Abstract
Aims and objectives/purpose/research questions:
The purpose of this paper is to clarify the grammatical constraints on discourse particles in Chinese–Japanese intra-sentential code switching in light of the general framework of the Matrix Language Frame (MLF) model augmented by the 4-M model.
Design/methodology/approach:
This study retrieves data collected for three years from three Chinese–Japanese bilingual children aged between 2;1 and 5;0.
Data and analysis:
The database consists of nearly 300 hours of spontaneous conversations that are audio-recorded from the families of the three bilingual children, as well as diary entries. It shows that a large number of code switching utterances involve discourse particles.
Findings/conclusions:
Qualitative analyses of the data indicate that discourse particles are generally constrained by the MLF, yet they do not fit into any category of the 4-M model. Morphologically bound, discourse particles represent the information structure of a sentence (as in the Japanese topic marker -wa) or encode constraints on the inferential processes (as in the Japanese complementizer -kara) rather than truth-conditional information. They manifest some idiosyncrasy at the interface of syntax and pragmatics, and set up the MLF at a discourse level. Thus, the MLF model is extended from a merely syntactic level to the syntax–discourse interface.
Originality:
The present work has contributed empirical evidence from a hitherto undocumented language pair of Chinese and Japanese, and made theoretical explorations on the linguistic constraints of discourse particles.
Significance/implications:
On one hand, it is work that provides support for the robust nature of universality of the MLF constraints on code switching. On the other hand, discourse particles exhibit typological features that need further theoretical exploration in order to make a more comprehensive account for the grammatical constraints on Chinese–Japanese code switching.
Keywords
Introduction
The alternate use of two languages within a single utterance, that is, intra-sentential code switching (hereafter CS), has captured much attention from linguists in the past several decades. Morphosyntactic descriptions on CS generally indicate that, despite the seemingly random behavior of CS utterances, there exist some underlying constraints that restrict the switching point and switchability of different syntactic categories involved in CS construction (Muysken, 1995).
However, little agreement has been reached on the characterization and formalization of such CS constraints. In particular, whether discourse particles are subject to any constraints or not remains a problem.
The present study attempts to investigate the grammatical constraints on CS of Chinese (Mandarin) and Japanese, a language pair that has barely been investigated in this field. The focus will be discourse particles in Chinese–Japanese CS utterances, in order to clarify the constraints on this category.
Discourse particles versus discourse markers
Generally speaking, most previous studies adopt the term of ‘discourse markers’ to refer to conjunctions, connectives, interjections, linkers, fillers or tags, etc., such as well, you know, and so on, yeah, so, I mean… (Brody, 1987; Goss & Salmons, 2000; Hlavac, 2006; Lantto, 2015; Maschler 1994, 2000; Matras, 2000, 2011; Torres, 2002). CS involving these discourse markers is regarded as emblematic CS, which is more like established borrowing. Structurally, they are extra-clausal and independent, and thus are subject to minimal syntactic restrictions.
Discourse particles in the present study, however, are not equal to the discourse markers in the previous studies. Referring to Schourup (1999) and Schiffrin (2001), discourse markers are syntactically optional expressions, which do not affect the truth-conditional content of utterances, but indicate the relationships between the utterance and the context. Discourse particles, on the other hand, are bound morphemes syntactically, sometimes phonologically, dependent on their adjacent words or phrases. Both Chinese and Japanese have a number of particles that are morphologically bound suffixes. They are often obligatory, affect the truth-conditional content of utterance and are mainly discourse oriented. Some are attached to a phrase and others to a clause. For instance, in Japanese, the topic marker -wa and an interjective particle -ne, which solicits involvement from the addressee (Iwasaki, 2013; Onodera, 2004), appear after a noun phrase in (1), and -kara ‘because’ in (2) functions as a complementizer.
(1) a. Kono hon this book- ‘This book is interesting.’ b. Koko-ga- here- ‘It hurts here.’ (2) Tukareta- tired-because home-at stay ‘Since I’m tired, I’ll stay at home.’
The properties of discourse particles in the present study are different from those of discourse markers in the previous studies. If discourse markers are totally in the realm of discourse, discourse particles in this study fall in the interface of syntax–discourse.
Discourse particles and constraints on code switching
In this paper we will examine the behavior of discourse particles in CS, mainly in light of the general framework of the Matrix Language Frame (MLF) model augmented by the 4-M model (Myers-Scotton, 1993, 2002).
The MLF model regulates that the surface word order of any bilingual CP (Projection of Complementizer) must be consistent with the Matrix Language (ML), and certain functional morphemes must come from the ML. On the other hand, Embedded Language (EL) is assumed to merely insert its lexical elements into this ML frame. Sometimes, the inserted part is an EL island in order to achieve sufficient congruence between the two languages involved. Hence, the distinction between lexical morphemes and functional morphemes is crucial. This distinction is elaborated more in the 4-M model (Myers-Scotton, 2002; Myers-Scotton & Jake, 2000, 2001, 2009). The model classifies morphemes into four subcategories: content morpheme; early system morpheme; bridge late system morpheme; and outsider late system morpheme. Content morphemes are morphemes that assign/receive thematic roles, such as nouns and verbs. Early system morphemes (e.g. determiners, plural -s) are the ones that are dominated by the maximal projection of the content morpheme that select them. Late system morphemes build syntactic structure rather than semantic structure. Bridge late system morphemes, such as possessive of, -’s, integrate content morphemes into a larger constituent. Included among outsider late system morphemes are agreement markers, clitics, case markers, etc., that depend on grammatical information outside of their own maximal projection, and only ‘outsiders’ are constrained by the MLF. The 4-M model makes a refinement for the MLF model in accounting for CS grammatical constraints. However, as we will see in the section of results and analyses, it is difficult to identify the equivalent category for discourse particles in the 4-M model. Thus, it remains vague whether discourse particles are constrained by the MLF model or not.
Besides the notion of ‘insertion’ demonstrated in the MLF model, the concept of ‘alternation’ has been proposed by Muysken (2000) in his typological approach to CS. Alternation refers to ‘switching at the clause periphery, syntactically unintegrated discourse markers, tag or emblematic switching, flagging and self-repair’ (Muysken, 2000, p. 231). An alternational type of CS is characterizable in terms of paratactic adjunction, and no constraints are correlated with this type of CS. However, the discourse markers, or other ‘functors’ defined by Muysken (2000), seem not to cover the notion of discourse particles exactly.
Generative approaches to CS, on the other hand, argue for no specific constraints on CS grammar. Thus, Chan (2008) claims that the existing syntactic apparatus that governs monolingual syntax also explains head-complement order, and that functional heads (D, I, C) always determine the order of their code-switched complements. In CS between the complementizer and its clausal complement, ‘the language of the complementizer determines the surface position of the clausal/IP complement, no matter which language that clausal/IP complement comes from’ (Chan, 2008, p. 800). This framework may account for the order within the maximal projection of a functional head (F), but it does not predict anything on the FP-internal or FP-external CS. Does F cause or impede CS? Or is it neutral in CS?
Therefore, more investigations based on various CS data are needed to clarify what role discourse particles play in CS, and how they are constrained.
Research method
Subjects
The present study retrieves CS data from three Chinese–Japanese bilingual children. The focal participant of the present study is a Chinese child, XIN, who was born in Beijing, and went to Japan with her Chinese parents at the age of 0;11. At home, she was addressed by her parents in Mandarin Chinese. 2 Meanwhile, she attended a Japanese nursery school from the age of 1;4, and spent approximately 40 hours per week in a monolingual Japanese environment. Therefore, from the earlier stage of her language acquisition, the infant had been exposed to the bilingual environment, which led her to the regular engagement of CS utterances.
A second subject, YOU, born in Japan in a Chinese family, shares a similar language and family background with XIN so that the two infants exhibited high frequencies of CS, the pattern of which shared many similarities. A third subject, TAN, went to Japan at nearly 3 years old with his Chinese parents, and was basically able to separate the two languages, with CS occurring occasionally.
Different from the ‘one parent–one language’ type of childhood bilingualism, which has been investigated more broadly, the present three cases belong to another type, that is, ‘non-dominant home language without community support’, as defined by Romaine (1989, p. 167). It is also known as the ‘one environment–one language’ type of bilingualism.
Data collection and transcription
The spontaneous conversations that took place mainly between the three children and their mothers at their homes were audio-recorded. The time and duration of the recordings were not fixed, in order to get as many natural utterances as possible. The situations of the recordings were interactions consisting generally of activities such as reading, playing with toys and conversing among family members during meals. For this reason, Chinese served as the dominant language input in most of the recordings.
For the focal child XIN, a total of 220 hours of data were accumulated for a time span of 24 months from XIN’s age of 2;1 to 4;0. Of all the recordings, a total of 24:19:50 hours (i.e. 1 hour per month on average) were selected, 3 and transcribed in accordance with the CHAT system in CHILDES (Child Language Data Exchange System; MacWhinney, 2000). The following quantitative study is mainly based on the data of this focal subject, because the data of XIN have been collected in a more regular and coherent way compared with the other two children.
For the other two subjects, YOU was recorded for nearly 50 hours from age 2;9 to 4;0, and TAN 15 hours from 3;9 to 5;0. Some segments of their recordings, especially CS utterances with relevant contexts, were also transcribed for complementary empirical support.
In addition to the audio recordings, diaries were kept by the researcher as a secondary source of data, in case any new types of CS utterances were produced by the subject outside the recording periods. Relevant contextual information has been provided as far as possible in the diary entries. To limit the inevitable selection bias, these diary data will only contribute to the counting of types for qualitative claims; while tokens for quantitative analyses come from the recordings only. The combination of audio recordings and diary data yields a rich database that provides a solid foundation for the present research.
The unit of analysis
This study adopts a bilingual CP as the unit of analysis. As defined by Myers-Scotton (2002, p. 54), a bilingual CP is a CS utterance with ‘the syntactic structure expressing the predicate-argument structure of a clause, plus any additional structures needed to encode discourse-relevant structure and the logical form of that clause’. Thus, each bilingual CP involving the switching of a discourse particle will be the unit of analysis.
Research questions
Through identifying the ML for the bilingual CP involving discourse particles in the current CS data, this study attempts to address the following questions.
What are the frequencies/occurrences of different discourse particles involved in Chinese–Japanese bilingual CP?
How should the discourse particles be categorized under the 4-M model?
Are they constrained by the MLF or not?
The hypothesis here is that discourse particles do not fit into any categories of the 4-M model. However, they partake in building the MLF at the discourse level, and hence are constrained by the MLF.
Results and analyses
Quantitative results
For the focal subject XIN, during the 24 months of the recorded period, this study identified a total of 2434 CS utterances from 8195 utterances within the total 24 hours of the transcribed data. The average ratio of CS occurrence was 29.7%, which was rather high compared with some other studies concerning CS (de Houwer, 1990; Taeschner, 1983, etc.). Figure 1 further traces the tendency of CS occurrence in the subject within the 24 months’ time span of examination.
Percentage and tendency of code switching utterance with age.
Discourse particles in Chinese–Japanese CS
Among the large quantity of CS utterances, discourse particles are found to occur frequently, especially Japanese discourse particles. This study has only examined those that occurred at relatively higher frequency, and those that co-occurred with a content morpheme from the other language in the bilingual CP. Two categories, ‘discourse postpositions’ and ‘complementizers’, are designated for the type and token statistic work, as shown in Table 1. 4 All particles are Japanese.
Numbers of bilingual Projection of Complementizers (CPs) with Japanese discourse particles.
Note that the type numbers include the diary entries in the parenthesis, and this is the reason why they are, in fact, larger than the tokens.
Discourse postpositions in bilingual CP
The first four particles always appear phrase-finally and are attached phonologically to a preceding word. From the perspective of information structure, they show the following pattern.
(3) a. XTHEME-wa/-ne/-sa + YRHEME
5
b. XRHEME-mo + YTHEME
The examples (4) are of the pattern (3a), and (5) follows the pattern of (3b), where X is Chinese and Y is Japanese.
(4) a. dàshù tree- ‘The tree is green, isn’t it?’ b. zhège this- ‘This, I got it.’ c. zhè-ge- this- ‘This, what is this?’ (5) hora, shǒu look hand-also have- ‘Look, it also has hands.’
When Japanese discourse postpositions are attached to Chinese X, the ML identification depends on the morpheme realization of Y. In examples (4a,b,c) and (5), since the Y parts are all in Japanese, the MLs are identified as Japanese.
The statistical analysis in Table 2 suggests that the ML is more likely to be Japanese (62%) than Chinese (38%) when Y is a rheme, while it tends to be Chinese (61.3%) instead of Japanese (38.7%) when X is a rheme. In other words, the unmarked ML seems to be determined by the language of a rheme part of the bilingual CP. Since the ML and information structure are interrelated, it is plausible that the ML is affected by the information structure that represents the degree of salience in bilingual CP.
Morpheme realization for the Y part.
In examples (6) and (7), X is Chinese as in (4) and (5), but Y is also Chinese. In other words, the whole sentences are Chinese except for discourse particles. In these cases, Chinese is identified as the potential ML, with some mixed EL islands inserted.
(6) a. zuǐba- mouth- b. zhè-ge- this- ‘This is a small mushroom.’ c. táozi- peach- ‘As for peaches, well, some are sweet, some are bitter.’ (7) Shǒu- hand-also not paint-finish- ‘The hand has not been painted, either.’
As pointed out in the previous literature review, the 4-M model classifies morphemes into four subcategories: content morpheme; early system morpheme; bridge late system morpheme; and outsider late system morpheme. Now, what are the Japanese discourse postpositions in examples (6) and (7)? Takagi (2008) classifies the topic marker -wa into content morphemes in her study of English-Japanese CS. However, it is apparent that, except for the focus marker -mo ‘also’, the Japanese theme particles -wa, -ne and -sa do not indicate any conceptual meaning, nor do they assign any thematic role contra the definition of the content morphemes under the 4-M model. Therefore, Japanese discourse postpositions should not be categorized into content morphemes.
Are they early system morphemes? No, because they are not selected by a content morpheme but by discourse, and they do not flesh out the meaning of their heads like determiners or plural -s.
They are different from bridge morphemes, such as of in English, in that they do not integrate content morphemes into a larger constituent. Furthermore, they activate language-specific pragmatic feature bundles and contribute to conceptual structures; -wa is a topic marker; -mo corresponds to ‘also’ in English; -sa and -ne are markers of involvement.
They are not outsiders, either. They do not depend on information outside of the elements with which they occur like case markers or subject–predicate agreement. Case markers are only grammatically constrained. However, discourse postpositions are not only constrained grammatically, but also constrained by discourse or pragmatic conditions.
Now we need a new category for syntax–discourse interface. Japanese discourse postpositions in this study, all of which indicate functional sentence perspective, and some of which solicit involvement from the addressee, do not fit in with any categories in the 4-M model. Let us call them Discourse Morphemes.
Note in passing that both Japanese and Chinese are topic-prominent languages, and Chinese also has some topic markers that are equivalent to the Japanese -wa. However, this study finds no cases of the ‘Japanese topic-Chinese topic marker’ and only few instances of the pattern ‘Chinese topic marker + Japanese comment’.
(8) cài- vegetables- ‘Vegetables, I do not quite like them.’
The fact that Chinese discourse particles are rare in CS may imply that the dominant language in the target child is Japanese, not Chinese, which proved to be true in Meng and Miyamoto’s study (2009) on language dominance. The child originally intends to set up a Japanese-oriented discourse frame, but the mother’s speech is Chinese-dominated. Hence, in order to achieve harmony during conversing with the mother, the child switches back and forth between the two languages. Japanese serves as the Matrix Frame into which a large Chinese constituent (EL island) is inserted. This is common in passive bilingualism, where children understand their parents’ heritage language, but respond mainly in the social community language. As indicated above, the unmarked ML is probably determined by the language of a rheme part of the bilingual CP, so that bilingual infants in Japan are inclined to use Japanese for rhemes.
These phenomena may be accounted for as ‘habitualization’, which means that
Habitual contact with the pragmatic norms of another language and its speech community may lead to a situation where speakers are unable to deactivate discourse forms from one language even when in psycholinguistic terms it may not be the ‘selected’ language of a communicative situation. (Hlavac, 2006, p. 1871)
The notion of habitualization appears to explain why Japanese discourse forms are preponderant in otherwise Chinese discourse to a large scale, particularly if Japanese has become the dominant language of the speaker.
Chinese has no equivalents to the other two Japanese discourse postpositions -ne and -sa. Thus, the presence of Japanese discourse postpositions is in a statistical sense, additive, not subtractive. They can also find their explanation in ‘habitualization’. The same arguments will be applied on complementizers in the following analysis.
Complementizers in bilingual CP
Complementizers are one of the most complicated categories in the morpheme classification of the 4-M model. Myers-Scotton and Jake (2009) claimed that the morpheme type of such categories depends on specific complementizers in specific languages. They may belong to bridge system morphemes (such as that-type complementizers), outsider system morphemes (such as the Arabic li?anhum ‘because’, which also functions as an agreement marker) or even content morphemes (such as the Spanish porque ‘because’, which is conceptually activated). Hence, there is no uniform classification for the complementizers.
How about Japanese complementizers? They are bound morphemes, and are phonologically attached to the clauses that precede them. Semantically, they encode procedural information rather than conceptual information, to use Relevance-Theoretic terms (Blakemore, 2002). In other words, they encode constraints on the inferential processes rather than truth-conditional information. They are discourse-oriented in that they ‘guide the hearer toward the speaker’s meaning by indicating the direction in which relevance is to be sought’ (Wilson & Sperber, 2012, p. 26). 7
In what way do Japanese complementizers indicate the inferential direction? One of their most salient characteristics is that they allow ‘ellipsis’: a subordinate clause appears with a complementizer but without its main clause, as below.
(9) a. Moo osoi-
already late-because ‘It’s late so…’ b. Kaigi-wa tyuusida-
meeting- ‘(They say) that the meeting has been called off.’
Elided clauses are recoverable with complementizers. Thus, the interpretation of (9a) involves an inferential procedure that results in the cognitive effect of contextual implication, where new information (for example, ‘Let’s go home.’) follows from the combination of the subordinate clause and the context.
This study examines four Japanese complementizers that involve CS in comparatively higher frequencies, as listed in Table 3. They usually occur in the following pattern: ‘
Frequency of the six patterns with complementizers.
To identify the ML for either CP1 or CP2, all the possible combinations of S1 and S2 must be examined first. Since S2 is optional for some complementizers, the combinations of S1 and S2 consist of six patterns:
(a) Japanese S1 -Comp. + ∅; (b) Japanese S1 -Comp. + Japanese S2; (c) Japanese S1 -Comp. + Chinese S2; (d) Chinese S1 -Comp. + ∅; (e) Chinese S1 -Comp. + Japanese S2; (f) Chinese S1 -Comp. + Chinese S2.
So, ML identification for the composite bilingual CP depends on the morpheme realization of both clauses. Table 3 summarizes the occurrence of the six patterns with those Japanese complementizers.
As shown in Table 3, not all patterns were observed. Below are some examples with -kara ‘because’; its frequency is the highest among the four complementizers in our data, and it is the only complementizer that appears in all the six patterns of bilingual CP.
(10) Pattern (a): Dà-de o-katazuke-suru- big- ‘Because I will put away those big ones (crayons).’ (11) Pattern (b): Sizuka-ni-ne, níwáwa ne-teru- quiet- ‘Be quiet, because the doll is sleeping, ok?’ (12) Pattern (c): Koko-ni yòng diànnǎo-ba, zhuōzi aru- here-at use computer- ‘You can use the computer here, because there is a desk.’ (13) Pattern (d): Bǎobao kàn-shū-ne- I read-book- ‘Because I am reading.’ (14) Pattern (e): Hǎoduō hǎoduō tabe-te, zhǎng gāo-gè’r- many many eat-and grow tall-size-because ‘I should eat many many (foods), because I can grow tall.’ (15) Pattern (f): Tàiyáng méiyǒu chū-lai-
sun not out-come-because sleep- ‘Because the sun did not come out, is it sleeping?’
The patterns (a) and (b) in examples (10) and (11) can be interpreted as a Chinese NP insertion into the ML of Japanese S1. For pattern (c) in (12), S1 involves a Chinese NP insertion into a Japanese ML, while S2 involves a Japanese PP insertion into a Chinese ML. In other words, the ML in (12) is altered from Chinese S2 to Japanese S1.
Although pattern (d) in (13) is only composed of a Chinese S1 plus the Japanese -kara, it is reasonable to consider that a large Chinese EL island is inserted into Japanese CP1 headed by complementizer -kara, and that the ML of CP2 is also Japanese, since the ellipsis of the main clause is a characteristic found only in Japanese.
For pattern (e) in example (14), Japanese S2 contained a Chinese NP insertion. Chinese S1, on the other hand, can be interpreted as an EL island inserted into Japanese CP1 framed by -kara.
For pattern (f) in example (15), similarly, Chinese S1 is interpreted as a large EL island inserted into Japanese CP1 headed by -kara. Then, ML alternation from Japanese CP1 to Chinese CP2 is supposed to occur at the clause boundary.
Let us move on to examples with the conditional complementizers -tara or -nara ‘if’, which always follow a predicate as a suffix. Morphologically, -tara is a fusion 9 of an aspect marker of perfect -ta and a C feature [+conditional mood]. Since the inflection head and complementizer head are both Japanese, S1/CP1 is naturally Japanese, as in (16). In fact, we could not find any Chinese S1 for the bilingual CP with -tara in the present data. As for -nara, it is etymologically a fusion of classic Japanese copula -nari and a C feature [+conditional mood], but now that the copula is obsolete, we can safely conclude -nara is a complementizer and it is possible to attach it to Chinese S1, as shown in (17), where Chinese S1 is first inserted into Japanese CP1 as an EL island, and then a ML alternation takes place at the boundary between the complementizer and Chinese S2.
(16) Pattern (c): Māma, Bǎobao ne- Mom I sleep-if- ‘Mom, if I go to bed, you are going to read book?’ (17) Pattern (f): Māma bú-kùn-
Mom not-sleepy-if read-book go kùn-
sleepy-if sleep- ‘Mom, if you are not sleepy, go to read; if you are sleepy, go to bed.’
Let us turn to the complementizer -to, which indicates either a quotation (‘that’) or condition (‘if’).
(18) Pattern (e): Wáwa bù-lěng-
doll not-cold-that say- ‘The doll said she is not cold.’ (19) Pattern (c): Kāi-dēng si-nai- turn on-light do-not-if who also look-not-see ‘If the light is not turned on, no one can see (anything).’
In (18), the Chinese adjective phrase bù-lěng ‘not cold’ is inserted into the Japanese ML as an EL island. In (19), a ML alternation occurs after the Japanese complementizer -to.
Lastly, the marker of coordination -de connects two or more clauses in parallel, like ‘and’ in English, as in (20).
(20) Pattern (f): Bàba yǒu shìqing-
Dad have things-and Mom go to school-and Bǎobao qù hoikuen. (11-APR-2008) I go kindergarten ‘Dad has something to do, Mom goes to school, and I go to kindergarten.’
We have thus far seen CS examples with Japanese complementizers. At this point, two questions arise. Firstly, can the complementizers be classified into one of the subcategories of the 4-M model? Secondly, are they constrained by the MLF at all?
The answer to the first question is no. The same logic with discourse postpositions applies here. Japanese complementizers, as well as discourse postpositions, are sharply distinguished as Discourse Morphemes from other system morphemes in the 4-M model.
As for content morphemes, Myers-Scotton and Jake (2009) treated subordinators, such as ‘because’ and ‘if’, as content morphemes, since they seem to give a ‘discourse role’, so that content morphemes are defined as conveyers of semantic meaning and pragmatic meaning. However, we claim that if content morphemes are conceptually activated, subordinators should not be included into this type. Many studies of pragmatics (Blakemore, 2002; Grice, 1975; Levinson, 2000) show that subordinators do not convey conceptual meaning, but they are triggers of implicature. They link two independent propositions that are truth-conditional, and subordinators themselves are not truth-conditional. Hence, the complementizers encode constraints on the inferential processes rather than truth-conditional information. Therefore, they should not be categorized into content morphemes, which are defined as conveyers of semantic meaning only in the present study.
Are they early system morphemes? No, because they are not selected by a content morpheme but by discourse, and they do not flesh out the meaning of their heads.
They are different from bridge morphemes in that they do not integrate content morphemes into a larger constituent. Furthermore, they activate language-specific pragmatic feature bundles and contribute to conceptual structures.
They are not outsiders, either. They do not depend on information outside of the elements with which they occur. Outsiders first and foremost play the role of making relations transparent between elements in the clause (e.g. arguments and verbs). If the role of outsiders is to disambiguate grammatical roles and provide argument structure, the function of Discourse Morphemes is to provide the information structure and clarify the pragmatic information. In this way, we can safely separate outsiders from Discourse Morphemes.
The second question is difficult to answer. One might argue that it is unnecessary to determine the ML of the whole CP2, because the unit of CS defined by Myers-Scotton (2002) is CP and the composite sentence is composed of two CPs. In fact, Table 3 shows that the occurrences of CS utterances look arbitrary.
Table 4 summarizes the quantitative results of languages for the two clauses. It apparently illustrates that Japanese ML is only weakly predominant, so that ML selection of CP1 and CP2 is independent, as predicted by MLF theory. This is partly correct, but not entirely. This is because it does not account for S2 ellipsis. As we saw above, Japanese complementizers are not always followed by a main clause. Suppose null S2 is licensed in Japanese by the complementizer of the subordinate clause S1, and then S2 deletion is only possible when C of CP1 is Japanese. In this sense, the ML of CP1 can sometimes affect the ML selection of CP2. Therefore, Japanese discourse particles are highly activated, while Chinese often appears as an EL island, or the alternated language at the clause boundary. In other words, the MLF at the discourse level for the child’s output is, in fact, mostly Japanese.
Language of S1 and S2.
Example (21) exhibits a longer CS utterance with Japanese as the ML at the discourse level.
(21) Jiē Bǎobao, ránhòu jiù outi-ni kaet-te, gohan-o tabe-
meet I then then home-to return-and dinner- Māma-to Bàba-
mom-and dad- Nǎinai Yéye-to Xin-tyan-
grandma grandpa-and I- ‘Pick me up (in the kindergarten), then come back home, after having dinner, Mom and Dad go to study, and Grandma, Grandpa and I go to sleep.’
Discourse morphemes and constraints on code switching
So far we have examined utterances of three Chinese–Japanese bilingual children to clarify discoursal constraints on CS with discourse particles.
Due to their morphological and pragmatic features, this study regards both Japanese discourse postpositions and complementizers as Discourse Morphemes, which designate the matrix frame at the interface of syntax–discourse. It is not implied that Discourse Morphemes are the fifth subcategory of the 4-M model. Since the MLF model and the 4-M model are independent, it is possible that Discourse Morphemes do not belong to any subcategory of the 4-M model, but they are constrained by the MLF at the syntax–discourse interface.
Our proposal also accounts for the fact that Discourse Morphemes can sometimes be preceded by a large EL island and followed by ML alternation. An EL Island is a kind of multi-item insertion. It was originally introduced as one of the compromise strategies to solve the problem of incongruence between the two participating languages, and realize smooth switching (Myers-Scotton, 2002). On the other hand, it also reflects a state of high activation of the EL during the language production process. It has been explored in relation to a broad range from phrasal items to multi-morphemic structures, even clause size items (Namba, 2009).
In this paper, we have shown that Discourse Morphemes are often attached to a group of words. The discourse postpositions in (22) are not preceded by simple content morphemes but by noun phrases. In (23), Japanese complementizers are attached to Chinese sentences. In other words, all the Discourse Morphemes can be adjoined to EL islands, which are underlined in (22) and (23).
(22) a. big dog-bowwow- ‘The big dog has gone home right?’ b. white-color- ‘Sharpen the white one, too.’ (23) a. I read-book- ‘Because I am reading.’ b. doll not-cold-that say- ‘The doll said she is not cold.’
Examples (22) and (23) indicate that EL islands can be inserted before Discourse Morphemes. ML alternation, on the other hand, freely occurs from a Japanese CP1 to a Chinese S2 at the boundary of the complementizer as in (14), etc. This indicates that the independence of CP1 is licensed by Discourse Morphemes, and it is inserted in the Chinese frame. In this sense, intra-sentential ML alternation is a kind of large EL island.
In other words, CS or EL island insertion is possible before and after Discourse Morphemes. This fact has three important theoretical implications.
Firstly, since EL islands are left-adjoined to Japanese Discourse Morphemes, the word order proves that Discourse Morphemes come from the ML. In this sense, CS with Discourse Morphemes is constrained by the MLF.
Secondly, given that EL islands are syntactically independent and Discourse Morphemes are in the discourse level, CS is possible at the syntax–discourse interface.
Thirdly, although the original definition on EL islands speculates that only EL morphemes are contained, both EL and ML morphemes can be enclosed in the EL islands and form mixed EL islands, as shown in the examples (6), (7) and (22), where Japanese postpositions are put between Chinese morphemes. In fact, such concept of the mixed EL islands have been defined by Backus (2003, p. 90) as some kind of multi-morphemic ‘lexical units’ to interpret some CS at the boundary between insertion and alternation.
If the suppositions are on the right track, we can say Discourse Morphemes are not only empirically but also theoretically motivated.
Conclusion
This study defines discourse particles as Discourse Morphemes, which differ from other subcategories defined in the 4-M model, such as case markers or inflections. Discourse Morphemes are in the peripheral positions of the sentence structure, such as topics, interjections and clause boundaries. The MLF model, but not the 4-M model, has been verified through Chinese–Japanese CS data, in that they partake in building the MLF at the discourse level, and thus are constrained by the MLF at the interface of syntax–discourse. Our hypothesis is therefore proved.
In addition, it is assumed that large EL islands and ML alternation are often concomitant with the occurrence of CS involving Discourse Morphemes. CS is possible at the syntax–discourse interface.
In summary, the general framework of the MLF has been extended from a merely syntactic hierarchy to a higher domain of syntax–discourse interface.
Further studies will continue to examine all particles involved in Chinese–Japanese bilingual CS, including discourse particles. The 4-M model will be eventually revised to provide a more effective and accurate account for the grammatical constraints on Chinese–Japanese CS utterances.
Footnotes
Appendix
In this appendix, some relevant contexts for most of the examples in the paper are provided for more comprehensive understanding. The underlined utterances are the examples.
*XIN – the focal subject
*YOU – another subject
*MOT – the mother
%eng – English translation
%act – accompanied action
(4a) Setting: XIN was coloring some book with MOT before sleeping.
%eng: The tree should be colored with green crayon, isn’t it?
*XIN: motto midori atta yo.
%eng: I found more crayons with the color of green.
*MOT: 嗯, 绿色.
%eng: Yes, green (crayon).
(4b) Setting: XIN and MOT were playing with some blocks after supper.
*XIN: zya koko sandoitti ni naru ka.
%eng: So here is it looks like a sandwich?
*MOT: 对, 这像一个三明治一样的.
%eng: Yes, it looks like a sandwich.
%eng: This (block), I see.
*XIN: koko ni osoto ii zyanai ka.
%eng: Let’s put it outside, shall we?
*MOT: 好啦.
%eng: All right, (it is outside now.)
(4c) Setting: XIN was watching a music CD of children’s songs with MOT after supper.
%eng: This scene, what is it about?
*XIN: omosiroi.
%eng: It is interesting.
*XIN: 散步呢?
%eng: Are they taking a walk?
*MOT: 这是去一个小动物园了吧.
%eng: Maybe they are walking around in a small zoo.
(5) Setting: XIN was drawing a picture of a girl with skirt, together with MOT before sleeping.
*XIN: 黄色的球球.
%eng: I want to draw some yellow circles (on the skirt).
*MOT: 干吗要画球球啊?
%eng: Why do you want to draw circles?
*XIN: datte 漂亮 ni natterun damon.
%eng: Because it will make the skirt look more beautiful.
*MOT: 哦, 在裙子上画点小花花什么的, 是不是.
%eng: Oh, you mean to draw some flowers on the skirt, right?
*XIN: kawaii, kawaii?
%act: drawing some circles on the skirt of a girl in the picture.
%eng: Is it cute?
%eng: Look, the girl also has hands.
%act: drawing two hands to the body of the girl.
*MOT: 哈哈哈, 好, 真棒, 不错不错.
%eng: Hahaha, good, great!
(6a) Setting: XIN was watching an English animation with MOT, which tells that a cow just gave birth to a little calf.
*XIN: 走路呢.
%eng: The little calf is walking.
*XIN: <它的 sippo nagai yo.>(×2)
%eng: It’s tail is really long.
*MOT: 哦, 长长的尾巴, 是不是?
%eng: Oh, long long tail, isn’t it?
%eng: Its mouth is red.
*MOT: 对, 有点粉色的.
%eng: Yes, a little pink.
*XIN: 红色的 zyanai no?
%eng: Isn’t it red?
*MOT: 粉粉的, 那个颜色.
%eng: Just a little pinky, that color.
*XIN: 手 wa?
%eng: What about its hand?
*MOT: 手？它是牛蹄子, 是黑色的.
%eng: Hand? It should be cattle hoof, and it is black.
(6b) Setting: XIN was reading some picture books together with MOT in the bed-room.
*XIN: nani kore?
%eng: What’s this?
*MOT: 哎，这是什么？
%eng: Ah, what’s this?
*MOT: Pizza, 是不是?
%eng: Pizza, isn’t it?
*MOT: 还记得咱们吃pizza吗?
%eng: Do you still remember that we once ate pizza?
*MOT: 买了pizza, 那叔叔得送到家里来, 是不是?
%eng: After you ordered it, the express delivery man had to deliver it to our home, remember?
%eng: This is mushroom.
*XIN: Xintyan 喜欢 小蘑菇 tuiteru no.
%eng: I like pizza with mushroom.
*MOT: 对.
%eng: Yes.
*XIN: 小蘑菇 to, 然后呢?
%eng: Mushroom, and what about this?
*MOT: 这是什么?
%eng: What is this?
*MOT: 小香肠.
%eng: Little sausage.
(6c) Setting: XIN was having lunch with MOT in the living room.
*MOT: 咱们先吃个一个梨吧, 好吗?
%eng: Let’s have a pear first, ok?
*XIN: ii yo.
%eng: Good.
*XIN: 桃子, 桃子 ga ii.
%eng: Peach, peach is better.
*MOT: 吃桃子呀, 好吧.
%eng: You want a peach, all right.
%eng: Peaches, but, even peaches, well, some are sweet, some are bitter.
*MOT: 哦,有的甜,有的苦啊.
%eng: Oh, some are sweet, some are bitter?
*XIN: 嗯, soo dayo.
%eng: That’s right.
(7) Setting: XIN was coloring a book with MOT before sleeping.
*XIN: 这个 ni suru, 胖虎的妹妹.
%eng: Let’s color this one, Gian’s sister.
*MOT: 这个已经都涂得差不多了.
%eng: This one you have almost finished coloring.
*MOT: 这只鞋没有涂完.
%eng: This shoe, you have not finished coloring.
*XIN: koko mo 鼻子 mo <没有涂完> ["].
%eng: Here, the nose has not been colored, either.
*XIN: koko mo 没有涂完 si, koko mo 没有涂完 si.
%eng: This is not finished, and this either.
*XIN: koko mo koko mo 没有涂完.
%eng: This and this are not finished.
%eng: Her hands have not been colored, either.
*XIN: ippai 没有涂完.
%eng: A lot of parts have not been colored.
(8) Setting: YOU was having supper with her parents. She only had a rice-ball, but her mother said she should eat more vegetables.
*MOT: 不加菜, 营养不够哦.
%eng: If you don’t have vegetables as well, the nutrition is not enough.
*YOU: 菜 kirai mon.
%eng: It’s that I do not like vegetables.
%eng: Vegetables, I do not quite like them.
*YOU: 菜sa, 菜 mo, 菜 wa, 菜 wa moo nakunattyatta mon.
%eng: Vegetables, there are no vegetables, anymore.
(10) Setting: XIN was coloring a book with MOT before sleeping. It’s too late, and MOT asked XIN to put those crayons back in the box, and get ready to sleep. XIN said she would put away those big (longer) crayons, and MOT should put away those shorter ones.
*XIN: zya 宝宝 大的 okatazuke.
%eng: So I will put those big (longer) crayons back in order, (while you deal with the shorter ones.)
*XIN: tyotto mattete ne.
%eng: Just wait, ok?
*XIN: kotti zyanai desyoo.
%eng: They should not be put here, right?
*XIN: kotti.
%eng: Here.
*XIN: mattete ne, tyotto.
%eng: Just wait a minute.
%eng: Because I am putting away those big crayons.
*XIN: 妈妈 ehon okatazuke 妈妈 ehon.
%eng: Mom, you can put away those picture books.
*MOT: 妈妈把书收起来, 啊.
%eng: Ok, Mom will put those books back in order.
(11) Setting: XIN was having lunch with MOT, meanwhile playing with her little doll. It was pretended that the doll was sleeping, so XIN asked MOT to be quiet.
*MOT: 小甜饼也挺好吃.
%eng: The sweet pancakes are also delicious.
*XIN: ima ne, 泥娃娃 ne, ima ne, 睡觉呢.
%eng: Now, the doll is sleeping.
*XIN: tyotto sii-sii.
%eng: Shush.
*MOT: 声音轻一点, 是不是?
%eng: I should make less noise, right?
*XIN: 嗯, yamete ne, 妈妈.
%eng: Yes, please don’t, Mom.
*MOT: 饺子也给你热几个吧.
%eng: I will also heat several dumplings for you.
%eng: Keep quiet, because the doll is sleeping, ok?
(12) Setting: After reading, MOT asked XIN to go to bed. Usually XIN sleeps inside the bed-room, while MOT studies outside in the living room. But this time XIN asked MOT to study in the bed-room, accompanying her.
*MOT: 好吧, 睡觉吧.
%eng: All right, go to bed.
*XIN: zya 泥娃娃 to Xintyan 睡觉.
%eng: So I will go to bed with my doll.
*XIN: kyoo wa 妈妈 看书?
%eng: Today Mom is going to study?
*MOT: 嗯, 妈妈看书.
%eng: Yes, I’m going to study.
*XIN: koko de?
%eng: Here?
*MOT: 不行, 妈妈得在外边.
%eng: Not here, Mom has to study outside.
*XIN: doosite?
%eng: Why?
*MOT: 妈妈要用电脑啊.
%eng: Because I have to use a computer.
%eng: You can use your computer here, since there is a desk here.
*XIN: 把书 koko ni oitete.
%eng: You can put your books here on the desk.
*MOT: 你倒挺会给妈妈安排呀.
%eng: How good you are at arranging things for me!
(13) Diary data. Setting: XIN has just got up in the morning, and was reading in bed.
*MOT: Xinying tyan 快穿衣服.
%eng: Xinying, put on your clothes quickly.
%eng: It’s just that I’m reading.
(14) Diary data. Setting: XIN was having supper with MOT.
*MOT: 多吃点.
%eng: Have more food.
%eng: Yes, I should eat much food, so that I can grow tall.
(15) Diary data. Setting: MOT was on her way to send XIN to the nursery school in bicycle, and they passed the place where they had ever found some snails.
*XIN: katatumuri?
%eng: Snails?
*MOT: 小蜗牛呢?
%eng: Where are the snails?
%eng: Since the sun has not risen, they might still be sleeping.
(16) Setting: After drawing, MOT asked XIN to go to bed.
*MOT: 好啦, 晚安.
%eng: All right, good night!
%eng: Mom, if I go to bed, are you going to read book?
*MOT: 嗯, 好.
%eng: OK.
*XIN: wakatta.
%eng: I see.
*XIN: oyasumi.
%eng: Good night!
*MOT: oyasumi.
%eng: Good night!
(18) Diary data. Setting: After lunch, XIN was playing with the doll. She took off the doll’s dress.
*MOT: 你怎么又给娃娃脱衣服了?
%eng: How come you took of the doll’s dress again?
%eng: Because the doll said she is not cold.
%act: XIN looked outside and saw that the sun had come out.
*XIN: moo 天气热了嘛.
%eng: The weather has already turned hot.
(19) Diary data. Setting: After having supper outside, MOT was taking XIN back home in bicycle.
*XIN: 回家先开暖气, 还有开灯.
%eng: When we arrive home, we should first turn on the heater, and also turn on the light.
%eng: If the light is not turned on, no one can see anything.
(20) Diary data. Setting: XIN was talking to herself on her way to the nursery school on MOT’s bicycle.
%eng: Dad has something to do, Mom goes to school, and I go to kindergarten.
(21) Setting: XIN was getting ready to go to bed, and MOT told her that grandparents are coming to visit next month.
*XIN: zya, 奶奶爷爷 kitemo, 妈妈也扎辫子, 啊!
%eng: So, even if grandma and grandpa come, Mom will also plait my hair, will you?
*MOT: 好吧.
%eng: All right.
*XIN: 然后宝宝臭美呀, 啊!
%eng: And then I will make some pose with my beautiful plait.
*MOT: 哈哈哈.
%eng: Hahaha.
*XIN: 然后就, 奶奶就宝宝就送了.
%eng: And then it will be grandma who takes me to the nursery school.
*XIN: 奶奶宝宝送.
%eng: Grandma takes me to the nursery school.
*MOT: 奶奶送宝宝.
%eng: Grandma takes you to the nursery school.
*XIN: 嗯.
%eng: Fine.
*XIN: 然后回来就爷爷.
%eng: And then when I come back, it is grandpa.
*MOT: 爷爷接宝宝呀?
%eng: Grandpa will pick you up back home?
*XIN: 嗯.
%eng: Yes.
*MOT: 好吧.
%eng: All right.
%eng: (Grandpa) will pick me up (in the nursery school), and then come back home, after having dinner, Mom and Dad will go to study, and Grandma, Grandpa and I will go to sleep.
*MOT: 呵, 你都知道啦?
%eng: Oh, how come you already know everything!
(22a) Diary data. Setting: XIN was on her way to the nursery school on MOT’s bicycle. Sometimes XIN and MOT can meet two dogs on the way, a big one and a small one.
*XIN: 今天大的狗汪汪 inai, 小的狗汪汪 atta.
%eng: Today the big dog was not there, and the small dog was there.
%eng: The big dog may have gone back home.
(22b) Setting: XIN was coloring some pictures with MOT before sleeping. Some of the color pencils need sharpening.
*XIN: 这个没有修一修呀, 妈妈这个.
%eng: This color pencil, you did not sharpen it?
*MOT: 嗯?
%eng: Which one?
*XIN: kiiroi no 没有修一修呀?
%eng: The yellow one, you did not sharpen it?
*MOT: 哦, 该修一修了, 用转笔刀转一转了, 是不是?
%eng: Oh, it needs sharpening, isn’t it?
*XIN: 嗯.
%eng: Yes.
*XIN: 还有紫色的呢!
%eng: And the purple one, too.
*MOT: 好, 妈妈给你拿铅笔刀.
%eng: Ok, Mom will fetch the pencil-sharpener.
*XIN: 还有绿色的.
%eng: And the green one.
*MOT: 好, 等着.
%eng: Ok, wait.
%eng: Sharpen the while one, too.
*MOT: 稍微等一下嘛!
%eng: Just wait a moment.
*XIN: kore mo site yo.
%eng: This one also.
*XIN: ippai site.
%eng: (We should sharpen) many of them.
Funding
This work was supported by the Humanity and Social Science Sector of the Ministry of Education in China (Grant Number 13YJC740067).