Lingua 118 (2008) 777–809 www.elsevier.com/locate/lingua
Code-switching, word order and the lexical/functional category distinction Brian Hok-Shing Chan Department of English, Faculty of Social Sciences & Humanities (FSH), University of Macau, Av. Padre Tomas Pereira S.J., Taipa, Macao Received 4 December 2004; received in revised form 27 March 2007; accepted 16 May 2007 Available online 3 July 2007
Abstract This paper claims that lexical categories (V, N) and functional categories (D, I, C) behave differently in bilingual code-switching: whereas functional heads always determine the order of their code-switched complements, lexical heads may not do so. This proposal thus deviates from many recent studies which suggest that all heads determine the order of their complements (e.g. Mahootian, 1993; MacSwan, 1999; Nishimura, 1997; Nishimura and Yoon, 1998). Assuming a ‘‘Null Theory’’ perspective (Mahootian, 1993; MacSwan, 1999), code-switching data are explained here in terms of existing syntactic apparatus which also governs monolingual syntax. It is proposed that word order between lexical categories and their complements is determined by head parameter instead of feature strength as an intrinsic property of the lexical heads. Nonetheless, head-complement order is inherently specified in functional categories. On this account, prepositions are functional heads instead of lexical heads. # 2007 Elsevier B.V. All rights reserved. Keywords: Code-switching; Word order; Head-complement order; Null Theory; Functional/lexical distinction; Prepositions
1. Code-switching, The Null Theory and the language faculty Are there grammatical constraints or principles governing code-switching? What are these constraints and principles? These questions have remained a central line of inquiry in codeswitching research since the pioneering studies by Pfaff (1979) or Poplack (1980). Proposals
E-mail address:
[email protected]. 0024-3841/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.lingua.2007.05.004
778
B.H.-S. Chan / Lingua 118 (2008) 777–809
range from universal constraints or models (e.g. The Matrix Language Frame Model—MyersScotton, 1993, 2002, etc.) at one end of the continuum to the view that there are no constraints (e.g. Bokamba, 1989) at the other, and somewhere in the middle lie models which allow for different constraints or principles for different language-pairs (Bhatt, 1997; Muysken, 1997, 2000; see the literature review in Chan, 2003 for a comprehensive summary). Recently, the debate seems to have taken a new twist: The idea which has gained momentum is that codeswitching and ‘‘pure’’ languages are governed by the same constraints or principles which form the language faculty or universal grammar—a position that has been called The Null Theory (Mahootian, 1993; MacSwan, 1999, 2000; Chan, 2003, etc.). It is obvious that The Null Theory has advantages over many previous models or constraints in being economical in theory: One need not posit constraints or principles which apply specifically to code-switching (Mahootian, 1993; MacSwan, 1999, 2000). There is another advantage which has been less frequently discussed but perhaps more crucial: That is, The Null Theory presents a more economical theory of the language faculty which does not contain an additional grammar for code-switching (Chan, 2003:57–58). This facet makes The Null Theory all the more preferable if one considers that code-switching is very likely to be the result of exposure to more than one language only after birth. In other words, all human beings, except the pathological cases, are endowed with the same language faculty (or Universal Grammar)—the classic argument in generative grammar (e.g. Chomsky, 1986a); however, depending on their exposure to different languages, they will end up speaking different languages (e.g. Chinese, Japanese, English or Swahili, etc.) or, in the case of more than one language, engaging in code-switching. Put it in another way, people do not code-switch not because they were born without ‘‘a code-switching grammar’’; rather, this is due to the absence of input of more than one language (in the case of monolinguals) or the lack of socio-cultural motivations (in the case of bilinguals).1 Those in favour of constraints specific to codeswitching may still assume that there is a code-switching grammar being dormant in monolinguals, but then the language faculty would become unnecessarily complicated. All in all, under The Null Theory, code-switching and ‘‘pure’’ languages are subject to the same linguistic constraints or principles; accordingly, code-switching is no longer a hybrid peripheral in linguistics but a potential source of data which reveals the architecture of the language faculty. This paper is an attempt in this enterprise by showing that code-switching patterns from various language-pairs do reflect a fundamental distinction in the language faculty that has long been assumed in linguistics, namely, functional categories and lexical categories. More specifically, I examined code-switching patterns in phrases where headcomplement order is different in the two participating languages. The data indicate that the language of lexical heads may not determine the position of their complements but the language of functional heads does so. In this paper, I aim to accommodate these data by outlining a model based on Chan (2003); I argue that this model is in compliance with The Null Theory while offering interesting insights into grammatical theory, in particular the theory of word order and the dichotomy of lexical categories versus functional categories. Section 2 provides some background to recent code-switching research concerning word order. I will turn to the data of relevant language-pairs in sections 3 and 4. Section 3 is devoted to lexical categories, namely, V and N, showing that they may not determine the order of their 1
See Chan (2004) for a general survey of the motivations of code-switching.
B.H.-S. Chan / Lingua 118 (2008) 777–809
779
code-switched complements.2 It is deduced that the complements of lexical categories appear in parametrized positions determined by word-order parameters, or they appear in specifier position of functional heads. Section 4 is devoted to functional categories, showing that they do determine the position of their code-switched complements. It is then suggested that head-complement order is specified as an inherent property of functional categories in terms of a ‘‘lexical parameter’’. Section 5 is the conclusion. 2. Code-switching, word order and the lexical/functional distinction The issue of word order has been central in studies on the grammar of code-switching. Poplack (1980) proposed The Equivalence Constraint that permits code-switching within phrases the linear orders of which are identical in the two participating languages. The Equivalence Constraint has been undermined by numerous counter-examples in various language-pairs (see Myers-Scotton, 1993; Santorini and Mahootian, 1995; Stenson, 1990; Halmari, 1997; Kamwangamalu, 1989), and most examples to be presented in later sections also disprove the constraint. Myers-Scotton (1993, 2002), in her Matrix Language Frame Model, suggests that it is the order of the matrix language which is followed (following The Morpheme Order Principle). The Matrix Language not only determines the word order of a code-switched sentence but it also supplies the system morphemes—most of which are closed class items or functional categories— according to The System Morpheme Principle.3 However, some data have been presented in which the matrix language cannot be clearly determined in a code-switched sentence/CP (see Bentahila, 1995; Boussofara-Omar, 2003; Muysken and de Rooij, 1995). Apart from these empirical problems, in the light of The Null Theory, The Equivalence Constraint and The Matrix Language Model both suffer from the same weakness; that is, they apply specifically to codeswitching and therefore they are not the optimal accounts. According to The Null Theory, wordorder in code-switching is best accounted for by mechanisms which also account for word order in ‘‘pure’’ languages (e.g. directionality parameters); word-order in code-switching is not explained by virtue of code-switching specific constraints or principles, such as The Equivalence Constraint or The Morpheme Order Principle as subsumed in the Matrix Language Frame Model. Notice that, in the case of the latter, if we appeal to The Morpheme Order Principle, we may be bound to accept a number of other principles and models as well under the rubric of The Matrix Language Frame Model (see an update in Myers-Scotton, 2002). Many proponents of The Null Theory have been preoccupied with word order too. Despite their different database and theoretical frameworks, they seem to have agreed that language of the head determines the order of the complement (Mahootian, 1993, 1996; Mahootian and Santorini, 1996; MacSwan, 1999; Nishimura, 1997; Nishimura and Yoon, 1998; Santorini and Mahootian, 2
The categories of A (i.e. adjectives/adverbs) or P (i.e. prepositions/postpositions), which are conventionally analyzed as lexical categories (Chomsky, 1986b; Ouhalla, 1991), are not considered for respective reasons: For the category A, most studies have focused on code-switching data involving attributive adjectives (see discussion in Santorini and Mahootian, 1995; Mahootian and Santorini, 1996), and adverbs which, on current account, are analyzed as either specifiers of functional heads (Cinque, 1994, 1999; Haegeman and Gue´ron, 1999) or adjuncts (Duffield, 1999). While a predicative adjective may be analyzed as a lexical head able to take a complement, I have not observed code-switching between a predicative adjective and its complement. As for the category P, I will argue in section 4 below that it is in fact a functional category, echoing recent works by Froud (2001) and Baker (2003). 3 In earlier accounts, Myers-Scotton (1993) appeared to link the system morphemes with the closed class items or functional categories. More recently, Myers-Scotton (1997, 2002) has made it clear such distinction does not directly correspond to the function/lexical distinction. See Chan (2003, ch. 2) for a more detailed discussion of these developments.
780
B.H.-S. Chan / Lingua 118 (2008) 777–809
1995). Mahootian (1993) sees head-complement order as a lexical property specified in the head: Under Tree Adjoining Grammar/TAG, which Mahootian (1993) adopted, a head, when accessed, projects a tree structure with the empty position of its complement already specified. Nishimura (1997) follows The Principles-and-Parameters framework, and head-complement order is also determined by the head, but here the mechanism is a ‘‘directionality parameter’’. MacSwan (1999) assumes the Minimalist Framework. In discussing code-switching between VO and OV languages, he suggests that the head verb is specified with a case feature. Assuming the framework of Kayne (1994), the underlying structure of VP is supposed to be VO. If the case feature is strong, the object raises, resulting in OV order. If the case feature is weak, the object remains in-situ, resulting in VO order. According to this theory, VO/OVorder is a lexical property of the V head. Granted the assumptions of The Null Theory and the theories of word order these researchers adopt, it would be logical and consistent that heads determine the position of their complements, irrespective of whether these complements are code-switched or not. What’s more, as these researchers have shown in their respective works, there are data in which the language of the head, in particular the verb, does determine the position of the code-switched complements. Mahootian (1996) sums up this position in the following statement: (1)
The language of a head determines the syntactic properties of its complements in codeswitching and monolingual contexts alike. (Mahootian, 1996:387)
In spite of its appeal, the claim that all heads must determine the position of (code-switched) complements seems too restrictive: It is possible that code-switched elements may be ‘‘assimilated’’ into the morphosyntactic structures of another language. Data which fit into this description have been well-documented from various language-pairs, in particular those works which presuppose a ‘‘matrix language’’ governing the morphosyntactic structure of a codeswitched sentence (e.g. Joshi, 1985; Kamwangamalu, 1989; Myers-Scotton, 1993, 2002, etc.). One need not accept the view that such a matrix language can be clearly identified in every codeswitched sentence and hence it is indeed represented in the competence of bilinguals (as suggested by Myers-Scotton, 2002). However, there is no denying that code-switched elements (being heads or non-heads in phrases) can or may take part in a syntactic structure (e.g. word order) characteristic of another language. Data from various language-pairs have suggested a difference of behaviour between lexical categories (N and V) and functional categories (D, I and C) in relation to the order of their codeswitched complements: The language of a lexical head may or may not determine the order of its complement, but the language of a functional category always determines the position of its complement (Chan, 2003). While this finding appears to defy a Null Theory account with the mechanisms in Mahootian (1993), Nishimura (1997) and MacSwan (1999), I argue in this paper that these data can indeed be captured without resorting to constraints or principles specific to code-switching, and therefore the finding is in fact not a contradiction to The Null Theory. After all, if there is really a major distinction between functional categories and lexical categories in the language faculty as commonly assumed, this difference is bound to be reflected in codeswitching in some way according to The Null Theory.4 4 Actually, a number of earlier works have suggested that functional categories and lexical categories behave differently in code-switching (Belazi et al., 1994; Joshi, 1985 and Azuma, 1993, 1997 in terms of ‘‘open class’’ and ‘‘closed class’’, Myers-Scotton, 1993). This paper, however, makes different assumptions and predictions, namely, functional categories and lexical categories have different word order properties and the language faculty does not have specific principles stipulating the role of the ‘‘matrix language’’ in code-switching.
B.H.-S. Chan / Lingua 118 (2008) 777–809
781
In brief, I propose a model in which heads merge with complements on both sides according to parameter values or lexical specification—contrary to the widely assumed ‘‘Antisymmetry framework’’ (Kayne, 1994) but consistent with Neeleman and Weerman (1999) or Saito and Fukui (1998). Furthermore, a production mechanism in Levelt’s (1989) fashion is incorporated to account for various code-switching patterns. The addition of this production mechanism is not seen as a complication of the grammar, since a production system, perhaps not thoroughly understood at the moment, is present anyway in the language faculty. Evidence for such a production system can be found in various types of speech errors which arise from errors at various stages of such a production system (see Garrett, 1980, 1982, 1988; Levelt, 1989). 3. Code-switching between lexical heads and complements 3.1. Code-switching between verbs and objects or other complements This section deals with code-switching between verbs and objects (i.e. ‘‘complements’’ of ‘‘head’’ verbs in terms of X-bar Theory). I will cite a first set of data, which shows that in many examples the head verb does determine word order. Then, I will show another set of evidence which suggest that the objects are moved or scrambled. These data indicate that the head verb does not solely determine the position of code-switched objects, just as the head verb does not in monolingual syntax. A third set of data will then be given, in which the position of the objects contrasts with that specified in the language of the verb. These data show that, if one assumes that code-switching patterns emerge from the same language faculty as that of ‘‘pure’’ languages, head-complement order is not a lexical property of the verb. These data are then accounted for by word-order parameters and Levelt’s (1989) production algorithm. Let us turn to the first set of data. In examples (2)–(5), a verb from a VO language takes a postverbal object. In examples (6)–(9), a verb from an OV language takes a preverbal object.5 These data can be captured by the models in Mahootian (1993, 1996), Nishimura (1997) or MacSwan (1999), despite a difference in theoretical frameworks they adopted. V from VO language, VO order (2)
Tell them you’ll buy xune-ye jaedid tell them you’ll buy house-POSS new ‘‘Tell them you’ll buy a new house. . .’’ (English–Farsi, Santorini and Mahootian, 1995:10, (13a))
(3)
nisei no jidai ni wa we never knew anna koto nanka Nisei POSS days P TOP we never knew such thing sarcasm ‘‘In the days of Nisei, we never knew such a thing as sarcasm.’’ (Japanese–English, Nishimura, 1985:76, (74))
5
The examples in this paper are quoted according to the following conventions in accordance with current practices in the literature (e.g. Santorini and Mahootian, 1995; MacSwan, 1999, etc.). The first line is the example. The language which appears first in the example is marked in normal face, whereas the language which appears later is marked in italics. The second line is the gloss. The third line is a free translation of the example in English. The fourth line is a caption which includes the language-pair of the example and the source. In the description of language-pairs, the language which first appears in the example is presented first in normal face, followed by the language which appears later in italics.
782
B.H.-S. Chan / Lingua 118 (2008) 777–809
(4)
He eats daliaa for dinner He eats porridge for dinner ‘‘He eats porridge for dinner.’’ (English–Hindi, Pandit, 1986:92, (19))
(5)
Hana, put cangnakam in the basket quickly and go home Hana, put toys in the basket quickly and go home ‘‘Hana, put the toys in the basket quickly and go home.’’ (Korean–English, Lee, 1991:142, (18)) V from OV language, OV order
(6)
ten dollars dad- e ten dollars gave-PERF ‘‘She gave ten dollars.’’ (English–Farsi, Mahootian, 1993:150, (104))
(7)
Only small prizes moratta ne only small prizes get-PAST ‘‘(We) got only small prizes.’’ (English–Japanese, Nishimura, 1986:128, (3))
(8)
Their American friends-eykey ilpwule Korean language sseyo their American friends-DAT intentionally Korean language use ‘‘(They) use Korean to their American friends intentionally.’’ (English–Korean, Park, 1990:75, (59))
(9)
ki Syria uske sath diplomatic relations kayam kare that Syria it with diplomatic relations establish do ‘‘. . .that Syria establishes diplomatic relations with it. (Hindi–English, Bhatt, 1997:228, (4b))
The head verb can hardly be the sole determinant of the order of the code-switched complements. There is another set of data in which the surface position of code-switched objects is not determined by the head verb: The objects may have undergone scrambling or movement, as in the case of objects in monolingual syntax. (10)
Tall buildings i nihon ga ti tsukuru tall buildings Japan NOM construct ‘‘Japan constructs tall buildings.’’ (English–Japanese, Nishimura, 1997:124, (20b))
(11)
Are-o i you have to learn ti. that-ACC you have to learn ‘‘That, you have to learn.’’ (Japanese–English, Nishimura, 1997:124, (19a))
(12)
hankuli-un I can read ti but I can’t understand ti Korean-TOP I can read but I can’t understand ‘‘Korean, I can read but I can’t understand.’’ (Korean–English, Park, 1990:88, (66))
B.H.-S. Chan / Lingua 118 (2008) 777–809
783
Mahootian (1993, 1996) and MacSwan (1999) had little to say about the possibility of scrambling or movement of code-switched objects. Neither did Nishimura (1997) suggest any principled account for the above examples she collected (i.e. (10) and (11) above). Nonetheless, these examples do not pose a lethal problem to The Null Theory, since the syntactic theories they assume (i.e. Tree Adjoining Grammar/TAG for Mahootian, 1993, 1996, Principles-and-Parameters Model for Nishimura, 1997 and the Minimalist framework for MacSwan, 1999) have different apparatus to deal with scrambling or movement.6 In some language-pairs, the verb comes from an OV language but the code-switched objects are postverbal. These objects appear immediately adjacent to the head verb, and hence it is not obvious that they have undergone movement or scrambling. As such, these data contradict the predictions of Mahootian (1993, 1996), Nishimura (1997) and MacSwan (1999). V from OV language, VO order (13)
pooTuruvaan letter put (3-SG-MASC-FUT) letter ‘‘He will write a letter.’’ (Tamil–English, Sankoff et al., 1990:79, (3))
(14)
I have to ttake¯ my hand I have to wash my hand ‘‘I have to wash my hand.’’ (English–Korean, Choi, 1991:889, (42))
Santorini and Mahootian (1995:10–11) actually noted the examples like (13), but they claim that their framework, according to which the head verbs determine the surface position of objects, can still handle these cases. This is because ‘‘objects can follow their verbs (in Farsi), as in many other underlyingly OV languages’’ (Santorini and Mahootian, 1995:10). Even though objects in Farsi can appear on both sides of the verb, Santorini and Mahootian’s (1995) statement does not necessarily carry over to Tamil and Korean. These two languages have been analyzed as strictly OV, and postverbal objects have a very limited distribution. Furthermore, these postverbal objects have special pragmatic functions and intonational characteristics. In the Tamil case, Herring (1994) concludes that postverbal objects can be divided into three subtypes with slightly different pragmatic functions and intonations. Structurally, these postverbal objects are either postposed (i.e. by movement) or simply detached from the argument structure of the preceding clause, but they are not base-generated as Santorini and Mahootian (1995) suggest for Farsi. As for Korean, Chang (1996) reports that the postverbal objects are also restricted, and uttered only after a pause. This is in contrast with other typical VO languages where postverbal objects are base-generated (e.g. English). To summarize, the postverbal objects in Tamil and Korean are either ‘‘moved’’ or detached from the previous clause, and thus they are not in a base-generated position as ‘‘parametrized’’ by the head verb. Considering the code-switched objects in (13) and (14) again, 6
It is well known that GB/minimalism analyzes (overt) movement as induced by the strong feature of a functional head (Chomsky, 1995). TAG treats movement as successive building-up of trees (Kroch, 1989). Less clear is how scrambling, traditionally known as an ‘‘optional movement’’, is analyzed in both frameworks. Chan (2003) assumes Saito and Fukui (1998), where scrambling is analyzed as a subcase of Merge and movement (e.g. topicalization, wh-movement) as a sub-case of Move. According to Saito and Fukui (1998), example (10) has to be scrambling because there is no agreement between the Japanese I and Spec-IP (i.e. subject). The IP is not closed off and subject to another merge operation. Examples (11) and (12) have to be topicalization: Since there is agreement between the English I and Spec-IP (i.e. subject), the IP is closed off and a new phrase, namely, a Topic phrase, has to be projected.
784
B.H.-S. Chan / Lingua 118 (2008) 777–809
their postverbal position, either moved or base-generated, is not licensed by the (Tamil and Korean) verb, hence violating Santorini and Mahootian (1995). In any case, Santorini and Mahootian (1995:11) rule out cases where a verb from a VO language takes a preverbal object. However, these cases can be attested in some language-pairs such as Mandinka–English (Haust and Dittmar, 1997) and Tsotsitaal7–English (Slabbert and Myers-Scotton, 1996). The following are two examples. V from VO language, OV order (15)
i ka i rectify 3PL TAM 2SG rectify ‘‘They usually rectify you.’’ (Mandinka–English, Haust and Dittmar, 1997:88, (6))
(16)
want on Tex laat ons daai (daardie) group join8 because old Tex make 1PL DEM group join ‘‘Because old Tex made us join that group.’’ (Tsotsitaal–English, Slabbert and Myers-Scotton, 1996:332, (3))
In the following example, a complement PP from Tsotsitaal appears preverbally, although the English verb would have licensed a postverbal complement. (17)
ek het met daai grootman ge-meet I AUX with DEM man PAST-PART-meet ‘‘I met with that man.’’ (Tsotsitaal–English, Slabbert and Myers-Scotton, 1996:332, (4))
There is yet another body of data where a verb from a VO language apparently takes preverbal objects (code-switched as in (18) and (19) or otherwise as in (20)–(23)), thus contradicting Mahootian (1993), Nishimura (1997) and MacSwan (1999). These data are pervasive in different language-pairs, involving what are commonly known as the mixed compound verbs.9 The following are some examples: (18)
asoko-e minna dump suru no over-there-to everything dump do PRT ‘‘(They) dump everything overthere.’’ (Japanese–English, Morimoto, 1999:35, (29))
(19)
mula khurcya¯ paint karta¯t boys chairs paint do+TENSE ‘‘Boys paint chairs.’’ (Marathi–English, Joshi, 1985:193, (4b))
7
Tsotsitaal is a variety of Afrikaans spoken in South Africa (Slabbert and Myers-Scotton, 1996). In terms of linear order, there is no code-switching between ‘‘group’’ and ‘‘join’’. However, the fact that the verb (‘‘join’’) takes a code-switched preverbal object (‘‘daai group’’) is against Mahootian (1993), Nishimura (1997) and MacSwan (1999). 9 The mixed compound verbs have been documented in Hindi–English (Kachru, 1978, etc.), Panjabi–English (Romaine, 1995), Tamil–English (Sankoff et al., 1990), Korean–English (Lee, 1991), Japanese–English (Nishimura, 1997), Turkish–Dutch (Backus, 1996), Moroccan Arabic–Dutch (Boumans, 1996) and Hindustani–Dutch (Muysken, 1993). Relevant to the present discussion are the language-pairs involving English, a VO language, and an Asian OV language. (Dutch is not a ‘‘pure’’ OV language—OV order appears in relative clauses but not main clauses.) 8
B.H.-S. Chan / Lingua 118 (2008) 777–809
(20)
kamalaa ne hamaare ghar par chicken taste kiyaa Kamla ERG our house at chicken taste did ‘‘Kamla tasted chicken at our house.’’ (Hindi–English, Pandit, 1986:106, (71))
(21)
vsi Panjabi learn kerni a we Panjabi learn do ‘‘We want to learn Panjabi.’’ (Panjabi–English, Romaine, 1995:140, (8))
(22)
anta car-ei drive paNNanum that car-ACC drive do + must ‘‘We must drive that car.’’ (Tamil–English, Sankoff et al., 1990:80, (8))
(23)
Too much money-ruˇl spend-haesso too much money-ACC spend-do+PAST ‘‘(He) spent too much money.’’ (English–Korean, Nishimura and Yoon, 1998:125, (4b))
785
The traditional understanding of these mixed compound verbs is that, as the name suggests, an ‘‘alien’’ verb, so-called the ‘‘host’’, forms a compound verb with the light verb, and a ‘‘mixed compound verb’’ behaves as a single verb in the language of the light verb (Kachru, 1978). Since these ‘‘mixed compound verbs’’ have been relexicalized or nativized into the language of the light verb, in (18)–(23), the preverbal position of objects is yielded. There are, however, two problems associated with the ‘‘compound’’ analysis: One, there is some evidence that the host and the light verb are two words instead of a compound. As noted in Chan (2003:77–78), the light verb can be separated by the negative marker: (24)
Enzyme jo¯ hai vah . . . reaction ko¯ initiate nahı¯ karta¯ hai, enzyme which is that reaction initiate not does balki rate of reaction ko¯ alter karta¯ hai rather rate of reaction alter does ‘‘An enzyme does not initiate the reaction; rather it alters the rate of reaction.’’ (English–Hindi, Kumar, 1985, 355)
(25)
bacea˜ nu˜ tusı˜ force n3i k3r sakde children ACC you force NEG do ‘‘You can’t force children.’’ (Panjabi–English, Romaine, 1995:140, (10))
The second problem is a theoretical one: The idea that a compound verb is ‘‘nativized’’ to the language of the light verb—and therefore OV order follows in the above examples—implicitly assumes the role of ‘‘that language’’ in determining the syntactic structure of the code-switched sentence. This assumption contradicts The Null Theory because ‘‘nativization’’ or the label of ‘‘language’’ should not be the ultimate explanation of the morphosyntactic structure of codeswitching; instead, it is the general linguistic principles or constraints, such as Agreement, which derive syntactic structures, ‘‘pure’’ or code-switched (MacSwan, 1999, 2000). Questions still abound as to the structure of these ‘‘mixed compound verbs’’ (or ‘‘bilingual verbs’’ in Muysken (2000)): Is the light verb a v/I element selecting a VP (Bhatia and Ritchie (1996)—see (26) below)? Alternatively, is the light verb a V head or N head (as the ‘‘alien’’ verb
786
B.H.-S. Chan / Lingua 118 (2008) 777–809
is sometimes nominalized in some language-pairs—see Muysken (2000, ch. 7) for details) adjoined to the host (see (27) below)?
Muysken (2000, ch. 7) concludes that all of these three structures—light verb as an auxiliary element (26), host as an adjoined V head (27) and host as an adjoined N head (27)—are instantiated in different language-pairs. In any case, for our purposes here, mixed compound verbs present some cases (if not all—where the host may be an N) where verbs from VO languages take preverbal subjects. Explanations of OVorder which resort to ‘‘the nativizing language’’ or ‘‘the nativization of verbs’’ are not satisfactory, if one takes The Null Theory seriously.10 10
As far as I understand it, there is still no definitive answer as to why the light verb has to emerge, thus resulting in mixed compound verbs. What’s interesting about this pattern is that similar structures (i.e.[V V]V) do not exist in the participating OV languages: What these OV languages have are compound verbs in which the light verb is combined with a noun or nominalized verb as ‘‘host’’ (i.e.[N V]V—see a detailed discussion in Muysken, 2000, ch. 7, also MyersScotton, 2002:134–139). In other words, what appears to be ‘‘the matrix language’’—the OV language—does not provide the ‘‘mixed compound verb’’ pattern, and hence this pattern is problematic in terms of The Matrix Language Frame Model, a point which Myers-Scotton (2002:136–137) has acknowledged. Myers-Scotton (2002) observes that mixed compound verbs primarily involve what Greenberg (1966) called ‘‘strict OV languages’’, but she remains undecided as to exactly how this leads to the ‘‘mixed compound verb’’ pattern. Having said all this, it would only be fair to point out that mixed compound verbs apparently pose a problem to The Null Theory as well, since they seem to be a ‘‘new’’ structure which evolves specifically from code-switching. Nonetheless, explanations along the line of The Null Theory (i.e. which make use of general linguistic principles only) are not completely impossible: Morimoto (1999), for instance, has suggested that the presence of the light verb is due to prosodic constraints rather than syntactic ones. In my understanding, The Null Theory does not necessarily prohibit ‘‘new’’ structures arising from code-switching or language contact (such as mixed compound verbs) so long as they do not violate any UG principles, just as UG does not preclude language change in the monolingual domain (Lightfoot, 1999).
B.H.-S. Chan / Lingua 118 (2008) 777–809
787
3.2. Word order of verbs and objects/complements in code-switching In brief, we have examined three different sets of data which involve verbs in codeswitching: (28)
Different patterns of VP in code-switching Dataset A: Head-complement order follows the language of the verb—examples (2) to (9) Dataset B: Objects are either topicalized or scrambled to clause-initial position— examples (10) to (12) Dataset C: Head-complement order follows the language of the complement— examples (13) to (17), and probably the so-called ‘‘mixed compound verbs’’ in (18) to (25) as well.
How do we capture these three datasets while keeping the most economical Null Theory by using existing syntactic apparatus independently posited for monolingual syntax? Notice that in datasets A and B, the verb behaves the same as if it does in monolingual syntax, and the ‘‘Null Theory’’ explanation seems most appealing. Most problematic is dataset C, which apparently is the result of mixing or switching languages. This phenomenon poses an immensely difficult dilemma for the theorizing of code-switching: On one hand, we may appeal to additional notions; for instance, the matrix language. That is, head-complement order is determined by the word-order of the matrix language (e.g. Myers-Scotton, 1993, 2002, etc.). However, this move would incur constraints or notions which apply specifically to code-switching. This kind of theories would hence be less economical than The Null Theory where code-switched and monolingual sentences are governed by the same syntactic constraints (e.g. Mahootian, 1993; MacSwan, 1999; Chan, 2003). On the other hand, it is extremely difficult for The Null Theory to account for the related fact that verbobject order appears to be optionally determined by the verb in code-switching: There is no other independent reason for the suggestion that in monolingual syntax heads optionally determine headcomplement order or, more drastically, verbs are adjuncts. After all, such a phenomenon—that objects appear in positions contrary to the language of the verb—seems to be induced by codeswitching. Chan (2003) appeals to Saito and Fukui’s (1998) theory in which word order is determined by directionality parameters and agreement. The former determines the direction of Merge, whereas the latter determines the direction of Move (supposed to be always leftward) as triggered by functional heads. Furthermore, scrambling is considered a sub-case of Merge, and it is thus governed by the directionality parameters. For instance, Japanese is head-final, and accordingly constituents under VP scramble to the left in order to preserve the verb-final order. Under this framework, dataset A is explained by the directionality parameters. That is, the ‘‘parameter value’’ of the language of the verb is set. Dataset B is jointly explained by the directionality parameters and agreement, as the former account for the direction of scrambling whereas the latter accounts for the direction of movement. Turning to the most problematic dataset C, Chan (2003) proposes that it can be subsumed under a Null Theory perspective too—the crux is that, under this framework, verb-object order is encoded as a parameter value rather than a lexical property (contra MacSwan, 1999; Mahootian, 1993, 1996; Santorini and Mahootian, 1995). That is, the parameter value, VO or OV in this case, is dissociated or separate from specific verbs (no matter from an OV or VO language); in other words, the parameter value is not part of the lexical properties or lemma (in terms of Levelt (1989)) of a
788
B.H.-S. Chan / Lingua 118 (2008) 777–809
particular verb. Then, a bilingual speaker presumably has two values which he/she has access to, that is, VO or OV in this case, although he/she has to choose either one in the majority of cases.11 It is thus possible that the bilingual speaker sets a ‘‘parameter value’’, say, VO or OV in this case, but he/she lexicalizes the verb from an OV or VO language, respectively, resulting in dataset C. One additional assumption is crucial here: word order (VO/OV) is set before a lexical verb is retrieved. This is consistent with our current understanding of the production model as outlined in Levelt (1989): In Levelt’s (1989:238–242) terminology, verb-object order is set through a ‘‘categorial procedure’’ for the predicate, and only afterwards will a lexical verb and the noun (head of object NP/DP)—plus the ‘‘lexical’’ modifiers of this noun—be retrieved.12 The functional elements (e.g. determiners, quantifiers, etc.) are supposed to be accessed separately in another process (‘‘frame-generation’’ in Levelt (1989) or ‘‘inflection’’ in Bock and Levelt, 1994). The issue remains open as to the nature of the conditions or constraints which govern a bilingual’s choice of one parameter value rather than the other (i.e. What are these constraints or conditions? Are they syntactic or sociolinguistic?). It may very well be the case, as suggested by a reviewer, that usually only the grammar of one language is activated in codeswitching in an ongoing conversation. Whereas this account may look all too similar to a ‘‘matrix language’’ account, there are at least, as far as I see it, two major differences between the account outlined above and The Matrix Language Frame Model (Myers-Scotton, 1993, 2002): One, The Matrix Language Frame Model (and indeed other ‘‘matrix language’’ accounts such as Kamwangamalu (1989) or Joshi (1985)) assumes that the matrix language governs the morphosyntactic structure of a sentence (or CP). The account proposed above, however, refers to the selection of a head parameter value only. Therefore, it does not exclude the possibility of other parameters being set according to another language, and hence a codeswitched sentence may have grammatical input from more than one participating language.13 Two, The Matrix Language Frame Model assumes a grammar specific to code-switching, and hence a bilingual always sticks to the rules of the matrix language as a result of bilingual competence. The account proposed above, on the other hand, presupposes a head parameter and a production mechanism which are supposed to be present anyway for monolinguals or bilinguals. A bilingual’s adherence to the grammar of one language is seen as just one of the strategies allowed by the language faculty or UG, not the only grammatical option required by the bilingual’s competence. This strategy may be motivated by sociolinguistic factors (e.g. the speakers are socially associated with that language, e.g. being native speakers, and they have been speaking in essentially that language) or processing reasons (e.g. it is easier to retrieve grammatical rules of one language rather than two languages). As such, this account is congruent with Muysken (2000), who suggests that different code-switching patterns reflect 11 Sometimes, both verbs from VO language and OV language appear and sandwich the object, resulting in the so-called ‘‘portmanteau constructions’’ (Nishimura, 1986; Sankoff et al., 1990). See examples in section 4 below. 12 This explains the following example in which both V and object NP/DP consist of English lexical elements despite OV order: i. ni i ye other language speak. . . COND you TAM other language speak ‘‘If you speak another language. . ..’’ (Mandinka–English, Haust and Dittmar, 1997:88, (11)) 13 The Matrix Language Frame Model actually also allows grammatical input from the less dominant ‘‘embedded language’’, but only under specific conditions, that is, the Embedded Language Islands (i.e. continuous phrases which consist of lexical elements all from the embedded language). See Myers-Scotton (2002) for an update.
B.H.-S. Chan / Lingua 118 (2008) 777–809
789
different strategies where one or both of the participating languages may be retrieved to various extents, namely, alternation, insertion or congruent lexicalization (see details in Muysken (2000)). Given the Minimalist view that linguistic variation may well be encoded in functional categories which vary in feature strength across languages (Chomsky, 1995, ch. 4), one may wonder whether OV/VO order can also be attributed to a functional category rather than a separate parameter, yielding an apparently more economical architecture. In fact, there has been a body of research suggesting that OV/VO order is determined by the functional category of AgrO (e.g. Chomsky, 1995, ch. 3, Zwart, 1997) or v (Chomsky, 1995, ch. 4, 2000; Koizumi, 199514). According to the AgrO theory, the underlying order of VP for all languages is VO/verbcomplement, following the ‘‘Antisymmetry’’ thesis of Kayne (1994). OVorder is yielded with the object raised overtly to Spec-AgrO by a strong ‘‘D-feature’’ in AgrO (Chomsky, 1995, ch. 3):
One problem for the AgrO theory is that the very category of AgrO, which is supposed to license the case of preverbal DP complements, does not explain the distribution of preverbal PP and/or IP complements (supposed to be without case), which is also characteristic of many OV languages (e.g. Japanese—Fukui and Takano, 1998, Hindi—Mohanan, 1994, Tamil— Lehmann, 1993, Turkish—Kornfilt, 1997, Panjabi—Bhatia, 1993 and Dutch—Neeleman, 1997). Another problem is that AgrO does not have morphological realizations in many OV languages (e.g. Japanese, Korean, Hindi, Turkish), which is supposed to be agreement inflections on the verb.15 Applying the AgrO theory to our code-switching data, one may say that the AgrO is strong whenever OV order is yielded whereas AgrO is weak whenever VO order is yielded. It is true that all data are covered, but it is difficult to confirm which language the AgrO element comes from. If AgrO is elusive because it is seldom realized morphologically, the category v is more concrete since it is supposed to be the position for the light verb (Chomsky, 1995, ch. 4, 2000). Nonetheless, the theory that v determines VO/OV order is problematic—That is, OV order is yielded by a strong v attracting the object to Spec-vP, whereas VO order is yielded if v is weak. If so, one expects that the light verb appears between the object and the verb in OV languages (i.e. (30)):
14 Koizumi (1995) also assumes AgrO which is lower than v. However, in his system, VO/OV order is determined by v rather than AgrO since both object and verb are assumed to raise overtly to the AgrO projection. 15 This is not to say that an AgrO-like Infl position is impossible across-the-board—it may well be necessary to account for phenomena like ‘‘past participle agreement’’ in French, where the verb does agree with a fronted object morphologically (Kayne, 1989).
790
B.H.-S. Chan / Lingua 118 (2008) 777–809
This prediction is not borne out. There seems to be no realization of v between object and verb. What has been commonly considered as the light verb appears in a clause-final position, in monolingual sentences (e.g. (31)) or code-switched sentences (e.g. (32)) alike. (31)
ilaa-ne mohan-ko pasand kiyaa Ila ERG Mohan ACC like do Ila liked (approved of) Mohan. (Hindi—Mohanan, 1997:437, (54a))
(32)
kamalaa ne hamaare ghar par chicken taste kiyaa Kamla ERG our house at chicken taste did ‘‘Kamla tasted chicken at our house.’’ (Hindi–English, Pandit, 1986:106, (71)—repeated from (20))
Notice that [DP V v] order—as in examples (31) and (32) above—cannot be derived even though the underlying tree is head-final or OV (see (33) below). In addition, notice that this underlying tree goes against the Antisymmetry thesis (Kayne, 1994), an assumption which has motivated and necessitated a ‘‘movement’’ account of OV structure in the first place:
There is a possibility that the order may result from V moving to a higher I position, but this is unlikely because of two reasons: Firstly, it is not clear what that position would be. It cannot be T(Tense) since it is the light verb v which is inflected for tense. Secondly, given that v is lexicalized, head movement of V to a higher I position across v would violate locality, in particular, The Head Movement Constraint. Even if V could move higher, the result would be an order of [V DP v], still not the required surface order of [DP V v].
B.H.-S. Chan / Lingua 118 (2008) 777–809
791
The remaining option in which [DP V v] can be derived is that v branches right in the underlying tree:
In (34) DP has to move to [Spec-vP] to yield the required surface order [DP V v]. The problem with (34) is that this structure is against Antisymmetry since the v head follows the VP complement. Let us also consider a structure where DP precedes V in its base-generated position inside VP:
The order of [DP V v] is also correctly yielded. However, DP need not move to [Spec- vP] in order to derive the order. In sum, it seems to me that the most economical account is that DP is basegenerated under VP in (35), despite its not being ‘‘antisymmetric’’. In its base-generated position, DP has little bearing with v. Notice that this account is also more economical than (34) which involves movement of DP to yield the required order. An alternative view is that the I-element (i.e. v or T) fixes OV/VO order. There seems to be a high degree of correlation between VO/OV order and I-initial/I-final order in the code-switching data. That is, VO order is predominantly associated with the I-initial order (i.e. I [VP V DP]—e.g. (2)16), and OVorder is predominantly associated with the I-final order (i.e. [VP DP V] I—e.g. (9)17). This correlation parallels the case of pure languages, where VO languages tend to be I-initial and OV languages tend to be I-final. As stated in Greenberg’s Universal 16 (Greenberg, 1966:85), VO languages tend to have [Aux V] order (Aux is I in Principle-and-Parameters Theory) whereas OV languages tend to have [VAux] order. However, such a generalization remains a tendency rather than an absolute: Out of the eight languages surveyed, there is an exception, namely, Guarani, where VO order is coupled with I-final order. Turning our attention to code-switching data again, we also find some exceptions where VO/OV order does not correlate to I-initial/I-final order, respectively: In examples (15) and (17) above, the code-switched sentences show verb-final order but a preverbal I-element, a tense-aspect marker (TAM) or an auxiliary (AUX). 16 17
I take the contracted form of will to be in I position. I assume the light verb ‘‘kare’’ to be in I position.
792
B.H.-S. Chan / Lingua 118 (2008) 777–809
Yet another possibility is that verb-initial/verb-final order follows the language of the I-element. This hypothesis would cover the Mandinka–English example (i.e. (15)) and the Tsotsittal–English example as well (i.e. (17)), assuming that both Mandinka and Tsotsittal are verb-final languages. The problem with this hypothesis, as far as I see it, is primarily a theoretical one: Why does the I-element have a bearing on the internal constituency of VP? One sensible conjecture is that the I-element is the head of the clause IP, and hence it determines the structure of IP (which includes verb-initial/verb-final order). Nevertheless, there seems to be no obvious syntactic mechanisms which guarantee a correlation between I-initial/I-final order and verbinitial/verb-final order in the Principles-and-Parameters account: Verb-object order is determined by a parameter. In the more recent Minimalist Framework, a verb merges with an object before I is merged. In Levelt’s (1989) framework (which Chan (2003) assumes) the I-element is inserted or lexicalized only after VO/OV order is set. Looking at code-switching data, there are instances in which I comes from a VO language but the code-switched VP is OV (see (60) below) and vice versa (see (62) below), defying the generalization that the language of I always determines VO/ OV order.18 One consequence of the parametric approach outlined above is that when code-switching takes place within a VP headed by a ditransitive verb or three-place predicate, the two objects both appear to the right or the left; there is no instance in which one object appears to the left of the verb but the other appears to the right. This is empirically borne out in the data: (36)
V + DP + PP John gave a book to ek larakii John gave a book to a girl ‘‘John gave a book to a girl.’’ (English–Hindi, Pandit, 1990:45, (26))
Indeed, Chan (2003) shows that in most code-switching examples all DPs and PPs (arguments or adjuncts) under VP (excluding the subject) appear on one side of the verb only. This is consistent with Saito and Fukui (1998) who suggest that the head parameter accounts for the word order of both complements and adjuncts which are merged in base-generated positions. See Chan (2003) for more data and discussion. 3.3. Code-switching between nouns and noun complements19 Code-switching between nouns and their complements is relatively rare, though not totally absent. By ‘‘noun complements’’ I assume, in line with standard works as Jackendoff (1977), Radford (1993) and Haegeman and Gue´ron (1999), that they are internal arguments of nouns.
18 Thanks are due to an anonymous reviewer who draws my attention to Zepter (2003), a recent attempt to account for mixed word order under Optimality Theory. Adopting the proposal, however, would entail the adoption of the OT framework with its assumptions of constraint ranking and interaction, which I see as a radical departure from the Principles-and-Parameters framework I have been following so far. In any case, I remain open to the potential of a satisfactory account in terms of OT. 19 In this paper, I take the DP Hypothesis according to which nominal projections are determiner phrases (Abney, 1987 and others): A D head takes an NP complement, and within NP an N may take a PP complement (projecting to N’). Some D heads have a specifier position where phrases may appear (e.g. the genitive marker ‘‘-‘s’’ in English—see below). There may be more than one functional head in the functional domain of N: Q or Num, etc.
B.H.-S. Chan / Lingua 118 (2008) 777–809
793
This treatment makes ‘‘noun complements’’ parallel to objects, which are internal arguments of verbs. The following are some examples of nouns (in bold) and their complements (in italics): (37)
a. b. c. d. e. f. g.
destroy the city/the destruction of the city (Jackendoff, 1977:70) own the house/the owner of the house (Farrell, 1992) That man in a tuxedo is the father of the bridegroom. The manager of that company is a Japanese. She loves the blue eyes of Europeans. Edinburgh is the capital of Scotland. Britain is part of Europe.
In (37a), the noun is a nominalization (i.e. ‘‘destruction’’) which licenses an internal argument as its related predicate (i.e. ‘‘destroy’’). In (37b), the noun is not a nominalization (i.e. ‘‘owner’’), but it is also related to a predicate (‘‘own’’). In (37c), the noun (i.e. ‘‘father’’) is a kinship term which is inherently relational. The related arguments are realized as the subject of the sentence (i.e. ‘‘That man in a tuxedo’’) and the internal argument of the nouns (i.e. ‘‘the bridegroom’’). There may well be other nouns which are inherently relational, for instance, a post term (e.g. ‘‘manager’’ in (37d)), a body-part term (e.g. ‘‘eyes’’ in (37e)) and others (e.g. ‘‘capital’’ in (37f), ‘‘part’’ in (37g)).20 The internal argument of a noun may also be a proposition realized as an IP/CP: (38)
The proposal that Mary should be invited is preposterous. (Haegeman and Gue´ron, 1999:441)
Contrary to the case of objects of verbs, noun complements do not appear immediately adjacent to the head noun. In English, for instance, the noun complements, as the above examples in (37) and (38) show, appear in post-nominal position. However, the preposition ‘‘of’’ is obligatory before DP’s, and so is the complementizer ‘‘that’’ before clauses.21 The preposition ‘‘of’’ is traditionally understood as a case marker, and the complementizer ‘‘that’’ is also inserted for some grammatical reason which is not fully understood at the moment.22 Now, let us assume that there is also a head parameter governing head-complement order in NP, on a par with the account for VP outlined above. Under this assumption, the data in (37) and (38) suggest that this parameter is head-initial for English. Problems arise as noun complements in English also appear in pre-nominal position. (39)
a. b.
The bride’s father (Haegeman and Gue´ron, 1999:413) recent student attacks (= ‘‘recent attacks on students’’—Radford, 1993)
On a closer look, however, there seem a lot more restrictions on the distribution of noun complements in this position: In (39a), the noun complement has to appear before the so-called 20
These classes of complement-taking nouns have been suggested in Tang (1989) for Mandarin Chinese. Haegeman and Gue´ron (1999) observe that the complementizer ‘‘that’’ in the following examples cannot be deleted. (i) The proposal *(that) Mary should be invited in preposterous. (Haegeman and Gue´ron, 1999:441) 22 Bernstein (2001), quoting earlier works by Szabolsci, suggests that the complementizer ‘‘that’’ is necessary to turn an IP into an argument. This, however, entails a separate account as to why ‘‘that’’ can be deleted if a clausal IP complement appears after verb. 21
794
B.H.-S. Chan / Lingua 118 (2008) 777–809
Saxon genitive ‘‘-s’’, a D-element (Abney, 1987; Haegeman and Gue´ron, 1999), but a clausal element cannot appear there: (40)
a. b.
?? The bride father (paraphrased from (39a)) *[(That) Mary should be invited]’s proposal is preposterous
These DP complements may have moved to the specifier position of D from lower positions (Fukui and Speas, 1986; Saito and Fukui, 1998; Radford, 1997). In other words, the pre-nominal position is not the base-generated and ‘‘parametrized’’ position of the noun, but it is attracted by the D-head, namely, the genitive marker ‘‘-’s’’. That the genitive marker may well be a case marker explains why this pre-nominal, ‘‘case-marked’’ position holds a DP but not an IP/CP complement (Lightfoot, 1999).23 As for (39b), Radford (1993) observes that noun complements in this position are restricted to NPs without person or number properties (Radford, 1993): (41)
recent (*the) student(*s) attacks
This noun ‘‘student’’ is not an argument position (recall that we have defined a noun complement as an internal argument of a noun) but arguably part of a compound (i.e. [student attack]N). Furthermore, notice that neither DP nor IP/CP complement can appear immediately before N head. (42)
*[(That) Mary should be invited] proposal is preposterous (paraphrased from (38))
Taking into account all these facts, English does appear to ‘‘parametrize’’ a base-generated post-nominal position for noun complements, which can hold both DP and CP complements. Nonetheless, noun complements which are DPs can raise to [Spec- DP] by means of movement from that position given appropriate trigger (i.e. ‘‘-s’’). This account is consistent with Saito and Fukui (1998) (which I have been assuming) in that movement is always leftwards triggered by functional heads, although the noun complement may branch rightwards in a base-generated position following N head in accordance with the head parameter. This account also maintains a parallelism of VP/IP structure and NP/DP structure as widely assumed in recent literature (Bernstein, 2001; Longobardi, 2001): Similar to objects of verbs, noun complements receive its theta role (e.g. theme) in post-head position (i.e. as determined by head parameter); it may surface in that position or raise to the specifier of a functional head (‘‘-s’’ for noun complements; I for objects of verbs) given the absence of other arguments higher in thematic hierarchy (i.e. possessor and agents in the case of NPs—see Longobardi, 2001). In this connection, the following code-switching data show that the English head nouns do not guarantee that the noun complements are distributed in post-nominal position, the position ‘‘parametrized’’ in English for head-complement order in NP:
23
IP/CP complements apparently do not receive case. In the following example, the preposition fails to assign case to IP/CP, hence the ungrammaticality. i. *The proposal of [(that) Mary should be invited]IP/CP Alternatively, we may say that the preposition does not c-select an IP/CP complement.
B.H.-S. Chan / Lingua 118 (2008) 777–809
(43)
go3 program ho2-ji5 waak6 chemistry leoi4-min6 CL program can draw chemistry in jat1 di1 molecule ge3 structure one CL molecule D structure ‘‘With the (computer) program, (you) can draw some structures of molecules in Chemistry.’’ (Cantonese–English, Chan, 1992)
(44)
joint family ke advantage bahot saare hain joint family D advantage ‘‘The advantges of a joint family are numerous.’’ (English–Hindi, Pandit, 1986:38, (12))
(45)
Religion-uDaya main purpose vantu oru supernatural being-la Religion-GEN main purpose (filler) a supernatural being-LOC oru belief create paNNaratu a belief create do-INF ‘‘The main purpose of religions is to create a belief in a supernatural being.’’ (English–Tamil, Sankoff et al., 1990:96, (61))
795
It appears that the pre-nominal position of these complements is licensed by the genitive markers or D-elements in respective languages. More research needs to be done to ascertain whether the noun complements indeed are moved to [Spec-DP], analogous to the analysis for the Saxon genitive in English, or they are base-generated in [Spec-DP].24 In any case, a Null Theory account would say that noun complements can appear in [Spec-DP] because simply that position licenses them, no matter in monolingual syntax or code-switching, and no matter which language the noun complement comes from. Accordingly, we do not need to appeal to explanations such as ‘‘English elements being assimilated into the syntax of another language’’, which appeal to the labels of ‘‘languages’’ and incur additional notions such as the ‘‘matrix language’’. Examples (43)–(45) above show that the head nouns are lexicalized in English even though the English head-initial NP parameter is not instantiated. Now, let us consider another codeswitching example which looks like the reverse: the English head-initial NP parameter is instantiated but the head noun does not come from English: (46)
24
The caparaasii of his office M.A. pass hai the peon of his office M.A. pass is ‘‘The peon of his office has passed his M.A. (exam.)’’ (English–Hindi, Pandit, 1990:50, (35))
This is possible because the properties of the genitive markers across languages may be different from the Saxon genitive in English. Take a look at the Cantonese–English example (43): The D element ‘‘ge3’’ may license both a DP and an IP in Cantonese: a. [go2 baan1 hok6-sang1]DP ge3 gaa1-zoeng2 DEM CL student D parents ‘‘The parents of those students’’ b. [Bei2 dak1 git3-fan1]IP ge3 siu1-sik1 Peter marry D news ‘‘The news that Peter will get married’’ A more detailed analysis of these facts is beyond the scope of this paper.
796
B.H.-S. Chan / Lingua 118 (2008) 777–809
Hindi nouns only license pre-nominal complements before the genitive marker ‘‘ki/kaa’’; yet, the noun complement in (46), ‘‘(of) his office’’, appears in post-nominal position. According to the present framework, the English parameter value of ‘‘head-initial’’ or ‘‘N-PP/DP’’ is set but a Hindi noun is drawn from the lexicon. Put in another way, the Hindi head noun does not automatically activate the Hindi word order for NP according to which the noun complement has to appear in pre-nominal position before a genitive marker. 4. Code-switching between functional heads and complements This section aims to show that in code-switching the language of functional categories (D, I and C), contrary to the case of lexical categories, does determine the position of the complements. This argument is assumed in both Mahootian (1993, 1996, also Santorini and Mahootian, 1995) and Nishimura (1997). Scattered examples were also given in Mahootian and Santorini (1996) and Nishimura and Yoon (1998). However, these works mainly discuss codeswitching data involving lexical categories. In what follows, I attempt to present a broader spectrum of data to show that the functional categories indeed determine the order of their codeswitched complements. Functional categories are thus different from lexical categories, as the latter do not always determine the position of their code-switched complements. The exact differences between functional categories and lexical categories are then discussed. 4.1. Code-switching between D and its NP complements The category D has been standardly identified with articles which express definiteness of its complement NP (Longobardi, 1994; Giusti, 1997). In this light, several language-pairs have been documented in which the participating languages have contrasting order of D and NP, namely, AdaEme-English, Wolof–English and Mandinka–English. When the articles in AdaEme, Wolof and Mandinka are code-switched with an English NP, these post-nominal D-elements remain in post-nominal position. The following are just some examples from the respective corpora: D from D-final languages, NP from D-initial languages (47)
e wo green dress ko he/she-PAST TONE wear green dress D/ART ‘‘He/she wore a green dress.’’ (AdaEme-English, Nartey, 1982:187, (6))
(48)
fexeel ba nekk ci teˆte de liste bi rek try-IMP until be P head of the list DEF ADV ‘‘Try to be only at the head of the list.’’ (Wolof–French, Poplack and Meechan, 1995:215, (23))
(49)
court officer-bi court officer-DEF ‘‘the court officer.’’ (English–Wolof, Haust, 1995:69, (28))
(50)
reason oo reason DEF ‘‘the reason’’ (English–Mandinka, Haust, 1995:48, (13))
B.H.-S. Chan / Lingua 118 (2008) 777–809
797
In these corpora, there are no examples in which the English articles appear. It is assumed that the absence of the English articles represents a gap in the data rather than a violation to the present theory, namely, the language of a functional head determines the position of its complements. What really violates the present theory is the appearance of English articles in post-nominal position. Why then are there no examples with English articles preceding codeswitched NP’s? Chan (2003) suggests that the functional categories from the speakers’ L2 are more difficult to access and hence appear infrequently. It may also be the case that in these cases the bilinguals always activate the functional elements from one language, usually L1 of the speakers, a phenomenon which has been termed the ‘‘Functional Element Effect’’ (Muysken, 2000). English articles appear to be one of these L2 functional categories for the African speakers whose data are collected in the above-mentioned corpora. Are there any language-pairs which, having contrasting orders for D, show examples where the D-element from both languages can be found to alternate with a code-switched NP? Giusti (1995) and Osawa (1998) have suggested that the post-nominal case particles are D-elements binding NP’s, although these case particles do not encode definiteness as articles apparently do. In other words, languages with morphological case, such as Japanese and Korean, are D-final. On this account, language-pairs such as Japanese–English, Korean–English, Hindi–English and Tamil–English consist of two languages with contrasting order for D and NP. The data of these language-pairs do show examples with an English D (i.e. an article) or a case particle alternating with a code-switched NP: D (i.e. article) from D-initial language, NP from D-final language (51)
I command you to do the nokum! I command you to do the recording ‘‘I command you to do the recording!’’ (English–Korean, Choi, 1991:889, (46))
(52)
He is such a nesshinna hita He is such an enthusiastic person ‘‘He is such an enthusiastic person.’’ (English–Japanese, Nishimura, 1985:99, (1))
(53)
raam kii bahan wants to start a naarii jaaritii aandolan. . . Ram GEN sister wants to start a women awakening movement ‘‘Ram’s sister wants to start a women’s awakening movement. . .’’ (Hindi–English, Pandit, 1990:49, (33))
(54)
So the Lord Jesus Christ is going to come back someday to take the maNavaaTTi So the Lord Jesus Christ is going to come back someday to take the bride ‘‘So the Lord Jesus Christ is going to come back someday to take the bride. (English–Tamil, Sankoff et al., 1990:85, (25))
D (i.e. case particle) from D-final language, NP from D-initial language (55)
System-i kantanhae system-NOM simple-is ‘‘(The) system is simple.’’ (English–Korean, Yoon, 1992:439, (8))
798
B.H.-S. Chan / Lingua 118 (2008) 777–809
(56)
One algebra question-o mark-shite. . . one algebra question-ACC mark-do ‘‘(We) mark one algebra question. . .’’ (English–Japanese, Nishimura, 1997:118, (5c))
(57)
Some Englishmen traditional Indian women-ko pasand karaten hain some Englishmen traditional Indian women-ACC like do are ‘‘Some Englishmen like traditional Indian women.’’ (English–Hindi, Pandit, 1990:44, (24))
(58)
naan pooyi paaDuvein Hindi song-ei I go-INF sing Hindi song-ACC ‘‘I will go and sing a Hindi song.’’ (Tamil–English, Sankoff et al., 1990:79, (1))
The above examples clearly show that the language of the D-element determines the surface position of the code-switched NP complement. There are no data in which the pre-nominal D-element (e.g. an English article) appears in post-nominal position, or a post-nominal D-element (e.g. a Japanese case-particle) appears in pre-nominal position. 4.2. Code-switching between I and its VP complements The category I has been subject to various interpretations: Pollock (1989) proposed the ‘‘Split-Infl Hypothesis’’ where I is further split into T(ense) and Agr(eement). Chomsky (1995, ch. 4, 2000) assumes T(ense) and v (light verb). It is assumed here that I refers to functional categories including T, v or other verbal inflections (e.g. Agr, Aspect, etc.) which heads an ‘‘extended projection’’ of the verb in Grimshaw’s (1991) sense. In this light, we see again that language-pairs including Japanese–English, Korean–English, Hindi–English and Tamil–English consist of two languages with contrasting orders for I and VP. The I-position in English are commonly assumed to be preverbal, which is filled by a modal/light verb ‘‘do’’, or which check the tense feature at LF. On the other hand, the I-element in the Asian languages is postverbal. The following examples show that the preverbal English I-element remains in preverbal position, despite the fact that the VP complement comes from an I-final language. I from I-initial language, VP from I-final language (59)
Can I nigeru?25 can I escape ‘‘Can I escape?’’ (English–Japanese, Nishimura and Yoon, 1998:127, (11a))
(60)
Don’t tambae pyo don’t cigarette smoke ‘‘Don’t smoke cigarettes.’’ (English–Korean, Nishimura and Yoon, 1998:127, (11d))
25
Following standard assumptions, the modal verb moves from an underlying I position which precedes the verb.
B.H.-S. Chan / Lingua 118 (2008) 777–809
799
In Choi’s (1991) data of Korean–English, the non-finite I in English—‘‘to’’—precedes a Korean verb/VP (i.e. the object appears to be dropped as in Korean). The following is one example: (61)
I didn’t tell her to je¯nhwa ke¯le¯ I didn’t tell her to phone call ‘‘I didn’t tell her to call (me).’’ (English–Korean, Choi, 1991:888, (38))
On the other hand, the postverbal I-element of the Asian languages remains in postverbal position. I from I-final language, VP from I-initial language (62)
Put the burden on myself-ha-ketunyo put the burden on myself-do ‘‘I put the burden on myself.’’ (English–Korean, Park, 1990:136, (142))
Recall that the some of the ‘‘mixed compound verbs’’ have been reanalyzed as non-compounds where the light verb is an I-element governing a VP (see examples in (24), (25) and structure in (26) in section 3.1 above, also Muysken, 2000, ch. 7). These instances might well be examples in which an I-element (i.e. the light verbs) remains in postverbal position as it is in ‘‘pure’’ languages. In sum, the distribution of a code-switched VP in relation to I (i.e. preceding or following I) must conform to the grammar of the language where the I-element comes from. 4.3. Code-switching between C and its IP complements The category C has been identified with complementizers which take a clausal/IP complement (e.g. ‘‘that’’ and ‘‘for’’ in English). Rizzi (1997) suggests that C, in parallel with I, may be further split up into various elements, including Top(ic) and Foc(us). In this paper, it is assumed that C refers to functional categories which select clauses or IPs. Again, language-pairs such as Japanese–English, Korean–English, Tamil–English and Hindi– English consist of two languages with contrasting orders for C and IP. The English complementizers are clause-initial, whereas the Asian complementizers, which have evolved from verbs of saying, are clause-final. C from C-final languages, IP from C-initial languages (63)
I’m gonna put troops on every highway and stop Japs from coming in I’m gonna put troops on every highway and stop Japs from coming in to iu yonna kiji ga arimasu yo C say like article NOM exist PRT ‘‘There is an article which says something like ‘I’m gonna put troops on every highway and stop Japs from coming in.’’ (English–Japanese, Azuma, 1991:35)
(64)
Even there, I am really lucky-nu collaNum Even there, I am really lucky-that say-must ‘‘Even there, one must say that I am really lucky.’’ (English–Tamil, Sankoff et al., 1990:91, (47))
800
(65)
B.H.-S. Chan / Lingua 118 (2008) 777–809
I am out of town ira malhaseyo I am out of town that tell ‘Tell (him) that I am out of town.’’ (English–Korean, Nishimura and Yoon, 1998:128, (16))
C from C-initial languages, IP from C-final language (66)
Soosuto, the same thing. Except that kore wa motto chiisa na mura na no ‘‘Then, (it’s) the same thing. Except that this is a lot smaller village.’’ (Japanese–English, Nishiumura, 1997:122, (15a))
(67)
I realized that nae hangukmal ajik yakhanko ‘‘I realized that my Korean is still weak.’’ (English–Korean, Nishimura and Yoon, 1998:127, (14b))
(68)
Everybody think that nay-ka yenge-lul cal hanta-ko sayngkakhayyo Everybody think C I-NOM English-ACC well do-C think ‘‘Everybody thinks that I’m a good English speaker.’’ (English–Korean, Park, 1990:103, (88))
(69)
Just because avaa innoru color and race engindratunaale just because they different color and race of-because ‘‘Just because they are of different color and race.’’ (English–Tamil, Sankoff, Poplack and Vanniarajan, 1990:93, (59))26
There are no data where the clause-initial English complementizer appears in clause-final position. Nor are there cases where the clause-final complementizers from the Asian languages appear in clause-initial position. Thus, to conclude, the language of the complementizer determines the surface position of the clausal/IP complement, no matter which language that clausal/IP complement comes from. 4.4. Excursus: are prepositions lexical or functional heads? Prepositions have been a problematic category in the functional/lexical category distinction. On one hand, prepositions pattern with the prototypical functional categories (i.e. determiners, modal auxiliary verbs or complementizers) in being a closed class (Emonds, 1985). On the other hand, prepositions side with the prototypical lexical categories (e.g. nouns and verbs) in assigning theta roles to arguments (Ouhalla, 1991). Looking elsewhere into the domain of speech errors, one still finds that prepositions display a dual character in sharing properties of prototypical lexical categories and prototypical functional/minor categories (Garrett, 1982). In the present framework, lexical categories may not determine the word order of their codeswitched complements, whereas functional categories must determine the word order of their complements. Assuming this is indeed the case, one may wonder how prepositions behave in 26
Examples (68) and (69) are the so-called ‘‘portmanteau’’ sentences with elements ‘‘doubled’’ from the two participating languages. In example (68), I assume that the subordinator ‘‘because/engindratunaale’’ is also a C-element as it selects and IP complement. As pointed out by a reviewer, these instances contain grammar rules and lexical elements from both participating languages, hence defying a ‘‘matrix language’’ account. See Azuma (1997) for some discussion.
B.H.-S. Chan / Lingua 118 (2008) 777–809
801
code-switching: We may look into language-pairs which have different orders for prepositions and their DP complements. If it turns out that the language of the prepositions always determines P-DP/DP-P order, we then have one additional argument that prepositions are functional categories. On the other hand, if the language of the preposition does not always determine P-DP/ DP-P order, we may conclude that prepositions are more likely to be a lexical category. Surveying language-pairs with conflicting P-DP/DP-P orders, such as Korean–English, Japanese–English, it is found that prepositions do always determine the order of their codeswitched complements. The following data is a sample: P from ‘‘prepositional’’ language, complement from ‘‘postpositional’’ language (70)
I feel sorry for halmoni I feel sorry for grandma ‘‘I feel sorry for grandma.’’ (English–Korean, Nishimura and Yoon, 1998:125, (5b))
(71)
She bought it for omiyage she bought it for souvenir ‘‘She bought it for a souvenir.’’ (English–Japanese, Nishimura, 1997:119, (8b))
(72)
Foreigners are not allowed in tirupatii baalaaji ke mandir foreigners are not allowed in Tirupati Balaji GEN temple ‘‘Foreigners are not allowed in Tirupati Balaji’s temple.’’ (English–Hindi, Pandit, 1986:102, (61))
(73)
I was talking to oru orutanooDa peesinDu irunten I was talking to one person talk-PROG be(1SG-PAST) ‘‘I was talking to one person.’’ (English–Tamil, Sankoff et al., 1990:93, (58))27
P from ‘‘postpositional’’ language, complement from ‘‘prepositional’’ language (74)
Native American-tul ilang. . . native American- PL with ‘‘With native Americans. . .’’ (English–Korean, Park, 1990:121, (115))
(75)
One of my friends to hanashite ta noyo one of my friends with talking was PRT ‘‘(I) was talking with one of my friends.’’ (English–Japanese, Azuma, 1991:26–27, (6))
(76)
ram woods par kavitaa likhataa hai Ram woods on poem writes is ‘‘Ram writes poems on woods.’’ (Hindi–English, Pandit, 1986:102, (60))
27
This example is another instance of ‘‘portmanteau’’ constructions. See footnote 26 above.
802
(77)
B.H.-S. Chan / Lingua 118 (2008) 777–809
Only kalaimagaLLataan movies-e patti peece ille only (magazine name)-LOC-only movies-ACC about talk-EMP NEG ‘‘Only in KalaimagaL there is no talk about movies.’’ (English–Tamil, Sankoff et al., 1990:81, (12))
Examples where P-DP/DP-P order does not follow the language of P are not attested. Thus, in the present framework, prepositions behave more like a functional category. In any case, the treatment of P as a functional category still faces the problem of why P is theta-assigning as are other lexical categories. Notice that not all prepositions are theta-assigning, the most obvious examples being ‘‘of’’ in English and ‘‘van’’ in Dutch (Neeleman, 1997). On the other hand, there are other categories which are apparently functional and yet theta-assigning: Subordinators (e.g. ‘‘because’’, ‘‘before’’ in English), for instance, appear to have more semantic content and select propositions as arguments, despite being a ‘‘closed class’’. They stand opposed to other apparent C-elements like conjunctions (e.g. ‘‘and’’, ‘‘but’’) or complementizers (e.g. ‘‘that’’) in that the latter are more like semantic operators (e.g. ‘‘that’’ encodes [-wh/+declarative], ‘‘and’’ encodes ‘‘&’’). Taking into account these facts, one may conjecture that the traditional distinction between functional and lexical categories is not fine enough. The tentative proposal put forward in Chan (2003) is that there might be two distinctions cross-cutting various categories, namely, closed class/open class words and conceptual/featural (or thetaassigning/non-theta-assigning) words. What is more certain is that more research is needed to clarify the related issues. For our purposes here, it is sufficient to note that P patterns with the functional categories in its behavior in code-switching. 4.5. Word order of functional heads and their complements This section aims to account for the differences between the behaviour of lexical categories and that of functional categories in code-switching. Assuming The Null Theory, these differences are attributed to the inherent properties of lexical categories and functional categories, instead of any grammatical constraints specific to code-switching. As we have seen above, the language of functional categories, namely, D, I and C (and presumably P as well) does determine the distribution of their complements, namely, NP, VP and IP, respectively. Functional categories thus differ crucially from lexical categories: Lexical heads may not determine the surface position of code-switched complements; the latter may be moved or appear in based-generated positions contrary to those parametrized by the language of the head. As we have seen in section 3, lexical heads, namely Vand N, may not determine the position of their complements. Assuming The Null Theory, these data are explained by a ‘‘mismatch’’ of ‘‘parameter-setting’’ from one language and ‘‘lemma-selection’’ from another language in the production process. That is, a bilingual may fix a head-complement order from one language (i.e. head-initial/head-final) and yet he/she may lexicalize the head from another language. This proposal implies that a lexical head (Vor N) does not inherently encode a parameter value (headfirst/head-final), apparently contradicting the assumption in Mahootian (1993), Nishimura (1997), MacSwan (1999) and others. Instead, a head parameter value is set before the head is lexicalized according to Levelt (1989). On the other hand, functional categories determine head-complement order. This phenomenon may well be explained by proposing that that functional heads, contrary to lexical heads, do encode a head parameter value (i.e. head-initial/head-final) so that headcomplement order is set when a particular functional head is merged with its complement,
B.H.-S. Chan / Lingua 118 (2008) 777–809
803
whether code-switched or not. In other words, this parameter value is an inherent property of the functional head. A difference between functional heads and lexical heads is consistent with The Null Theory, since in monolingual contexts both types of categories are also supposed to be distinct, either syntactically or semantically (Abney, 1987, etc.) or in terms of psycholinguistics where they are supposed to be accessed separately (Garrett, 1980; Levelt, 1989). One possible concern arising from the proposal is economy, as word order is specified in each word which is a functional category. However, as functional categories form closed classes, it is not too uneconomical to have head-complement order specified in the lexical entries of functional categories. Furthermore, the idea that head-complement order is a lexical property allows cases where different words from the same functional category show different headcomplement orders. This is apparently the case for is the INFL system in Mandinka, as manifest in the following example of Mandinka–English code-switching: (78)
n buka wo understand – noo 1SG TAM that understand – AUX ‘‘I’m not able to understand that.’’ (Mandinka–English, Haust and Dittmar, 1997:87, (4))
Apparently, the TAM marker and AUX are different I-elements in Mandinka. Whereas Mandinka is an ‘‘exotic’’ language which remains obscure in the linguistics literature, a similar case of ‘‘split functional category’’ has been attested in a much more familiar language, namely, Hindi. Bayer (1999) notices that there are two C-elements in Hindi which have different headcomplement orders: The complementizer ‘‘ki’’ is head-initial, preceding its clausal complement. The complementizer ‘‘ke liye’’, on the other hand, is head-final, following its clausal complement. The C-system in Hindi thus defies a single C-initial or C-final parameter. This phenomenon suggests yet another difference between functional categories and lexical categories: Whereas different items within a single category (I or C) may have different orders with respect to its complements, this is not conceivable for lexical categories; that is, we seldom find a language in which some verbs precede objects while other verbs follow objects systematically. One related issue is whether head-final order of functional categories is yielded with their complements moving to the specifier positions of respective functional heads, as envisaged by ‘‘Antisymmetry’’ which stipulates a universal underlying ‘‘specifier-head-complement’’ order (e.g. Haegeman and Gue´ron, 1999; Kayne, 1994). That is, D, I or C may bear a strong feature and hence induce overt movement of the whole complement NP, VP or IP to [Spec-DP], [Spec-IP] or [Spec-CP], resulting in D-final, I-final or C-final orders in monolingual or code-switched sentences. This analysis is not preferable for empirical reasons: Firstly, an NP or IP does not move independently of D-head or C-head, the so-called ‘‘Function Head Constraint’’ (Radford, 2004).28 Secondly, the analysis contradicts canonical movements (i.e. Saxon genitive ‘‘-s’’, subject-raising, wh-movement) in which a functional category (D, I, C) triggers movement of only part but not the whole of its lexical complement (i.e. arguments within NP, VP and IP, respectively). Put it in another way, movement of a constituent is usually triggered by a functional head higher than the one immediately dominating it. Accordingly, what seems more viable and economical is an account in which the ‘‘complement-head’’ order is yielded when a functional 28
This is to be distinguished from another ‘‘Functional Head Constraint’’ (Belazi et al., 1994) in the code-switching literature which prohibits code-switching between a functional head and its lexical complement.
804
B.H.-S. Chan / Lingua 118 (2008) 777–809
head is merged with its complement, in the case of code-switching or monolingual syntax. See Bayer (1999) for more elaborate arguments for a similar position on functional categories based on ‘‘hybrid languages’’. 5. Conclusions This paper aims to show that functional categories (D, I, C) and lexical categories (V, N) behave differently in code-switching in terms of word order properties. More specifically, the language of lexical heads may not determine the word order of their complements (codeswitched or otherwise), but functional categories must do so. Assuming The Null Theory, this difference is not attributed to constraints or principles which apply specifically to codeswitching. Nor is ‘‘The Matrix Language’’ appealed to even though, descriptive speaking, a lexical head may be in a head-complement order characteristic of another language, that is, what appears like ‘‘a matrix’’. Instead, it is assumed that, this difference reflects a fundamental difference between the two categories in the language faculty which prevails in code-switching and monolingual contexts alike. For lexical categories (V, N), head-complement order is set by parameter values. However, in code-switching, a value (head-initial/head-final) may be set and yet the head is drawn from a language other than that associated with this value. This is possible because the lexical head is retrieved after the parameter is set (Levelt, 1989) and a bilingual speaker has access to the lexical head in both languages. In addition, code-switched complements, just as complements in ‘‘pure’’ languages, may appear in a ‘‘non-parametrized’’ position via agreement (with functional heads) or scrambling, in accordance with Saito and Fukui (1998). For functional categories (D, I, C), head-complement order is specified as a lexical property, and the complements of these heads, code-switched or otherwise, seldom move out of these ‘‘parametrized’’ positions.29 Finally, under this framework, prepositions have been found to behave as a functional category, which sheds new light into the perennial puzzle as to whether P is a lexical or functional (see Baker, 2003 and Froud, 2001 for an update discussion). Notes The following are the abbreviations used in the glosses: 1: 2: 3: ACC: ADV: ART: AUX: C: CL: COND: D: 29
first person second person third person accusative case marker adverb article auxiliary verb complementizer classifier conditional determiner
Radford (2004) points out that in English VP complements of certain I-elements (modal verbs, infinitive ‘‘to’’) can move out of VP, thus defying this generalization.
B.H.-S. Chan / Lingua 118 (2008) 777–809
DAT: DEF: DEM: ERG: EMP: FUT: GEN: I: IMP: INF: LOC: MASC: NOM: NEG: P: PART: PASS: PERF: PL: POSS: PROG: PRT: SG: TAM: TOP:
805
dative case marker definite article demonstrative ergative case marker emphatic marker future genitive marker inflection imperative infinitive locative case marker masculine nominative case marker negation preposition/postposition participle passive marker perfective plural possessive progressive particle singular tense-aspect-mood topic marker
Acknowledgement Many thanks are due to an anonymous reviewer for very helpful and inspiring comments. The remaining errors are mine. References Abney, S., 1987. The English Noun Phrase and its Sentential Aspect. Doctoral Dissertation, MIT. Azuma, S., 1991. Processing and Intra-sentential Code-switching. Doctoral Dissertation. University of Texas, Austin. Azuma, S, 1993. The frame-content hypothesis in speech production: evidence from intrasentential code switching. Linguistics 31, 1071–1093. Azuma, S., 1997. Lexical categories and code-switching: a study of Japanese/English code-switching in Japan. The Journal of the Association of Teachers of Japanese 31 (2), 1–24. Backus, A., 1996. Two in One—Bilingual Speech of Turkish Immigrants in the Netherlands. Tilburg University Press. Baker, M., 2003. Lexical Categories. Cambridge University Press, Cambridge. Bayer, J., 1999. Final complementizers in hybrid languages. Journal of Linguistics 35, 233–271. Belazi, H., Rubin, E., Toribio, J., 1994. Code-switching and X-bar theory: the functional head constraint. Linguistic Inquiry 25, 221–237. Bentahila, A., 1995. Review of ‘‘Duelling languages: Grammatical structure in codeswitching’’. Language 71 (1), 135–140. Bernstein, J.B., 2001. The DP hypothesis: identifying clausal properties in the nominal domain. In: Baltin, M., Collins, C. (Eds.), The Handbook of Contemporary Syntactic Theory, pp. 536–561. Bhatia, T.K., 1993. Panjabi. Routledge.
806
B.H.-S. Chan / Lingua 118 (2008) 777–809
Bhatia, T.K., Ritchie, W.C., 1996. Light verbs in code-switched utterances: derivational economy in I-language or incongruence in production? BUCLD20 Proceedings, pp. 52–62. Bhatt, R.M., 1997. Code-switching, constraints and optimal grammars. Lingua 102, 223–251. Bock, K., Levelt, W., 1994. Language production: grammatical encoding. In: Gernsbacher, M.A. (Ed.), Handbook of Psycholinguistics. Academic Press, London, pp. 945–984. Bokamba, E.G., 1989. Are there syntactic constraints on code-mixing? World Englishes 8, 277–292. Boumans, L., 1996. Embedding verbs and Collocations in Moroccan Arabic/Dutch code-switching. In: Eid, M., Parkinson, D. (Eds.), Perspectives on Arabic Linguistics IX: Papers from the Ninth Annual Symposium on Arabic Linguistics. John Benjamins, pp. 45–67. Boussofara-Omar, N., 2003. Revisiting Arabic diglossic switching in light of the MLF model and its sub-models: the 4–M model and the Abstract Level model. Bilingualism: Language and Cognition 6 (1), 33–46. Chan, B.H-S., 1992. Code-mixing in Hong Kong Cantonese-English Bilinguals: Constraints and Processes. Unpublished M.A. dissertation. Department of English, Chinese University of Hong Kong. Chan, B.H-S., 2003. Aspects of the Syntax. Pragmatics and Production of Code-switching—Cantonese and English. Peter Lang, New York. Chan, B.H-S., 2004. Beyond contextualization: code-switching as a textualization cue. Language and Social Psychology 23 (1), 7–27. Chang, S.-J., 1996. Korean. John Benjamins. Choi, J., 1991. Korean-English code-switching: Switch-alpha and Linguistic constraints. Linguistics 29, 877–902. Chomsky, N., 1986a. Knowledge of Language. Praeger, New York. Chomsky, N., 1986b. Barriers. MIT Press, Cambridge, Massachusetts. Chomsky, N., 1995. The Minimalist Program. MIT Press, Cambridge, Massachusetts. Chomsky, N., 2000. Minimalist inquiries: the framework. In: Martin, R., et al. (Eds.), Step by Step. MIT Press, Cambridge, MA, pp. 89–156. Cinque, G., 1994. On the evidence of partial N-movement in the Romance DP. In: Cinque, G., et al. (Eds.), Paths Towards Universal Grammar: Studies in Honor of Richard S. Kayne. Georgetown University Press, Washington, DC, pp. 85– 110. Cinque, G., 1999. Adverbs and Functional Heads. Oxford University Press, Oxford. Duffield, N., 1999. Adjectival modifiers and the Specifier-Adjunct Distinction. In: Adger, D., et al. (Eds.), Specifiers: Minimalist Approaches. Oxford University Press, Oxford, pp. 126–145. Emonds, J., 1985. A Unified Theory of Syntactic Categories. Foris, Dordrecht. Farrell, P., 1992. Null noun complements in English. Linguistic Inquiry 23 (1), 147–156. Froud, K., 2001. Prepositions and the lexical/functional divide: aphasic evidence. Lingua 111, 1–28. Fukui, N., Speas, P., 1986. Specifiers and projections. In: Fukui, N., et al. (Eds.), MIT Working Papers in Linguistics, pp. 128–172. Fukui, N., Takano, Y., 1998. Symmetry in syntax: merge and demerge. Journal of East Asian Linguistics 7, 27–86. Garrett, M.F., 1980. Levels of processing in sentence production. In: Butterworth, B. (Ed.), Language Production. Vol. 1: Speech and Talk. Academic Press, London. Garrett, M.F., 1982. Production of speech: observations from normal and pathological language use. In: Ellis, A. (Ed.), Normality and Pathology in Cognitive Functions. Academic Press, London, pp. 19–76. Garrett, M.F., 1988. Processes in sentence production. In: Newmeyer, F. (Ed.), The Cambridge Linguistic Survey. iii, Cambridge University Press, Cambridge, pp. 66–96. Giusti, G., 1995. A unified structural representation of (abstract) case and article. In: Haider, H., et al. (Eds.), Studies in Comparative Germanic Syntax, pp. 77–93. Giusti, G., 1997. The categorial status of determiners. In: Haegeman, L. (Ed.), The New Comparative Syntax. Addison Wesley Longman, pp. 95–123. Greenberg, J., 1966. Some universals of grammar with particular reference to the order of meaningful elements. In: Greeenberg, J. (Ed.), Universals of Language. second ed. MIT Press. Grimshaw, J., 1991. Extended Projections. Manuscript. Brandeis University. Haegeman, L., Gue´ron, J., 1999. English Grammar. Blackwell, Oxford. Halmari, H., 1997. Government and Code-switching: Explaining American-Finnish. John Benjamins. Haust, D., 1995. Code-switching in Gambia: Eine Sociolinguistische Untersuchung van Mandinka, Wolof und Englisch in Kontakt (in German). Doctoral Dissertation. Hamburg University. Haust, D., Dittmar, N., 1997. Taxonomic or functional models in the description of code-switching? Evidence from Mandinka and Wolof in African contact situations. In: Jacobson, R. (Ed.), Codeswitching Worldwide. Mouton de Gruyter, Berlin, pp. 79–90.
B.H.-S. Chan / Lingua 118 (2008) 777–809
807
Herring, S., 1994. Postverbal position in Tamil. In: Butt, M., et al. (Eds.), Theoretical Perspectives on Word Order in South Asian Languages. CSLI Publications, pp. 119–152. Jackendoff, R., 1977. X-Bar Syntax: A Study of Phrase Structure. MIT Press, Cambridge, MA. Joshi, A.K., 1985. Processing of sentences with intrasentential code-switching. In: Dowty, D.R., Karttunen, L., Zwicky, A. (Eds.), Natural Language Parsing. Cambridge University Press, Cambridge, pp. 190–205. Kachru, B., 1978. Toward structuring code-mixing: an Indian perspective. International Journal of the Sociology of Language 16, 27–46. Kamwangamalu, N.M., 1989. The morphosyntactic aspects of French/English-Bantu code-mixing: evidence for Universal Constraints. In: CLS 25, Parasession on Language in Context. Chicago Linguistic Society, Chicago, IL, pp. 157–170. Kayne, R., 1989. Facets of romance past participle agreement. In: Benica, P. (Ed.), Dialect Variation and the Theory of Grammar. Foris, Dordrecht, pp. 85–103. Kayne, R., 1994. The Antisymmetry of Syntax. MIT Press, Cambridge, Massachusetts. Koizumi, M., 1995. Phrase Structure in Minimalist Syntax. Doctoral Dissertation. MIT. Kornfilt, J., 1997. Turkish. Routledge, London. Kroch, A., 1989. Asymmetries in long-distance extraction in a tree-adjoining grammar. In: Baltin, M.R., Kroch, A.S. (Eds.), Alternative Conceptions in Phrase Structure. Chicago University Press, Chicago, pp. 66–98. Kumar, A., 1985. Hindi-English mixing in scientific discourse. World Englishes 4 (3), 355–358. Lee, M.-H., 1991. A Parametric Approach to Code-mixing. Doctoral Dissertation. State University of New York at Stony Brook. Lehmann, T., 1993. A Grammar of Modern Tamil. Pondicherry Institute of Linguistics and Culture. Levelt, W., 1989. Speaking: From Intention to Articulation. MIT Press, Cambridge, Massachusetts. Lightfoot, D., 1999. The Development of Language. Blackwell. Longobardi, G., 1994. Reference and proper names: a theory of N movement in syntax and logical form. Linguistic Inquiry 25 (4), 609–665. Longobardi, G., 2001. The structure of DPs: some principles, parameters and problems. In: Baltin, M., Collins, C. (Eds.), The Handbook of Contemporary Syntactic Theory, pp. 562–604. MacSwan, J., 1999. A Minimalist Approach to Intra-sentential Code-switching. Garland, New York. MacSwan, J., 2000. The architecture of the bilingual language faculty: evidence from intrasentential code switching. Bilingualism: Language and Cognition 3 (1), 37–54. Mahootian, S., 1993. A Null Theory of Code-switching. Doctoral Dissertation. Northwestern University. Mahootian, S., 1996. A competence model of code-switching. In: Arnold, J., et al. (Eds.), Sociolinguistic Variation: Data, Theory and Analysis—Selected Papers from NVAW 23 at Stanford. CSLI Publications, US, pp. 387–399. Mahootian, S., Santorini, B., 1996. Code-switching and the complement/adjunct distinction. Linguistic Inquiry 27 (3), 464–479. Mohanan, T., 1994. Argument Structure in Hindi. CSLI Publications. Morimoto, Y., 1999. Making words in two languages: a prosodic account of Japanese-English language mixing. International Journal of Bilingualism 3 (1), 23–44. Muysken, P., 1993. Constraints on code-switching: Indic mixed compound verbs. Papers for the Code-switching Summer School, Strasbourg. European Science Foundation, pp. 195–207. Muysken, P., 1997. Code-switching processes: alternation, insertion and congruent lexicalization. In: Putz, M. (Ed.), Language Choices: Conditions, Constraints and Consequences. John Benjamins, pp. 361–380. Muysken, P., 2000. Bilingual Speech: A Typology of Code-mixing. Cambridge University Press, Cambridge. Muysken, P., de Rooij, V., 1995. Review of ‘‘Social motivations for code-switching: Evidence from Africa’’ and ‘‘Duelling languages: Grammatical structure in codeswitching’’. Linguistics 33, 1043–1066. Myers-Scotton, C., 1993. Duelling Languages: Grammatical Structure in Codeswitching. Oxford University Press, Oxford. Myers-Scotton, C., 1997. Afterward. Duelling Languages: Grammatical Structure in Codeswitching, New edition. Oxford University Press, Oxford. Myers-Scotton, C., 2002. Contact Linguistics: Bilingual Encounters and Grammatical Outcomes. Oxford University Press, Oxford. Nartey, J.S., 1982. Code-switching, interference or faddism? Language use among educated Ghanaians. Anthropological Linguistics 24, 183–192. Neeleman, A., 1997. PP complements. Natural Language and Linguistic Theory 15, 89–137. Neeleman, A., Weerman, F., 1999. Flexible Syntax: A Theory of Case and Argument. Kluwer Academic Publishers, Dordrecht.
808
B.H.-S. Chan / Lingua 118 (2008) 777–809
Nishimura, M., 1985. Intrasentential Code-switching in Japanese and English. Doctoral Dissertation, University of Pennsylvania. Nishimura, M., 1986. Intrasentential code-switching: the case of language assignment. In: Vaid, J. (Ed.), Language Processing in Bilinguals: Psycholinguistic and Neuropsychological Perspectives. Lawrence Erlbaum, Hillsdale, NJ, pp. 123–144. Nishimura, M., 1997. Japanese/English Code-switching—Syntax and Pragmatics. Peter Lang, New York. Nishimura, M., Yoon, K.-K., 1998. Head directionality and intrasentential code-switching: a study of Japanese Canadians and Korean Americans’ bilingual speech. In: Silva, D. (Ed.), Japanese/Korean Linguistics, vol. 8. CSLI Publications, pp. 121–130. Osawa, F., 1998. The emergence of the D system and the demise of morphological case in English. UCL Working Papers in Linguistics 10, 467–488. Ouhalla, J., 1991. Functional Categories and Parametric Variation. Routledge, London. Pandit, I., 1986. Hindi-English Code-switching—Mixed Hindi-English. Datta Book Centre, Dehli. Pandit, I., 1990. Grammaticality in code-switching. In: Jacobson, R. (Ed.), Code-Switching as a Worldwide Phenomenon. Peter Lang, New York, pp. 34–69. Park, J., 1990. Korean/English Intrasentential Code-switching: Matrix Language Assignment and Linguistic Constraints. Doctoral Dissertation, University of Illinois at Urbana Champaign. Pfaff, C.W., 1979. Functional and structural constraints on syntactic variation in code-switching. Language 55, 297–318. Pollock, J.Y., 1989. Verb movement, Universal Grammar and the structure of IP. Linguistic Inquiry 20, 365–424. Poplack, S., 1980. Sometimes I’ll start a sentence in English y terminol Espanol: towards a typology of code-switching. Linguistics 18, 581–618. Poplack, S., Meechan, M., 1995. Patterns of language mixture: nominal structure in Wolof-French and Fongbe-French bilingual discourse. In: Milroy, L., Muysken, P. (Eds.), One Speaker: Two Languages: Cross-disciplinary Perspectives. Cambridge University Press, Cambridge, pp. 199–232. Radford, A., 1993. Head-hunting: on the trail of the nominal Janus. In: Corbett, G.G., et al. (Eds.), Heads in Grammatical Theory. Cambridge University Press, pp. 73–113. Radford, A., 1997. Syntactic Theory and the Structure of English: A Minimalist Approach. Cambridge University Press, Cambridge. Radford, A., 2004. Minimalist Syntax: Exploring the Structure of English. Cambridge University Press, Cambridge. Rizzi, L., 1997. The fine structure of the left periphery. In: Haegeman, L. (Ed.), Elements of Grammar. Kluwer Academic Publishers, pp. 281–337. Romaine, S., 1995. Bilingualism, second ed. Blackwell. Saito, M., Fukui, N., 1998. Order in phrase structure and movement. Linguistic Inquiry 29 (3), 439–474. Sankoff, D., Poplack, S., Vanniarajan, S., 1990. The case of the nonce loan in Tamil. Language Variation and Change 2, 71–101. Santorini, B., Mahootian, S., 1995. Codeswitching and the syntactic status of adnominal adjectives. Lingua 96, 1–27. Slabbert, S., Myers-Scotton, C., 1996. The structure of Tsotsitaal and Iscamtho: code-switching and in-group identity in South African townships. Linguistics 34, 317–342. Stenson, N., 1990. Phrase structure, congruence, Government and Irish-English code-switching. In: Hendrick, R. (Ed.), The Syntax of Modern Celtic Languages. Academic Press, San Diego, California, pp. 169–197. Tang, T.-C., 1989. Studies on Chinese Morphology and Syntax 2 (in Chinese). Student Book Co. Ltd., Taiwan. Yoon, K.-K., 1992. New perspectives in intrasentential code-switching: a study of Korean-English. Applied Psycholinguistics 13, 433–449. Zepter, A., 2003. Phrase Structure Directionality: Having a Few Choices. Unpublished doctoral dissertation. Rutgers University. Zwart, J.-W., 1997. The Germanic SOV languages and the universal base hypothesis. In: Haegeman, L. (Ed.), The New Comparative Syntax. Addison Wesley Longman, pp. 246–267.
Further reading Bhatt, R.M., 1995. Code-switching and the functional head constraint. In: Fuller, J.M., et al. (Eds.), Proceedings of ESCOL ‘94, pp. 1–12. Boumans, L., 1998. The Syntax of Codeswitching: Analysing Moroccan Arabic/Dutch Conversations (Studies in Multilingualism 12). Tilburg University Press. Chomsky, N., Beletti, A., Rizzi, L., 2002. On Nature and Language. Cambridge University Press, Cambridge.
B.H.-S. Chan / Lingua 118 (2008) 777–809
809
Clyne, M., 1987. Constraints on code switching: how universal are they? Linguistics 25, 739–764. Fukui, N., 2001. Phrase structure. Baltin, M., Collins, C. (Eds.), The Handbook of Contemporary Syntactic Theory, Chapter 12, pp. 374–406. Koopman, H., Sportiche, D., 1991. The position of subjects. Lingua 85, 211–258. Matsumoto, Y., 1996. A syntactic account of light verb phenomena in Japanese. Journal of East Asian Linguistics 5 (2), 107–149. Muysken, P., 1995. Code-switching and Grammatical theory. In: Milroy, L., Muysken, P. (Eds.), One Speaker, Two Languages: Cross-disciplinary Perspectives. Cambridge University Press, Cambridge, pp. 177–198.