeds that of monolinguals（Muñoz-Sandoval，Cummins，Alvarado，& Ruef，1998；Oller & Eilers，2002；Oller，Pearson，& Cobo-Lewis，2007）.It has also been claimed that the conceptual knowledge basis built up in the L1 expedites learning of an L2（Cummins，1991；Verhoeven，2007），and that being bilingual means to be advantaged cognitively in such domains as enhanced metalinguistic awareness and the executive control that facilitates L2 learning（Bialystok，2001）.At the same time，however，there is a possibility that being bilingual is not advantageous in all areas of cognitive development.Rather than simply believing that bilinguals have an advantage over monolinguals or vice versa，it is meaningful to identify which cognitive domains are enhanced by bilingualism（Scheele，Leseman，& Mayo，2010）.

The current study has examined within- and across-language relationships between grammatical forms and semantic/communicative functions by focusing on such complex grammatical measures as transitive and intransitive verbs，active and passive voices，durative and completive forms，and referential topic management.If all of the children had narrated the two scenes using these devices in more or less identical ways，we could have claimed that they all operate with one linguistic system.The real story was found to be much more complicated.Some bilingual children did perform in almost identical ways in some domains of both languages，whereas others performed very differently when narrating stories in the two languages.Similarities between the narrations in the two languages may suggest a flexible use of conceptualizations（or schematizations） that is available to fluent bilinguals，depending upon the cognitive processes required by the task at hand.

It is important to note that the interdependence between schematizations and grammatical domains is not evident everywhere.As reported earlier，for instance，eight narrators who chose the progressive in English selected the completive in Japanese.Bilingual children，when narrating in English，may adopt a process-focused point of view while using a result-focused point of view when narrating in Japanese.This may correspond to what Hopper and Thompson（1980） termed the distinction between foregrounding and backgrounding.Narrative consists of two types of structures：temporal structure（foregrounding） and durative/descriptive structure（backgrounding）.More specifically，foregrounding refers to the parts of the narrative that relate a sequence of events with respect to a timeline，thus belonging to the skeletal structure of the narrative，and，in contrast，backgrounding refers to supportive narrative（e.g.，orientation，which presents static descriptions of the scene） that does not itself narrate the main events.If we interpret that the completive corresponds to foregrounding，and the durative corresponds to backgrounding，then we may claim that bilingual narrators tend to emphasize backgrounding when narrating in English but foregrounding when narrating in Japanese.Although these differences are language-specific features，we may be allowed to hypothesize as follows：Because foregrounding constitutes the narrative skeleton（for relating chronological events） and because backgrounding，in a sense，provides additional but elaborated information，the aforementioned eight children's inclusion of the progressive form indicates that their English narration is sophisticated and advanced.

The choice of the passive voice may possibly also be considered an opposite example of language-specific aspects，at least in part.As Minami（2011） reports，Japanese speakers effectively use passive forms and subject-referencing markers in order to relate a clear chronological sequence of events.This cross-linguistic difference may be attributable to various functions of passive structures in Japanese that are not available in English；specifically，whereas only transitive verbs are allowed in passive constructions in English，both transitive and intransitive verbs can be used in Japanese for the construction of the passive voice.Some examples are，“I was fallen by rain（=I got rained on）” and “I was died by my pet dog（=My pet dog died on me）”（Makino & Tsutsui，1986；Maynard，1990）.As long as Japanese narration is concerned，as seen earlier in the use of the passive voice in the deer scene，as children age and advance in school，they come to encode agency appropriately in their narrations in Japanese，their mother tongue.

However，if we take the results of the chasing scene into consideration，the aforementioned assumption does not necessarily hold.Instead，we realize that we may not be able to assert the positive relationship between the children's age（or year in school） and their use of passive forms.Recall that when describing the chasing scene，four narrators chose the passive voice in both languages（see Table 5），and among these four narrators，two were Mike and Misa，first and second graders respectively.Rather，the differences between the active voice and the passive voice are more likely to be scene specific and related to schematization，or，more specifically，perspective-taking，i.e.，the speaker has more than one way to express a certain event by emphasizing one of the characters involved in the event，without altering the logical contents of the event（Bamberg，1997a，1997b；Kuno，1987）.For example，the meanings of the two constructions of the event，“The deer threw the boy into the river” versus “The boy was thrown into the river” in Picture 17，are different not only in terms of what is placed into the subject/actor and direct object positions，but also in terms of what is emphasized or schematized in the event or action.

Thus，while the discrepancies have been emphasized to some extent so far in this discussion section，we should not forget that the two languages are，at the same time，interlocking in the light of schematization.In some cases，furthermore，we should interpret the degree of interlockingness in the light of cross-language transfer as well.Or rather，it may be safe to say that cross-language transfer and schematization are inseparable at least in some cases.For example，whereas 14 narrators chose a transitive verb when narrating the deer scene in both languages，11 children used an intransitive verb（e.g.，ochiru） in Japanese and also（e.g.，fall） in English（see Table 11）.The fact that the boy falling is punctual or completed in both languages seems to suggest a result-focused orientation，which is characteristic of Japanese and appears in English as well.This can be interpreted as a result of cross-language transfer from the L1（Japanese） to the L2（English）.

It is interesting to note that some children failed to include causality either in Japanese or in either of the two languages.The results of the initial screening tests，which included a standardized receptive vocabulary test，did not show that one of the languages appeared to be significantly weaker than the other.But the study has revealed that all children are not equally successful in narrating stories.Narrative competence requires semantic and syntactic skills，i.e.，the skills needed to organize information（which relates to the verbalization of a schema）.In the case of bilinguals，these skills are required in both languages.We may need to assist certain types of narrators，narrators with language-specific encoding patterns and narrators with random and/or inconsistent encoding patterns，in the development of their bilingual fluency.In the case of children who showed successful English-specific encoding patterns，we recommend that their Japanese parents support their children in the development of their L1 skills.