Th= is is the html version of the file Google= automatically generates html versions of documents as we crawl the web.
Tip: To quickly find your search term on this page, press Ctrl+F<= /b> or =E2=8C=98-F (Mac) and use the find bar.
Nonfinite verbs and negotiating bilingualism in codeswitching: Impli= cations for a language production model
Page 1<= /font>
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Bilingualism: Langua= ge and Cognition: page 1 of 15 C
o Cambridge University Press 2013 doi:10.1017/S1366728913000758<= /font>
Nonfinite verbs a= nd negotiating
bilingualism in c= odeswitching:
Implications for = a language
production model<= /b>
Michigan State U= niversity
Midlands Technic= al College, Columbia, SC
(Received: June 21, = 2013; final revision received: November 13, 2013; accepted: November 13, 20= 13)
This paper argues= that a set of codeswitching data has implications for the nature of cognit= ive control in bilingualism and for
models of languag= e production in general. The data discussed are Embedded Language (EL) nonf= inite verbs that occur in
Matrix Language (= ML) frames with appropriate ML inflectional morphology in some codeswitchin= g (CS) corpora. Notably
EL infinitives ar= e involved, as in wo mu con=C3=A7evoir be nu=C9=96e . . . =E2= =80=9Cthey don=E2=80=99t imagine that something . . . =E2=80=9D (from Ewe= =E2=80=93French
CS). The main arg= ument is that such nonfinite forms are selected because they only need chec= king at the lexical-conceptual
level of abstract= structure with the speaker=E2=80=99s intended semantic-pragmatic meaning. = That is, they do not project information
about syntactic a= nd argument structure that is included in the abstract structure of finite = verbs. Nonfinite EL verbs occur
because they bett= er satisfy the speaker=E2=80=99s intentions regarding semantic and pragmati= c meaning than NL finite verbs. The
employment of non= finite EL verbs instead of EL finite verbs partially explains why codeswitc= hing in general and such verb
phrases in partic= ular is perceived as fast and effortless. How one lexical entry (the EL non= finite verb) can take on the
morphosyntactic r= ole of another one (the ML finite verb) implies flexibility in cognitive co= ntrol at an abstract level. It also
implies a certain= malleability at an abstract level in the ML morphosyntactic frame that mak= es it possible to take in a
nonfinite verb in= a slot for a finite verb.
Keywords: codeswitch= ing, language production model, Matrix Language Frame model, nonfinite verb= s
Introduction<= /nobr>
This paper suggests = how structural patterns that occur in
natural codeswitchin= g (CS) demonstrate the fine tuning
and flexibility in t= he bilingual=E2=80=99s cognitive control of both
languages. Such patt= erns also indicate the cross-linguistic
accommodation that m= ust be occurring at an abstract
level. The nature of= the control in many instances of
CS has implications = for models of language production
in general. The spec= ific data discussed here are verbs
=E2=88=97 We would like to thank the three anonymous revie= wers for their
insightful and const= ructive comments. We also thank the BLC editors,
particularly Carmen = Silva Corval=C3=A1n for her invaluable advice and
constant support. We= also appreciate the many authors of the examples
employed that were n= ot our own. Special thanks go to Evershed
Amuzu, Jeanette Sak= el, Leigh Swigart, and Jeanine Treffers-Daller
who uncovered relev= ant examples in their data when we asked for
their help. The aut= hors are listed in alphabetical order. This is a much
revised version of = the co-authored talk titled =E2=80=9CWhat does it cost?
Codeswitching and i= ts implications for language production=E2=80=9D. Myers-
Scotton gave this p= resentation at the 8th International Symposium on
Bilingualism (ISB8)= in Oslo, Norway, in June, 2011. She is grateful
to the Department o= f Linguistics and Scandinavian Studies of the
University of Oslo = for making her presence possible. She also thanks
the audience at Pen= n State University where she presented another
earlier version of = this paper.
Address for corresp= ondence:
Carol Myers-Scotton= , Michigan State University, Department of Linguistics, Germanic, Slavic, A= sian and African Languages, 619 Red Cedar Road,
Room B331Wells Hall= , East Lansing, MI 48824, USA
from one language w= hich occur within the grammatical
frame supplied by a= nother, and inflected with the second
language=E2=80=99s = inflectional morphology. The data differ a
great deal from tho= se studied by those psycholinguists
who increasingly ar= e interested in bilingualism. However,
at least one of the= goals is similar. It is to address
the following quest= ion: What is the nature of the
cognitive control t= hat must underlie rapid switches from
one language to ano= ther? The reason at least some
psycholinguists tur= n to studies of bilingualism is that =E2=80=9Cthe
presence of two lan= guages provides a lens into the way that
cognitive systems i= nteract that cannot otherwise be seen in
research that is re= stricted to speakers of a single language=E2=80=9D
(Kroll, Dussias, Bo= gulski & Valdes Kroff, 2012, p. 231).
Psycholinguistic an= d neurolinguistic studies have made
a good deal of pro= gress in recent times in recognizing that
interaction charac= terizes bilingual speech production. In
1989, Grosjean (19= 89) put it this way: =E2=80=9CThe bilingual
is not two monolin= guals in one person=E2=80=9D. Since then,
these scientists h= ave demonstrated several findings about
bilinguals and the= ir speech. First, there is always parallel
activation of the = speaker=E2=80=99s two languages; that is, speakers
do not switch off = one of the languages, even when it is not
being spoken (e.g.= Dijkstra, 2005). Psycholinguists refer

Page 2
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Carol Myers-Sco= tton and Janice L. Jake
to this as =E2=80= =9Cnon-selective access=E2=80=9D. Second, neuroscience
advances indicate = that a bilingual=E2=80=99s two languages involve
the same neural sy= stems (e.g. Abutalebi & Green, 2007).
Third, not only ca= n the L1 influence the L2, but eye
tracking studies s= how that the L2 can influence the L1
so that certain sy= ntactic patterns in the L1 may change
(Dussias, 2003). F= ourth, the extent to which switching
languages has a co= st in terms of Response Time (RT)
in an experimental= setting remains a debate (e.g. Meuter
& Allport, 199= 9), but at least some studies show no or
reduced cost (e.g.= Costa & Santesteban, 2004; Gullifer,
Kroll & Dussia= s, 2013; Moreno, Federmeier, Kara & Ku-
tas, 2002). Kroll = et al. (2012) offers a current overview and
Kroll and De Groot= (2005) is an edited volume surveying
the complexity of = issues and approaches. Fifth, bilinguals
=E2=80=93 even in = experimental conditions =E2=80=93 rarely are affected by
the context to the= point that they unintentionally switch;
this has led many = researchers to argue that bilinguals have
a degree of inhibi= tory control (e.g. Green, 1998). Finally,
the degree of accu= racy that bilinguals show in selecting
the =E2=80=9Cright= =E2=80=9D language in lexical decision tasks suggests a
plasticity in thei= r cognitive control.
These findings do = not surprise those linguists who
study naturally-oc= curring CS. Although they have not
focused on cogniti= ve issues, these researchers recognize
that switching pat= terns showing elements from both
languages must inv= olve resolving competition between
them. But a surfac= e feature of CS that has long been
very obvious to CS= linguists is that most switching of
languages, even wi= thin a clause, seems to be fast and
effortless, with f= ew or no hesitations. As early as 1980,
Sridhar and Sridha= r (1980) recognized this and suggested
that CS may have a= small cost or none at all. Just recently,
Charles Bwenge, a = Tanzanian linguist who studies CS,
spontaneously char= acterized the bilingual language usage
of the Tanzanian e= lite, including members of the Bunge
(Parliament) in th= is way: =E2=80=9CWhat has never stopped amaz-
ing me is the flue= ncy . . . The educated elite in Tanzania
and the MPs [Membe= rs of Parliament] in particular speak
more fluently and = confidently when they codeswitch than
when =E2=80=98forc= ed=E2=80=99 to speak in English or Swahili only=E2=80=9D
(personal communic= ation, October 9, 2013).
However, even thou= gh they may recognize the
effortless quality= of CS, linguists have not studied the
=E2=80=9Cproductio= n costs=E2=80=9D systematically. In contrast, switching
time has been stud= ied extensively by psycholinguists.
The two topics, la= nguage switching (what psycholinguists
study) and CS (wha= t linguists and sociolinguists study)
differ in many res= pects. Below we survey the main
differences briefl= y.
How CS and lang= uage switching studies differ
First, language sw= itching (LS) studies may study real
linguistic element= s, but they do so under experimental
conditions that in= volve a manipulation of the data. CS
studies almost alw= ays consider strictly naturally-occurring
data =E2=80=93 lan= guage =E2=80=9Cin the wild=E2=80=9D aptly characterized by
Gullberg, Indefrey= & Muysken (2009, p. 37). Sometimes
the data occur in = interviews, but they are even more likely
to occur in ordina= ry conversations.
Second, the nature= of what is being switched is
different. LS stud= ies largely have examined unconnected
words, digits, or = even unconnected sentences. The words
studied are often = concrete nouns. Recently, some LS
studies involve la= rger phrasal units, such as determiners
and their nouns, o= r progressive vs. perfect verb phrases.
But the majority o= f LS studies still focus on single lexical
switches. Linguist= s study CS at every linguistic level,
from its phonetics= to discourse or narrative structure
(see Bullock &= Toribio, 2009, an edited volume).
They frequently an= alyze which language provides the
grammatical frame = for the clause (the Matrix Language, or
ML) and what kinds= of elements from the other language
(the Embedded Lang= uage, or EL) occur. Sociolinguists
study social motiv= ations for CS in their most general
sense (e.g. Auer, = 1998; Gardner-Chloros, 2009; Myers-
Scotton, 1993b, inter alia). For many linguists, the interest
is in patterns in = grammatical structure and this generally
involves studying = intra-clausal CS. Today, there are a few
LS studies that ar= e referred to as code switching studies,
but most CS resear= chers would not agree that the data
studied approximat= e naturally-occurring CS.
Third, for psychol= inguists, a major issue is whether
a switch across la= nguages incurs any processing costs;
such costs are mea= sured in terms of Response Time. The
idea is that RT is= assumed to reflect processing difficulty:
the longer the RT,= the more difficult something is to
process, whether t= o produce or to comprehend. As already
indicated, linguis= ts have rarely noticed any lapses of time
in naturally-occur= ring CS, although some may identify
some switches as = =E2=80=9Cflagged=E2=80=9D (Poplack, 1980).
Fourth, to measure= RT involves laboratory experimen-
tal conditions wit= h many controls. Increasingly today =E2=80=93
although this was = not always a practice earlier =E2=80=93
psycholinguists ev= aluate their subjects carefully to ensure
that they meet the= experiment=E2=80=99s requirements regarding
language proficien= cy; often very fluent bilinguals are the
subjects. In contr= ast, linguists are concerned less with
measuring proficie= ncy; many respondents are judged to be
proficient in both= or all the languages involved. CS itself
does not require e= qual proficiency; speakers with limited
proficiency in the= EL can produce single words in a clause
framed by the ML. = Proficiency is more required in the ML.
Also, the CS data = that are collected normally do not occur
under artificially= -controlled conditions, although some
controls are possi= ble, such as recruiting equal numbers
of males and femal= es or persons in certain age groups.
A fifth difference= is the methodology employed in the
analysis of result= s. Most LS studies use methodologies that

Page 3
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Nonfinite verbs= and codeswitching
promote objectivity= . In many studies, neither the subject
nor the researcher = can affect the data collected, beyond the
research design. Fo= r example, eye-tracking methodologies
are frequently used= to assess comprehension. Participants=E2=80=99
eye movements are r= ecorded while they listen to auditory
messages, but the p= articipants are not aware of what
is being studied. M= ore recent studies measure brain
activity under expe= rimental conditions (Kutas, Moreno
& Wicha, 2009).= In contrast, CS analysts rely almost
entirely on empiric= al, naturally-produced CS data rather
than laboratory-gen= erated data (see Gullberg et al., 2009).
The analyst seeks t= o support an argument regarding CS
structure with rele= vant examples; the data studied may
reflect the researc= h paradigm of the analyst. Often, the
point of discussion= is the relative roles of the different
participating langu= ages and constituent types. Counting
the number of examp= les produced is generally not the
point; demonstratin= g patterns in the data is. Because CS
analysis is almost = always a usage-based approach, some
analysts relate the= ir findings to such syntactic theories as
Cognitive Grammar (= Langacker, 2008; Talmy, 2000, inter
alia). Other= linguists produce grammaticality judgments
to support their ar= guments; some of these linguists relate
their claims to gen= erative linguistic theories such as the
Minimalist Program = (Chomsky, 1995, inter alia).
Goals of the pap= er
We have shown how p= sycholinguistic studies focus on
measures of process= ing costs as evidence of underlying
cognitive control. = However, the focus of this paper is not
on measuring cost, = but rather on HOW COST IS AVOIDED
in one specific typ= e of naturally-occurring CS involving
nonfinite verbs. Th= us, this paper argues that CS studies
provide clear evide= nce of another type of flexibility in
the cognitive syste= m in addition to those results from
psycholinguistic ex= periments.
This discussion fo= llows from the Matrix Language
Frame (MLF) model (= Myers-Scotton, 1993a, 2002a, inter
alia), a mod= el designed to analyze naturally-occurring
CS. In brief, under= the MLF model, the language
with the most criti= cal grammatical contributions to the
bilingual clause is= called the Matrix Language (ML), and
the other participa= ting language, which largely supplies
some content elemen= ts in the clause, is called the
Embedded Language (= EL). The ML construct interacts
with several other = assumptions about language and
language production= . Relevant to the notion of ML is
the Uniform Structu= re Principle (USP), which applies to
all language. It st= ates that any constituent/construction
has a uniform struc= ture that must be observed whenever
it appears (Myers-S= cotton, 2002a, p. 8). The critical
provision of the US= P for CS is that it preferences uniform
structure from the = ML in bilingual constituents.
The data considere= d in this study come from
bilingual clauses = exemplifying CS in naturally-occurring
conversations. EL = verbs in such bilingual clauses as in
(1) and (2) below = are the specific topic in regard to how
production cost in= terms of perceived switching time is
avoided when they = appear. (EL verbs are underlined in
the examples; abbr= eviations used in glosses are listed at
the end of the pap= er.)
(1) Tanzanian S= wahili=E2=80=93English (Bwenge, 2010, p. 31)
sababu ny-ingi
lakini sababu
CL10-LOC reason C= L10-many but reason
moja ni kwamba ma-= benki y-etu
katika ma-zingira<= /nobr>
CL6-NONPST-operate in
sana . . .<= /div>
CL6-difficult very
=E2=80=9CThere are= many reasons but one of them is that our
banks are operatin= g in very unfavorable conditions.=E2=80=9D
(MP, discussing a = bill in the Tanzanian Parliament)
(2) Ewe=E2=80= =93English (Amuzu, 2010, p. 155)
ny=C9=94nu-w=C3=B3= , all the time a, w=C3=B3 a
woman-PL all the time TP = 3PL POT be
wo-respect-= m . . .
=E2=80=9CWomen, th= ey are supposed to respect you all the
time . . . =E2=80= =9D
Rapid switching of= EL verbs, while receiving
inflections from t= he ML, but with little or no perceived
response time cost= s, may be the most unexpected
phenomenon in natu= rally-occurring CS. Moreover, the
various ways an EL= verb is integrated into the
morphosyntax of th= e ML also provide evidence regarding
the nature of this= flexibility and for why production costs
seem low. The dile= mma is how to avoid the expected
production cost of= the assumed need to check the EL verb
for congruence wit= h an ML counterpart and/or with the
requirements of th= e ML frame.
Illustrating CS= : How verbs are different
Of all EL elements= that occur in CS corpora, singly-
occurring nouns ar= e almost always the most frequent;
verbs are less fre= quent. For example, in Pfaff=E2=80=99s (1979)
study of Spanish= =E2=80=93English CS, only 71/932 (7.6%)
switched elements = are verbs; in contrast, 818 (87.8%)
switched elements = are nouns. Similarly, in Poplack (1980),
there are 141 noun= s that are switched, as compared to
13 verbs and 13 ve= rb phrases; switched nouns are 10.84
times more common = than verbs in this Spanish=E2=80=93English
corpus. In Okasha= =E2=80=99s (1999) Arabic=E2=80=93English corpus, in
the Generation 1 d= ata, only 23 English verbs occur,
as compared to 139= singly occurring nouns; nouns are
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Carol Myers-Sco= tton and Janice L. Jake
switched six times= more frequently. In her Generation
2 data, verbs are = even less frequent (8 verbs vs. 838
nouns). Backus (19= 92, 1996) presents two different
pictures. In his 1= 992 data, 108 singly-occurring Dutch
nouns and 21 NPs o= ccur in Turkish grammatical frames;
41 single-occurrin= g verbs and 12 VPs occur. That is,
nouns and NPs are = 2.4 times more frequent than verbs.
However, in the Ba= ckus (1996) corpus, only 15 verbs
occur out of 259 i= ntrasentential switches. Treffers-Daller
(1994, pp. 98=E2= =80=9399) points out that in her Brussels Dutch=E2=80=93
French corpus, =E2= =80=9Cnouns form the largest category of
French borrowings = [single word switches]=E2=80=9D. French nouns
represent 58.4% ou= t of all lexical categories and verbs
represent 8.9%. In= some CS data sets, verbs are much
more frequent. Mye= rs-Scotton (1993a) reports 91 English
verbs in Swahili f= inite clauses; 24.6% of the switches.
Nouns represent 46= .5% of the switches (141 types/174
tokens). In one Ac= holi=E2=80=93English corpus (Myers-Scotton
& Bernsten, 19= 95), 89 English nouns and N-bars occur
(60% of the intras= entential switches) in comparison with
only 48 English ve= rbs. Of these 48 verbs, 10 are gerunds
and occur as NPs (= e.g. subjects of a clause); only 38
(25.5%) English ve= rbs occur in verb positions in Acholi-
framed clauses. No= t all CS analyses report quantitative
data. For example,= Boumans (1998) explains that he does
not, observing tha= t data sources may be so heterogeneous
that quantitative = data would be misleading.
As just illustrate= d, nouns are the most commonly
switched constitue= nt type in intra-sentential CS. Example
(3) illustrates an= EL noun in a noun phrase headed by an
ML determiner. Whi= le nouns receive thematic roles, they
do not supply them= . Thus, when an EL noun occurs in an
ML frame, the only= congruence checking that is needed
is that its semant= ic/pragmatic features make it le mot juste
from the speaker= =E2=80=99s point of view. From a production point
of view, EL nouns = are =E2=80=9Cfree=E2=80=9D in regard to grammatical
(3) Spanish=E2= =80=93English
(Jake, Myers-Scott= on & Gross, 2002, p. 81)
... eso ya<= /div>
this already DET.M.= OBJ put.1PL.PRET
dentro del<= /div>
within of.DET.M.S
=E2=80=9C . . . we= already included this within the time.=E2=80=9D
In some data sets,= larger EL constituents, EL islands, are
also frequent. The= y consist of well-formed phrases in the
EL and generally i= nclude a noun. Such examples of EL
elements can be ex= plained relatively easily because many
EL islands are adj= uncts or formulaic in nature. Example
(4) illustrates a = PP EL island (met een brusselaar), and
there is a formula= ic EL island in (2) above (all the
time). EL i= slands of any type, PP EL islands, NP EL
islands or VP EL i= slands, are not the subject of this
(4) French=E2= =80=93Brussels Dutch
(Treffers-Daller, = 1994, p. 209)
... quand elle est=
mari=C3=A9e [PAUSE= ] met
when 3S.F COP.3S married
een brusselaar<= /b>
=E2=80=9C . . . wh= en she married a Brussels person.=E2=80=9D
EL verbs are anoth= er matter. In contrast to nouns, verbs
carry a good deal = of grammatical information relevant to
the phrase structu= res in which they occur; they assign
thematic roles. Fu= rther, verbs determine how thematic
roles are mapped o= nto predicate=E2=80=93argument structure. For
this reason, one p= rediction is that switching a verb should
carry a higher pro= duction cost than a noun. Earlier, it
was assumed under = the MLF model that all EL verbs
must be checked fo= r congruence with the requirements of
the ML grammatical= frame (see Myers-Scotton, 1993a;
Myers-Scotton &= ; Jake, 1995). However, we now revise
that analysis, as = is discussed below. We now posit that
not all aspects of= the grammatical structure projected by
EL verbs are check= ed for congruence in CS. Similar to
nouns, only the le= xical-conceptual structure of EL verbs
is checked for con= gruence with the ML, as is discussed
Example (5) illust= rates switching of an EL verb
inflected with ML = morphology. The ML supplies
inflections. In th= is example, the ML supplies the EL verb
with elements that= mark it as a finite verb. Such EL verbs
occurring in ML fi= nite slots are the main topic of this
paper. Under the S= ystem Morpheme Principle of the MLF
model, verbal infl= ections such as these must come from
the ML, and this i= s the case with all EL verb forms that
are discussed here= .
(5) Swahili=E2= =80=93English (Myers-Scotton, 1993a, p. 30)
na maji moto,
1S-CONSEC-OBJ.3S-was= h with water hot
S-CONSEC-OBJ.3S-rub wit= h
=E2=80=9CAnd I was= hed it with hot water, and then I rubbed
it with [a] cloth.= =E2=80=9D
Other assumptio= ns about bilingual production:
The Abstract Le= vel model
In addition to the= MLF model and the USP, the analysis
explaining the rel= ative ease of switching verbs assumes
another model, the= Abstract Level model (Myers-Scotton,
2002a; Myers-Scott= on & Jake, 1995). This is a model
of the abstract le= xical structure of entries in the mental
lexicon. According= to the Abstract Level model, the
levels of abstract= structure include lexical-conceptual
structure, predica= te=E2=80=93argument structure, and the level
of morphological r= ealization patterns. Thus, the reason
many EL nouns occu= r is that they need only convey
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Nonfinite verbs= and codeswitching
a speaker=E2=80=99s= intended message at the abstract lexical-
conceptual level; t= hat is, an EL noun need not match
an ML counterpart a= t the levels of predicate=E2=80=93argument
structure and morph= ological realization patterns.
Unlike EL nouns, E= L verbs occur only infrequently in
some language pairs= , although very often in other pairs,
as noted above. Thi= s makes their switching something of
a puzzle. The analy= sis here offers at least a partial answer,
including the concl= usion that EL verbs in naturally-
occurring CS imply = a sense in which the bilingual
cognitive system is= open to modifications. As noted
above, some psychol= inguistic studies imply that the
L1 may be influence= d by the L2; similarly, some CS
researchers disting= uish what is referred to as classic
CS from composite C= S, that is, CS with convergence
phenomena (see e.g.= Amuzu, 2010, published online April
22, 2013; Fredsted,= 2008; King, 2001, inter alia). In this
paper, we only cons= ider classic CS in which all the critical
grammatical element= s in bilingual constituents come from
one language, the M= L.
Codeswitching in= a model of language production
CS may take differe= nt forms, but many bilingual speakers
often switch from o= ne language to another in the same
conversation or eve= n the same clause. There may be
several levels in p= roduction at which one language is
inhibited or not. E= arly on, in language production, at
least in regard to = the grammatical frame of the bilingual
clause, the initial= competition is always resolved in
favor of one of the= languages, called the ML under
the MLF model. This= resolution occurs at the planning
stage (the conceptu= al level in the production model we
follow, a modified = version of Levelt, 1989). A set of
pre-linguistic fact= ors helps determine the language of the
grammatical frame f= or a bilingual clause (the ML) at this
stage, including bo= th sociolinguistic and psycholinguistic
factors. For exampl= e, the sociolinguistic profile of the
macro-community may= influence the selection. But more
critical in specifi= c interactions is what would be the
unmarked choice if = the interaction involving CS were
negotiated monoling= ually. That is, which language is
considered primary = in monolingual speech in interactions
of a given type is = likely to become the ML of a bilingual
clause in those int= eractions. It may be the L1 of some or
all of the speakers= , or it may be a lingua franca that differs
from the L1s of mos= t or all of the speakers (for example,
in Swahili=E2=80=93= English CS, Swahili is a lingua franca for most
speakers). Various = interpersonal factors are weighed when
one language is con= sidered more appropriate or unmarked
for an interaction = (see Myers-Scotton, 1993b, inter alia,
on the notion of th= e unmarked choice). But when speakers
are communicating v= ia CS, and especially when switching
is at length or fre= quent, neither language is the unmarked
choice, although th= e language of the grammatical frame,
the ML, is more un= marked; in fact, CS itself may be the
unmarked choice. W= hat is important is that CS allows the
speaker to have th= e best of two worlds =E2=80=93 the best of the
socio-pragmatic me= ssages that are associated with the use
of elements from m= ore than one language.
Obviously, a major= psycholinguistic factor that
influences selecti= on of the ML is the relative proficiency
of the speakers in= both languages. The reason is that
speakers must be s= ufficiently proficient regarding the
grammatical frame = of whatever linguistic variety is the
ML. Of course this= need not be the standard variety of
the language invol= ved; it is whatever variety is expected
as appropriate in = the interaction. For the variety that is the
EL when speakers a= re engaging in CS, less access to the
grammatical frame = of an EL is needed if all a speaker does
is select nouns or= other content morphemes from the EL
that have enough c= ongruence with the ML that permits
them to fit in ML = constructions.
As noted above, ps= ycholinguists have demonstrated
experimentally in = LS studies that both languages are
active, even when = only one language is cued. CS also
offers empirical e= vidence that both languages are active
whenever a bilingu= al engages in CS. Under the MLF
model, the ML supp= lies the grammatical frame of the
bilingual clause. = A critical characteristic of the ML is that
the whole store of= generalized grammatical knowledge
associated with th= at language is available, too. In bilingual
speech, positing a= n ML assumes that there is an EL, but
not an EL that can= project the same structures as the
ML can. Thus, an E= L form may occur in the ML frame
which is fleshed o= ut by the ML generalized grammatical
knowledge, plus th= e specific grammatical information
that also informs = the rest of the clause and ties the
verb to the clause= . That is, the ML also determines how
such ML morphology= as subject or object inflections,
tense inflections,= and passive or causative morphology
maps predicate=E2= =80=93argument structure onto morpholog-
ical realization p= atterns at the positional (surface)
The ML can inhibit= the EL in various ways while
accepting its part= icipation in other ways. Specifically,
empirical evidence= makes it clear that the EL can
contribute concept= ually-salient lexical elements to a
bilingual clause t= hat satisfy a speaker=E2=80=99s intentions. For
example, the frequ= ently-occurring EL nouns do this.
What naturally-= occurring CS tells linguists
In many CS corpora= , one language continues as the
ML throughout all = clauses in the data set; however, in
some corpora, the = ML does change from one clause to
another, but still= one variety generally dominates as the
ML across the data= set. Obviously, selecting the ML (and
necessarily inhibi= ting the non-ML language for frame-
building functions= ) at the planning stage and sticking with
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Carol Myers-Sco= tton and Janice L. Jake
Figure 1. Producti= on model.
it throughout the = corpus of the bilingual interaction would
seem to be more ef= ficient and would lower production
cost. But switchin= g MLs within a conversation or even
within one speaker= =E2=80=99s turn is evidence that both languages
are generally avai= lable without pauses and that plasticity
characterizes the = language system. Speakers may switch
for a single word = or for longer stretches; the motivations
are largely socio-= pragmatic; the speaker is negotiating
some change in the= =E2=80=9Cfooting=E2=80=9D of the conversation (see
Goffman, 1981, on = footing, Chapter 3). In another sense,
competition does n= ot end with designation of the ML.
The main reason sp= eakers speak at all is to communicate
through meaningful= elements and those elements may
come from either l= anguage.
We emphasize the r= ationale for CS because it takes
advantage of the s= ocio-pragmatic features from two or
more languages; no= t only can switching provide =E2=80=9Cthe right
word=E2=80=9D, but= changing languages can also signal changes in
the speakers=E2=80= =99 relationships or the topic or other features of
the interaction (M= yers-Scotton, 1993b, on the Markedness
model). Thus, beca= use of their content, EL verbs may meet
a speaker=E2=80=99= s desire to encode a particular message better
than an ML counter= part can.
At an abstract lev= el in the production model that
we propose, semant= ic and pragmatic feature bundles
underlying meaning= ful elements that satisfy a speaker=E2=80=99s
communicative inte= ntions are selected at the conceptual
level, along with = the ML (see Figure 1). The bundles
become salient as = language-specific lemmas at the level of
the mental lexicon= . Lemmas enable the mapping of lexical
patterns onto patt= erns of ML grammatical structure in the
sense of Talmy=E2= =80=99s conceptual approach to language. He
states that this a= pproach =E2=80=9Cis concerned with the patterns
in which and the p= rocesses by which conceptual content is
organized in langu= age=E2=80=9D (Talmy, 2000, vol. 2, p. 2). At the
conceptual level, = lemmas underlie semantically-salient
morphemes that con= trast with the structurally-assigned
units that only be= come salient when further mapping
occurs at the leve= l of the formulator. These structurally-
assigned elements = are called late system morphemes
under the 4-M mode= l of morpheme classification (see
Myers-Scotton, 200= 2a; Myers-Scotton & Jake, 2009, inter
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Nonfinite verbs= and codeswitching
alia); they = are not discussed further because they are not
especially relevant= here.
However, what conc= erns us here are not the advantages
of CS, but rather i= ts cost: what makes using two languages
together sufficient= ly economic to warrant engaging in CS?
That is, its advant= ages are not without assumed production
costs, and we turn = now to a cost/benefit analysis of one
aspect of CS. As al= ready indicated, the lemmas that
underlie verbs are = more complex than those for nouns,
even though both no= uns and verbs are content morphemes
in the model of mor= pheme classification followed here, the
4-M model.
As indicated above= , earlier proposals regarding
selection of EL ele= ments assumed congruence checking
must occur, althoug= h there was little consideration of the
production cost. Th= e Blocking Hypothesis in the original
formulation of the = MLF model is that a blocking filter
blocks any EL conte= nt morpheme that is not sufficiently
congruent with the = ML with respect to three levels of
abstraction regardi= ng subcategorization (Myers-Scotton,
1993a, p. 120). Lat= er, Myers-Scotton and Jake (1995)
argued that all EL = forms are checked twice =E2=80=93 once
at the conceptualiz= er and once at the mental lexicon.
That is, at the con= ceptual level, the activation of the EL
form must satisfy t= he semantic/pragmatic features of the
speaker=E2=80=99s i= ntentions. Then, within the level of the mental
lexicon, we suggest= ed that three levels of abstract lexical
structure were chec= ked for congruence. Congruence at
the level of lexica= l-conceptual structure was requisite
for any CS. We sugg= ested congruence at the levels
of predicate=E2=80= =93argument structure and morphological-
realization pattern= s determined the way an EL form might
be integrated into = an ML frame =E2=80=93 as a singly-occurring
morpheme, or perhap= s as a larger EL island. The analysis
considered here rev= ises this assumption; we illustrate
below our argument = that EL verbs are only checked for
lexical-conceptual = structure.
What is the cost= of inflecting EL verbs in an ML
But if CS is viewed= in a cost/benefit analysis, CS has
to make the benefit= it provides worth the cognitive costs
(in utterance time = and other possible ways) of employing
more than one langu= age. No matter what, all EL elements
must undergo some s= ort of congruence check at the
conceptual level fo= r meeting the speaker=E2=80=99s intentions. But
after that, we now = hypothesize that EL forms are =E2=80=9Cfree=E2=80=9D
once they pass this= check, even verbs. This may be one
reason why EL nouns= occur frequently in many data sets.
Still, in some CS c= orpora, such as those with a Bantu
language as the ML,= many EL verbs occur with ML
inflections, as see= n in the Swahili=E2=80=93English example above
(example (5)). The = example below illustrates a clause that
consists almost ent= irely of EL (English) content words, but
with a Bantu ML fr= ame (SiSwati). The English EL verb
discuss is = inflected with the siSwati subject agreement (i-)
and the perfect as= pect suffix (-ile).
(6) siSwati=E2= =80=93English (Kamwangamalu, 1994, p. 75)
Tennis associat= ion i-discuss-ile le-problem ku
meeting yab= o ye kugcina
=E2=80=9CThe tenni= s association discussed that problem at
their last meeting= .=E2=80=9D
In comparison to t= he distribution of EL nouns in mixed
constituents, the = appearance of EL verbs across corpora
reflects variation= in the way they are integrated, variation
of a qualitative a= s well as quantitative nature. That is,
even though EL ver= bs are less frequent than EL nouns,
the grammatical fr= ames in which they occur are more
varied. Of interes= t to a model of language production is
the way this varia= tion provides an answer to the question
of why CS appears = effortless. That is, while EL verbs
are integrated int= o bilingual clauses less frequently and
in more diverse gr= ammatical structures than nouns, CS
involving verbs ap= pears effortless in the same way as the
insertion of singl= y-occurring EL nouns.
The EL is active i= n CS at the level of lexical-conceptual
structure, when an= EL verb is selected as the lemma that
best satisfies the= speaker=E2=80=99s intentions. Thus, the EL verb
brings along its m= eaning, but it is the ML that integrates
it into predicate= =E2=80=93argument and morphological realization
patterns. That is,= how thematic roles are realized in
the syntax is dete= rmined by the ML; similarly, the ML
determines how ten= se and agreement are realized.
However, the non-o= ccurrence of EL verbs with ML
inflections is a h= int that insertion of EL verbs, in those pairs
where it does occu= r, is not straight forward =E2=80=93 at least not in
the same way as in= sertion of EL nouns. This is because the
well-formed occurr= ence of a verb entails production costs
beyond satisfying = a speaker=E2=80=99s intentions regarding desired
semantic/pragmatic= messages. If verbs bring with them
directions to the = formulator in the language production
model, referring t= o the levels of predicate=E2=80=93argument struc-
ture and morpholog= ical realization patterns, the insertion
of an EL verb, pre= sumably, would mean that these levels
have to be checked= for congruence with the directions
underlying the ML = grammatical frame. Because verbs
would appear in a = phrase called by the ML frame, they
would occur with s= ystem morphemes that may co-index
subjects and objec= ts as well as refer to other grammatical
elements outside o= f their immediate grammatical projec-
tion. This implies= that EL verbs would have to be checked
with potential ML = counterparts concerning the specific
directions they co= ntain regarding predicate=E2=80=93argument
structure and morp= hological realization patterns.
Yet, if EL verbs w= ere to undergo extensive checking
that EL nouns do n= ot face, how would we explain that
when EL verbs do o= ccur, the conversations appear to
be effortless and = fast? In brief, the problem is this:
Downloaded: 17 Dec= 2013
IP address: 68.61.= 148.173
Carol Myers-Sco= tton and Janice L. Jake
how to reconcile t= he apparent =E2=80=9Clow cost=E2=80=9D in terms of
response time that= is perceived when EL verbs occur in
some data sets. Th= ere is a confluence of evidence that
points to a soluti= on, including evidence from specific
insertion strategi= es across language pairs, evidence from
ongoing language c= ontact situations, and, most obviously,
evidence from revi= siting and reanalyzing the EL verb
forms occurring in= ML finite frames.
EL nonfinite ve= rbs as the cost/benefit solution
First, our analysi= s looks for the type of EL verb that
does not direct ma= pping of thematic roles onto predicate=E2=80=93
argument structure= or assign morphological realization
patterns. The answ= er is that nonfinite verbs fit this low-
cost scenario. We = propose that nonfinite verbs do not
carry the same cos= ts as finite forms because their levels
of predicate=E2=80= =93argument structure and/or morphological
realizations patte= rns are not salient in structure building.
Their only salient= level of abstract structure is the
level of lexical-c= onceptual structure, as is the case with
nouns. Thus, such = verb forms as infinitives and present
participles can ta= ke ML verbal inflections without creating
any congruence pro= blems regarding the abstract levels
referring to gramm= atical structure. That is, they do not
send directions re= garding EL late system morphemes that
need to be accommo= dated =E2=80=93 and, presumably, inhibited =E2=80=93
in CS, if an EL fi= nite verb were employed.
Because of this, n= onfinite verbs fit the low cost scenario
that can be observ= ed in CS for the phrases that would
contain finite ver= bs in monolingual data. Examples such
as (7) make it ver= y clear that the English verb is not a finite
form because the s= peaker=E2=80=99s intentions call for a past tense
marking, but wa= tch-ed does not occur; the past meaning
comes only from th= e ML inflectional morphology that
watch does = not influence.
(7) Swahili=E2= =80=93English (Myers-Scotton, 1988)
... a-li-vyo
kuwa karibu ku-= faint-i
wakati tu-li-po-watch
filmi y-a
time 1PL-PAST-LOC-watch film CL9-ASSOC
Dracula . . .
Dracula . . .
=E2=80=9C . . . ho= w she was near to fainting when we watched
[the] film of Drac= ula . . . =E2=80=9D
Second, earlier an= alyses (by us and other CS
researchers) did n= ot discuss in detail the nature of EL
verbs. Without a m= odel of abstract lexical structure
with multiple leve= ls that can be split and re-combined
across lexemes and= a compatible production model, such
discussions did no= t seem relevant. Further, the variety of
EL verb forms in C= S initially did not seem amenable to
generalizations. A= closer look at the shared characteristics
of the EL verb for= ms does provide an answer. In the data
considered here, t= he EL verb is always a nonfinite form.
For example, a sur= face English verb could be assumed
to be finite becau= se it appeared in a frame for finite verbs.
No one seemed to c= onsider that these verbs might be
nonfinite verbs th= at the ML frame could accommodate.
Also, no one consi= dered the psycholinguistic costs of
having both the EL= and ML send directions to the
Formulator for act= ivating production of finite verbal
morphology. Furthe= r, if the verbs in question were finite
verbs, at some poi= nt, the levels of predicate=E2=80=93argument
structure and morp= hological realization patterns that
would underlie an = EL finite verb would have to be checked
with those levels = underlying the ML frame. There would
be a psycholinguis= tic cost. Either the EL would be checked
with the ML for su= fficient congruence at all levels (see
Myers-Scotton &= ; Jake, 1995) or the ML would inhibit all
of the checking, n= ot allowing any EL verbs at all.
To this point, the= n, we have argued against the
likelihood that EL= finite verbs would occur in ML frames
with ML inflection= s. We have also argued that EL nonfinite
verbs are better c= andidates to occur in the finite verb slot
in ML frames becau= se they only need to be congruent at
the abstract level= of lexical-conceptual structure, meeting
the requirements o= f the speaker=E2=80=99s intentions. That is, their
other levels of ab= stract structure are not activated. What
remains is to argu= e that those EL verbs that occur in such
examples as we hav= e already given (examples (1) and (2)
and (5)=E2=80=93(7= )) can be identified as nonfinite verbs.
As we already note= d, the nature of these verbs did
not attract our at= tention (and that of other CS researchers)
because we were no= t relating the ease of CS to the abstract
nature of these ve= rbs. However, a second factor may be that
English is often t= he EL in many the CS corpora studied to
date. The problem = with English as the prototype EL is its
bare infinitives a= re identical to most present tense finite
forms. So when an = EL verb occurs, which is it? Of course
the full infinitiv= e is most common in monolingual English
(a bare verb form = plus the morpheme to). But the presence
of to depen= ds on the construction. Consider small clauses
in monolingual Eng= lish, such as go in Let that man go. Or
compare the infini= tive in She likes to play the piano with
this =E2=80=9Cdo= =E2=80=9D construction, Did she play the piano as a child?
Thus, the ambiguit= y of the English infinitive is a reason
to overlook the po= ssibility that the English verb occurring
as an EL form in M= L finite frames is an infinitive.
Of course the natu= re of the infinitive is not necessarily
the same across la= nguages. But there are sources of
ambiguity in other= languages. For example, phonology
of other EL verbs = in other language pairs is potentially
ambiguous. We turn= to French, which is the EL in a
number of data set= s, especially in North Africa and Sub-
Saharan Africa. Fr= ench has more than one conjugation,
but infinitives in= the most common one end in -er.
In conversational = data, it is hard to distinguish the
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Nonfinite verbs= and codeswitching
pronunciation of th= is infinitive from some finite French
forms. However, whe= n CS data contain French EL verbs
from other conjugat= ions, these data provide evidence that
it is the French in= finitive that occurs in the ML slot for
finite verbs, even = for -er verbs.
Exemplifying non= finite EL verbs in ML frames
Examples of CS with= a number of different languages
as the ML illustrat= e the role of French infinitives serving
where finite verbs = would be assumed to occur. Clear ex-
amples come from Ew= e=E2=80=93French CS in Togo in West Africa
(Amuzu, 2011) (see= promettre in (8) and con=C3=A7evoir in (9)
below), and from c= ases in Central Africa, such as Chiluba=E2=80=93
French CS from Con= go (Kamwangamalu, 1987, 1994)
(see rendre in (10) and soumettre in (11)). In addition
to occurring with = agreement morphology, the EL verb in
(10) occurs with t= he reciprocal verbal extension -angana,
essentially coinde= xing the subject and object. Also, there
are examples from = eastern Congo Swahili=E2=80=93French and
Lingala=E2=80=93Fr= ench that make it clear that it is the French EL
infinitive that ap= pears in the ML slot for a finite verb.
(8) Ewe=E2=80= =93French (Amuzu, 2011)
e la
promettre = na w=C3=B2 be . . .
3S POT promise
=E2=80=9CHe will = promise you that . . . =E2=80=9D
(9) Ewe=E2=80= =93French (Amuzu, 2011)
wo mu
con=C3=A7evoir= be
nu=C9=96e<= /div>
3PL NEG.AOR imagin= e
COMP something
tsa e la
s=E2=80=99acco= mpagner
too 3S POT-can 3S= .REFL-accompany
=E2=80=9CThey don= =E2=80=99t imagine that something else can
accompany it.=E2= =80=9D
(10) Chiluba=E2= =80=93French (Kamwangamalu, 1994, p. 75)
ba-rendre-= angana
3PL-AUX.PAST 3PL-gi= ve-RECIP.HAB visits
ya bungi quand= elle etait ici
a lot
when she was here=
=E2=80=9Chey used= to visit each other a lot when she was
(11) Chiluba=E2= =80=93French (Kamwangamalu, 1987, p. 169)
Ndi n-ku-t=C3= =A9l=C3=A9phoner
1S-2S.OBJ-INF-tele= phone COP
ku-ku-informe<= /b>(r) ne
INF-2S.OBJ-inform COMP 1S.NEG-have
pas oubli=C3= =A9 probl=C3=A8me u-wu-vwa
NEG forget problem
2S-1S.OBJ-have.FV<= /nobr>
=E2=80=9CI=E2=80= =99m calling you in order to tell you that I haven=E2=80=99t
forgotten the pro= blem [that] you have submitted
to me.=E2=80=9D
Further, in regard= to a Brussels Dutch=E2=80=93French corpus,
Treffers-Daller (1= 994, p.110) observes that the =E2=80=9C[t]he
French verb is int= egrated into Brussels Dutch in
the following way:= the morpheme -er is attached to the
French verb stem (= for example: bless + -er) and then the
verb stem is conju= gated according to the Dutch rules for
the regular verbs= =E2=80=9D. Elsewhere, Treffers-Daller (personal
communication, Sep= tember 9, 2013) comments: =E2=80=9Cthere
are lots of exampl= es [from the 1994 corpus] where French
verbs are fixed wi= th the -er suffix and then all kinds
of inflections can= be added=E2=80=9D. Note that the -er suffix
marks the infiniti= ve in the most common conjugation in
French. The examp= le below shows that this construction
=E2=80=9Cis still= very common in the Brussels Dutch dialect=E2=80=9D
(Treffers-Daller,= personal communication, September 9,
(12) Brussels = Dutch=E2=80=93French (Treffers-Daller, 2006)
Hij ... pakt zijn= vest
en hij
takes his cardiga= n and strokes
carress-ee= r-t
zijnen hond terwi= jl
[carress-INF-3S] h= is
dog while<= /div>
det kikvors in ze= ijne zak
that frog<= /div>
into his
pocket sneaks
=E2=80=9CHe takes= his cardigan and strokes his dog while
that frog sneaks = into his pocket.=E2=80=9D
Examples (13) thr= ough (15) below make clear that
both ML prefixes = and suffixes, including suffixes that
change the argume= nt structure, do occur with EL nonfinite
verbs in the stru= cture. In (13) and (14), the English verb
(spoil and= select) receive several affixes from Swahili,
and a final vowel= to satisfy Swahili phonotactics. In both,
the -i- = =E2=80=9Capplied=E2=80=9D, with a meaning =E2=80=9Cto=E2=80=9D or =E2=80=9C= for=E2=80=9D, and a
second suffix = -w- =E2=80=9Cpassive=E2=80=9D precedes the final vowel -a.
Both English verb= s also have Swahili subject agreement
morphology. In (1= 5), three English verbs occur in bilingual
clauses with Swah= ili affixes, affect, deal, and backfire. Two
have subject and/= or object=E2=80=93verb agreement morphology;
deal has a= n infinitive prefix.
(13) Swahili= =E2=80=93English (Myers-Scotton, 1993a, p. 120)
vyangu ni-me-s= poil-i-w-a
CL8.thing 1S-PERF-spoil-APPL-PASS-FV
=E2=80=9CMy thing= s were stolen from me.=E2=80=9D
[Literally: =E2= =80=9CMy things, I was spoiled from/because
of my things.=E2= =80=9D]
(14) Swahili= =E2=80=93English (Myers-Scotton, 2013)
kule down = ...
2S-PERF-select-APPL-PASS-FV there down
=E2=80=9CYou had = told me that [it was] you had been
selected [demoted= ] . . . =E2=80=9D
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Carol Myers-Sc= otton and Janice L. Jake
(15) Swahili= =E2=80=93English (Myers-Scotton, 2013)
I hope mim= i si-ta-end-a
I hope me 1S.NEG-<= font style=3D"font-size:9px">FUT-go-FV= place
CL9-FUT-1S.OBJ-affect<= /font>
=E2=80=9CI hope I= won=E2=80=99t be in a place [that] it will affect
[infect] me.=E2= =80=9D
ku-deal na=
1S.NEG-want-NEG.FV INF-deal with
i-ki-backfire<= /b>,
CL2-person CL2.many CL9-COND-backfire
hata ku-kaa Kenya=
1S.NONPST-can-FV even INF<= /font>-stay Kenya
=E2=80=9CI hope I= won=E2=80=99t be in a place that will infect me;
I don=E2=80=99t w= ant to deal with many people. If it
backfires, I can = even stay in Kenya.=E2=80=9D
EL infinitives wi= th special overt ML or EL markings,
another type of n= onfinite verb, seem to be less frequent in
finite verb slots= than the unmarked English (EL) nonfinite
verb in many Bant= u=E2=80=93English corpora, such as those
just illustrated.= In Arabic=E2=80=93Spanish data from Morocco,
Spanish infinitiv= es receive ML verbal inflections, as in
(16) and (17). In= (18) and (19), the French infinitive is
inflected with Ar= abic agreement affixes, as Arabic verbs
are inflected. On= ce the finding that EL infinitives do occur
in the ML finite = slot becomes understood, they may be
recognized elsewh= ere in a more extensive review of other
data sets.=
(16) Moroccan = Arabic=E2=80=93Spanish
(Vincente & Z= iamari, 2008, p. 462)
K=C2=AFa y=C7=9Dm= =C5=A1=C8=8B ytrbajaru k=E1=BD=94ll yawm f-=C7=9Ds-sentro
=E2=80=9Cils vont= travailler chaque jour au centre-ville.=E2=80=9D
[=E2=80=9CThey co= me to work every day in the city center.=E2=80=9D]
(17) Moroccan = Arabic-Spanish
(Vincente & Z= iamari, 2008, p. 462)
f =C7=9Dd-d=C2=AF= ar
COP.PERF.1S in DET-hou= se and
plancarit<= /nobr>
la ropa
planchar[iron].PERF.1S DE= T clothes
all.3S.F [kull-(h= )a]
=E2=80=9CI was at= home (in the house) and I ironed all of her
(18) Moroccan = Arabic=E2=80=93French
(Bentahila & = Davies, 1983, cited in Backus,
1996, p. 215)
Tajbqa jconfro= nter ces idees
=E2=80=9CHe keeps= confronting his ideas=E2=80=9D
(19) Moroccan = Arabic=E2=80=93French
(Bentahila & = Davies, 1992, p. 450)
hija lli tatse= cr=C3=A9terna les hormones ...
=E2=80=9Cit=E2=80= =99s that which secretes [Arabic inflections] for us
the hormones . . = . =E2=80=9D
In her analysis o= f Moset=C3=A9n grammar, Sakel (2004, p. 6)
points out how Sp= anish verbs are integrated into Moset=C3=A9n,
a variety of the = Moset=C3=A9nan family with speakers in the
foothills of the = Bolivian Andes and the Amazon basin:
=E2=80=9CSpanish = verbs are expressed in the infinitive forms and
followed by a Mos= et=C3=A9n ending. In this way, Spanish lexical
elements are trea= ted as non-verbal Moset=C3=A9n elements
appearing with th= e marker [-yi-] that assures Moset=C3=A9n
cross-reference c= an be added=E2=80=9D. In an extensive discussion
of contact phenom= ena, Matras (2009) observes somewhat
similar strategie= s can be found to integrate EL verbs into
other languages.<= /nobr>
(20) Moset=C3= =A9n=E2=80=93Spanish (Sakel, 2004, p. 6)
. . . Por lo meno= s gracias a esos se=C3=B1ores neustros
tatarabuelos =E2= =80=93
. . . j=C3=A4en= =E2=80=99-tom m=C3=A4y=C3=ABdy=C3=AB=E2=80=99 sufrir-yi-in!
how-COM day
=E2=80=9C . . . A= t least thanks to these men great
grandfathers =E2= =80=93 how many days have they suffered.=E2=80=9D
The sentence in (= 21) is another example of the type
of verb form of i= nterest here, with Wolof as the ML.
Here, an EL verb = (French organizer) appears as organize-
waat, with= the Wolof suffix meaning =E2=80=9Crepeated action, do
again=E2=80=9D. S= wigart (personal communication, March, 2002)
states: =E2=80=9C= For example, organiize has a special form of the
suffix beginning = with a w, -waat. If the verb ended in a
consonant, it wou= ld be -aat. So the infinitive ending -e is
clearly part of h= ow the French verb is used in the Wolof
context=E2=80=9D.= The Wolof pronouns are conjugated to indicate
the tense and asp= ect of the verb.
(21) Wolof=E2= =80=93French (Swigart, 1994, p. 187)
War na=C3=B1oo organiize-waat
should 1PL
organize.INF-do.again ball
dinaa la
time [Fr. temp= s] PL.DET 1.FUT 2PS.OBJ
invite [Fr. in= viter]
=E2=80=9CWe have = to organize another dance these days; I=E2=80=99ll
invite you.=E2=80= =9D
Further, another = language pair, Acholi=E2=80=93English in
Uganda, provides = additional confirming data for the
central argument = of this paper, that only nonfinite forms
of EL verbs occur= inflected by the ML. Acholi, a West
Nilotic language,= is a language that must mark aspect, so
an English bare i= nfinitive is an insufficient match. Instead,
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Nonfinite verb= s and codeswitching
another nonfinite = form, the present participle, is activated
because it carries= the feature of aspect that satisfies the
ML=E2=80=99s requi= rements. See i-boarding in (22) and gi-doing
in (23) from Achol= i=E2=80=93English.
(22) Acholi=E2= =80=93English (Myers-Scotton, 2005, p. 12)
Chances me=
accident p= ol ka
chances ASSOC accident many if
i-boarding tax= i
2S-boarding taxi
=E2=80=9C(The) ch= ances of (an) accident (are) many if you
board (a) taxi.= =E2=80=9D
(23) Acholi=E2= =80=93English (Myers-Scotton & Bernsten, 1995)
Gi-doing l= abongo lunch
without lunch
=E2=80=9CThey do = without lunch.=E2=80=9D
In some language = pairs, more than one nonfinite
form of the EL ver= b can be affixed with a =E2=80=9Cnativizer
suffix=E2=80=9D (B= olonyai, 2005) before being integrated into the
syntax of the ML. = For example, in her discussion of
the acquisition of= various CS strategies by a Hungarian=E2=80=93
English speaking c= hild, Bolonyai (2005) discusses how
the Hungarian verb= alizer -ol/=C3=B6l/el is added to integrate
English verbs in C= S. See example (24) below. This is
similar to the Mos= et=C3=A9n strategy for integrating Spanish
verbs. Bolonyai ob= serves that this verbalizer is also
added to other non= finite verb forms of the English verb,
although this is l= ess frequent. See example (25), in which
a present particip= le, socializing, has been activated to
realize aspect, i.= e., the incomplete nature of the event.
Still, this presen= t participle is not inflected directly
in the Hungarian f= rame; the verbalizer is also added
to such EL nonfini= te verb forms. Similarly, in some
Hungarian=E2=80=93= English CS, the past participle of the EL verb
may be selected to= satisfy the speaker=E2=80=99s intentions to
convey completed a= spect (see Bolonyai, 2005, for more
examples).<= /div>
(24) Hungarian= =E2=80=93English (Bolonyai, 2005, p. 320)
egy bike-jack<= /b>-et
order-VBZ-PAST/3SG/INDEF <= /font>a
=E2=80=9CHe order= ed a bike-jack.=E2=80=9D
(25) Hungarian= =E2=80=93English (Bolonyai, 2005, p. 320)
Nem lenne<= /div>
jobb ott-marad-ni= egy
not be/COND better there-stay-INF a
kicsi-t social= izing-ol-ni
little socializin= g-VBZ-IN= F
=E2=80=9CWouldn= =E2=80=99t it be better to stay and socialize a little?=E2=80=9D
We assume that th= e choice of nonfinite EL verb
depends on the req= uirements of the ML grammatical
frame and how ML s= atisfies those requirements. In his
study of incorpora= tion and complex predicates, Haig
(2002) discusses h= ow Turkish participles are the forms
taken in Kurdish f= rames. Clearly, more research is needed
in CS involving l= anguage pairs in which different nonfinite
EL verb forms are= selected to satisfy ML frames.
Modified EL ve= rbs in ML verb conjugations
Another clue that= EL verbs must be nonfinite comes
from strategies f= or assigning EL verbs to an ML class,
usually indicated= by the infinitive form. Although the
speaker=E2=80=99s= intentions activate socio-pragmatic processes
that call for an = EL verb, the frame of the bilingual
clause may call f= or a finite form. For example, Pfaff
(1979, p. 300) re= ports an example from Texas Spanish=E2=80=93
English CS of an = English verb assigned to the -(e)ar
Spanish infinitiv= e class (los hombres me trust-earon
=E2=80=9Cthe men = trusted me=E2=80=9D) in an ML slot receiving ML
finite inflection= s. In this example, the nonfinite English
verb occurs with = additional ML, Spanish, derivational
morphology. Backu= s (1996, pp. 218=E2=80=93222) provides an
overview of =E2= =80=9Cnativized=E2=80=9D EL verbs which illustrates this
phenomenon. In Ge= rman=E2=80=93English CS, there are examples
of English verbs = accommodated into a conjugation of
German, typically= the weak conjugation (ge-verb-t). For
example, in conve= rsations among American university
students and Germ= an exchange students, Fuller (2012)
recorded such exa= mples as this reference to baseball
(26) German=E2= =80=93English (Fuller, 2012)
Die werden ge-= draft-et von einer High School.
=E2=80=9CThey are= drafted from a high school.=E2=80=9D
=E2=80=9CDo=E2= =80=9D constructions
In many other dat= a sets, a different strategy to integrate
EL verbs in mixed= constituents is employed, the =E2=80=9Cdo=E2=80=9D
construction. It = is found in many data sets across
typologically div= erse languages. Still, it is a strategy
related to the an= alysis proposed here because it relies on
EL nonfinite verb= s, too, to express speaker intentions, but
nothing else. Thi= s strategy is employed more frequently
than the insertio= n of a nonfinite form into a finite ML
frame. What this = and other strategies discussed above have
in common is that= they all avoid a second checking of
congruence betwee= n ML and EL verbs. This construction
consists of an ML= verb form that encodes the meaning
=E2=80=9Cdo=E2=80= =9D, but is largely bleached of any meaning; the critical
meaning in the cl= ause comes from a nonfinite verb, often
the infinitive, i= n the predicate called by the =E2=80=9Cdo=E2=80=9D verb. Even
though this const= ruction adds some complexity to the
clause, again, as= pointed out above, there is no congruence
checking between = the EL verb and the ML frame, or
ML and EL finite = verb counterparts, so production cost
associated with l= anguage switching remains minimal. All
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Carol Myers-Sc= otton and Janice L. Jake
necessary ML verb= al inflections occur with the =E2=80=9Cdo=E2=80=9D verb,
not the nonfinite= EL form.
Generally in =E2= =80=9Cdo=E2=80=9D constructions, the EL nonfinite
form is an infini= tive, but in some data sets it may be a
present participl= e as it is in Hausa=E2=80=93English CS (Bickmore,
1985). Examples (= 27) through (29) are typical examples
of =E2=80=9C=E2= =80=98do=E2=80=99 verb constructions=E2=80=9D (with the =E2=80=9Cdo=E2=80= =9D verbs
(27) Chichewa= =E2=80=93English
(Simango, 1995, c= ited in Myers-Scotton,
2002b, p. 217)
Nthawi zambiri, <= b>if you don=E2=80=99t follow
times many=
if you don=E2=80= =99t follow
directions= ,u-ma-chit-a
get lost
directions 2S-FUT-= do-FV get lost
=E2=80=9CMany tim= es, if you don=E2=80=99t follow directions, you get
(28) Turkish= =E2=80=93Dutch (Backus, 1996, p. 238)
Ja<= /div>
maar toch,= millet kijken
yeah but still pe= ople watch-INF
do-PROG-3S<= /div>
=E2=80=9CYeah, bu= t still, everybody is watching you.=E2=80=9D
(29) Tamil=E2= =80=93English (Annamalai, 1989, p. 51)
Avan enne conf= use-paNNiTTaan
=E2=80=9Che confu= sed me=E2=80=9D
Mapping EL the= matic roles onto ML
predicate=E2= =80=93argument structure
Data from Xhosa= =E2=80=93English CS in Africa show that in some
contact situation= s, English verbs are integrated into Bantu
grammatical frame= s by adding of one or more verbal
extensions that o= vertly mark the verb as transitive, as they
are in the ML. Th= at is, the valence of verbs in English as
transitive or int= ransitive is not overt; therefore, languages
such as Xhosa do = not recognize it in terms of mapping
thematic roles on= to predicate=E2=80=93argument structure in.
Simango (2011) re= ports ML verbal extensions added to
English transitiv= e verbs, further transitivizing them, as
in (30) and (31) = below. In (30) a causative extension
is added to an En= glish transitive verb. In (31) an
applied extension= is added to an English transitive
verb. Such exampl= es further show that while conceptual
structure of the = EL verb is available, the EL verb
does not project = and overtly mark predicate=E2=80=93argument
structure. Instea= d the ML projects the predicate=E2=80=93argument
structure and map= s it onto the morphological realization
patterns.<= /div>
(30) isiXhosa= =E2=80=93English (Simango, 2011, p. 131)
Ngubani oyena mnt= u okanye eyona celebrity
=E2=80=9CWhich ot= her person or another celebrity do you
admire?=E2=80=9D<= /nobr>
=E2=80=9Cwhom you= admire=E2=80=9D
(31) isiXhosa= =E2=80=93English (Simango, 2011, p. 132)
u-bukel-e<= /div>
kanye u-visit-= el-e
2S-watch-FV CL9-TV= or
CL10-friends CL10.your<= /font>
=E2=80=9CDo you w= atch TV or visit your friends?=E2=80=9D
Examples such as = these show that, more than lack of
congruence of dir= ectionality of argument structure or case
assignment, it ap= pears that the psycholinguistic processes
involved in codes= witching simply inhibit checking and
also production o= f EL processes beyond the level of
lexical-conceptua= l structure.
In summary, the d= iscussion here offers an explanation
for the EL verbal= structures that are identified in this
paper as occurrin= g in mixed constituents in the bilingual
CS clause. It doe= s this by looking at CS from the
standpoint of abs= tract levels of lexical structure and
production costs,= and this changes our view of where
explanations can = be found. The basic claim is that the
limited structura= l role and the nonfinite nature of EL
verbs that appear= in bilingual clauses prevent costly
competition with = the ML in these clauses. Not just EL
nouns, but also n= onfinite verbs, carry desired meanings
without impinging= on the ML=E2=80=99s grammatical procedures.
That is, nonfinit= e verbs differ from finite verbs, whose
levels of predica= te=E2=80=93argument structure and morphological
realization patte= rns are always available to participate in
the grammatical f= rame. Those levels are not salient in
nonfinite verbs. = Another way of looking at this analysis is
to recognize that= it explains the lack of bilingual clauses
with EL late syst= em morphemes that indicate dependency
relations in gram= matical structure. That is, EL finite verbs
do not occur, and= the role of indicating dependency
relations is rese= rved for the ML (compare the MLF
model and the USP= , the latter of which preferences ML
Initially, explai= ning how EL verbs fill the role of finite
verbs in the ML f= rame was not obvious because CS
language pairs em= ploy many different morphosyntactic
strategies to int= egrate verbs. However, a commonality
of the insertion = of all nonfinite forms and =E2=80=9Cdo=E2=80=9D verb
constructions is = that the various strategies allow for
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Nonfinite verb= s and codeswitching
EL nonfinite verbs= that provide lexical meaning to be
integrated into th= e ML grammatical frame. Either the
form of the EL ver= b is nonfinite or ML morphology is
added to an EL ver= b that allows it to be integrated into the
ML frame. For exam= ple, a Hungarian verbalizer marks
English verbs so t= hey can be inflected in the same way
as Hungarian verbs= . Similarly, adding the Moset=C3=A9n -yi- to
a Spanish infiniti= ve integrates the EL verb sufficiently so
that the ML frame = can inflect it. What may be considered
compromise strateg= ies (e.g. the =E2=80=9Cdo=E2=80=9D construction) appear
to occur where map= ping of predicate=E2=80=93argument structure
and inflections, s= uch as case, would be problematic,
perhaps in terms o= f directionality of mapping. Like =E2=80=9Cdo=E2=80=9D
verbs, Bantu verba= l extensions are a transparent strategy
accommodating the = mapping of EL thematic roles onto
the predicate=E2= =80=93argument structure of the ML.
However, a key to= a common explanation for these
diverse strategies= is recognizing that, at the abstract level,
nonfinite verbs do= not undergo congruence checking
with the requireme= nts of the ML frame. That is, the
thematic roles of = the EL verb are checked at the level
of the conceptuali= zer so that they satisfy the speaker=E2=80=99s
intentions; howeve= r, how the thematic roles are mapped
onto ML predicate= =E2=80=93argument structure, and, further, how
this argument stru= cture is realized in the inflectional
morphology of the = ML, is not checked. This means that
congruence checkin= g is only at the conceptual level for EL
verbs because they= are nonfinite. This argument has been
supported by data = from typologically diverse language
pairs in which EL = verbs do appear in ML frames. Still, the
proposed analysis = does not provide the kind of evidence
that controlled ps= ycholinguistic experiments might offer
about actual time = comparisons between instances of
inflected ML verbs= vs. EL verbs in a frame of ML
inflections, or EL= nouns in mixed NPs vs. EL nonfinite
verbs with ML infl= ections; psycholinguists may be able to
offer more innovat= ive experiments that consider data that
approximate natura= lly-occurring CS.
One consequence o= f our analysis of the insertion of
nonfinite EL verbs= in mixed constituents is that it predicts
limited congruence= checking for all inserted EL lexemes.
For example, we ha= ve suggested that to explain why costs
of all CS seem to = be low, including costs of insertion of
EL verbs in biling= ual constituents, only lexical-conceptual
structure of the E= L lemma is checked. Other features
directing mapping = of structures at the levels of predicate=E2=80=93
argument structure= or morphological realization patterns
are not checked; i= nstead, they follow from ML processes
directing these as= pects of abstract lexical structure. This
results in two phe= nomena. First, EL morphosyntactic
processes are inhi= bited. Second, ML processes provide the
requisite ML syste= m morphemes. One specific hypothesis
is that in CS with= in a clause, only EL verbs which are
nonfinite occur in= an ML frame.
From the psycholi= nguistic point of view, what is
attractive about = this analysis of how EL verbs are
integrated into M= L frames is that it supports claims about
flexibility in th= e cognitive structure. This suggests that
the bilingual can= fine tune his or her cognitive control
of both languages= . In terms of the production model, the
bilingual=E2=80= =99s intentions and cognitive control have automatic
consequences rega= rding how languages participate in the
bilingual clause.= Once lexical-conceptual structure has
activated a verb = lemma, the division of labor between the
ML and the EL aut= omatically determines which language
can inflect the v= erb, free of cost.
Thus, the puzzle = of =E2=80=9Crapid, apparently effortless=E2=80=9D
speech when speak= ers are engaged in CS is partially
solved. The varie= d ways EL verbs can be modified
to satisfy an ML = frame depends on their structure
at an abstract le= vel, but it also depends on a certain
malleable quality= in the abstract nature of the ML frame
itself that allow= s it to accept EL elements. That is, the
morphosyntactic s= tructure of the ML is shown to be
very flexible in = regard to the nature of EL elements it
can accommodate, = advancing our understanding of CS in
general. =E2=80= =9CThe key to any advance is to be able to explain
the complex visib= le by some simple invisible=E2=80=9D (Carroll,
2013, citing Nobe= l physicist Jean Baptiste Perrin).
Abbreviations = in glosses
As much as possib= le, we have preserved the glosses that
the author(s) inc= luded with the example. Where there are
no glosses, the a= uthor(s) did not supply any.
1, 2, 3
first, second, th= ird person
auxiliary<= /div>
causative<= /div>
noun class [+numb= er]
COMPAR comparative
CONSEC consecutive
final vowel
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Carol Myers-Sc= otton and Janice L. Jake
masculine<= /div>
non past
potential<= /div>
PAST, PST past=
reflexive<= /div>
singular, subject=
topic marker
VBL, VBZ verbalizer
Abutalebi, J., &a= mp; Green, D.W. (2007). Bilingual language
production: The n= eurocognition of language representation
and control. N= eurolinguistics, 20, 242=E2=80=93275.
Amuzu, E. (2010).= Composite codeswitching in West Africa: The
case of Ewe=E2= =80=93English codeswitching. Saarbrucken: Lambert
Academic Publishi= ng.
Amuzu, E. (2011).= Ewe contact database (unpublished corpus).
University of Gha= na, Legon.
Amuzu, E. Produci= ng composite codeswitching: The role of the
modularity of lan= guage production. International Journal
of Bilingualis= m, doi:1367006913481139. Published online
by Sage, April 22= , 2013.
Annamalai, E. (19= 89). The language factor in code mixing.
International = Journal of the Sociology of Language, 75,
Auer, P. (ed.) (1= 998). Code-switching in conversation:
Language, inte= raction and identity. London: Routledge.
Backus, A.M. (199= 2). Patterns of language mixing, a
study of Turki= sh=E2=80=93Dutch bilingualism. Weisbaden: Otto
Backus, A. M. (19= 96). Two in one: Bilingual speech in Turkish
immigrants in = The Netherlands. Tilburg: Tilburg University
Bentahila, A., &a= mp; Davies, E. D. (1983). The syntax of Arabic=E2=80=93
English code swit= ching. Lingua, 59, 301=E2=80=93330.
Bentahila, A., &a= mp; Davies, E. D. (1992). Code-switching and
language dominanc= e. In R.J. Harris (ed.), Cognitive
processing in = bilinguals, pp. 443=E2=80=93458. Amsterdam:
Elsevier.<= /div>
Bickmore, L. S. (= 1985). Hausa=E2=80=93English code-switching. MA
thesis, Departmen= t of Linguistics, UCLA.
Bolonyai, A. (200= 5). English verbs in Hungarian/English
codeswitching. In= J. Cohen, K. T. McAlister, K. Rolstad &
J. MacSwan (eds.)= , Proceedings of the 4th International
Symposium on B= ilingualism, pp. 317=E2=80=93327. Somerville,
MA: Cascadilla Pr= ess.
Boumans, L. (1998= ). The syntax of codeswitching: Analyzing
Moroccan Arabi= c/Dutch conversations. Tilburg: Tilburg
University Press.=
Bullock, B., &= ; Toribio, A. J. (eds.) (2009). The Cambridge
handbook of li= nguistic code-switching. Cambridge:
Cambridge Univers= ity Press.
Bwenge, C.M. (201= 0). The tongue between: Swahili &
English in Tan= zanian parliamentary discourse. Munich:
Carroll, S. B. (2= 013, March 11). Solving the puzzles of mimicry
in nature. New= York Times, Section D, p. 3.
Chomsky, N. (1995= ). The Minimalist Program. Cambridge, MA:
The MIT Press.
Costa, A., & = Santesteban, M. (2004). Lexical access in bilingual
speech production= : Evidence from language switching in
highly proficient= bilinguals and L2 learners. Journal of
Memory and Lan= guage, 50, 491=E2=80=93511.
Dijkstra, T. (200= 5). Bilingual visual word recognition and lexical
access. In Kroll = & De Groot (eds.), pp. 178=E2=80=93201.
Dussias, P. E. (2= 003). Syntactic ambiguity in L2 learners: Some
effects of biling= uality on L1 and L2 processing strategies.
Studies in Sec= ond Language Acquisition, 25, 529=E2=80=93557.
Fredsted, E. (200= 8). Convergence in verb phrases. Linguistics,
46, 949=E2= =80=93982.
Fuller, J. (2012)= . German=E2=80=93English corpus (unpublished).
Southern Illinois= University.
Gardner-Chloros, = P. (2009). Code-switching. Cambridge:
Cambridge Univers= ity Press.
Goffman, E. (1981= ). Forms of talk. Philadelphia, PA: University
of Pennsylvania P= ress.
Green, D. W. (199= 8). Mental control of the bilingual lexico-
semantic system. = Bilingualism, Language, and Cognition,
1, 67=E2= =80=9381.
Grosjean, F. (198= 9). Neurolinguists, beware! The bilingual is
not two monolingu= als in one person. Brain & Language,
36, 3=E2= =80=9315.
Gullberg, M., Ind= efrey, P., & Muysken, P. (2009). Research
techniques in the= study of code-switching. In Bullock &
Toribio (eds.), p= p. 21=E2=80=9339.
Gullifer, J., Kro= ll, J. F., & Dussias, P. E. (2013). When language
switching has no = apparent cost: Lexical access in sentence
context. Front= iers in Psychology, 4, 1=E2=80=9313.
Haig, G. (2002). = Noun-plus-verb complex predicates in Kur-
manji Kurdish: Ar= gument-sharing, argument incorpora-
tion, or what? Sprachtypologie und Universalienforschung,
55, 15=E2= =80=9348.
Jake, J. L., Myer= s-Scotton, C., & Gross, S. (2002). Making the
minimalist approa= ch to codeswitching work: Adding the
Matrix Language. = Bilingualism, Language and Cognition,
5, 69=E2= =80=9391.
Kamwangamalu, N. = M. (1987). French/Vernacular code-mixing
in Zaire: Implica= tions for syntactic constraints. In B. Need,
E. Schiller &= A. Bosch (eds.), Papers from the 23rd
Annual Meeting= of the Chicago Linguistic Society, Part
1: The General= Session, pp. 166=E2=80=93180. Chicago IL: Chicago
Linguistic Societ= y.
Kamwangamalu, N. = M. (1994). SiSwati=E2=80=93English code-switch-
ing: The matrix l= anguage principle and linguistic
constraints. S= outh African Journal of African Languages,
14, 70=E2= =80=9377.
Downloaded: 17 De= c 2013
IP address: 68.61= .148.173
Nonfinite verb= s and codeswitching
King, R. (2001). <= i>The lexical basis of grammatical borrowing:
A Prince Edwar= d Island French case study. Amsterdam:
John Benjamins.
Kroll, J. F., &= ; De Groot, A. M. B. (eds.) (2005). Handbook
of bilingualis= m: Psycholinguistic approaches. New York:
Oxford University= Press.
Kroll, J. F., Duss= ias, P. E., Bogulski, C. A., & Valdes Kroff,
J. R. (2012). Jug= gling two languages in one mind: What
bilinguals tell u= s about language processing and its
consequences for = cognition. Psychology of Learning and
Motivation, 56= , 229=E2=80=93262.
Kutas, M., Moreno,= E., & Wicha, N. (2009). Code-switching in
the brain. In Bul= lock & Toribio (eds.), pp. 289=E2=80=93306.
Langacker, R. W. (= 2008). Cognitive grammar: A basic
introduction. New York: Oxford University Press.
Levelt, W. J. M. (= 1989). Speaking, from intention to articulation.
Cambridge, MA: Th= e MIT Press.
Matras, Y. (2009).= Language contact. Cambridge: Cambridge
University Press.=
Meuter, R.F.I., &a= mp; Allport, A. (1999). Bilingual language
switching in nami= ng: Asymmetrical costs of language
selection. Jou= rnal of Memory and Language, 40, 25=E2=80=9340.
Moreno, E.M., Fede= rmeier, K.D., Kara, D., & Kutas, M.
(2002). Switching= languages, switching palabras: An
electrophysiologi= cal study of code switching. Brain &
Language, 80, = 188=E2=80=93207.
Myers-Scotton, C. = M. (1988). Nairobi Swahili=E2=80=93English corpus
(unpublished). Mi= chigan State University.
Myers-Scotton, C. = M. (1993a). Duelling languages: Grammati-
cal structure = in codeswitching. Oxford: Oxford University
Press. [2nd edn. = in 1997, with new =E2=80=9CAfterword=E2=80=9D.]
Myers-Scotton, C. = M. (1993b). Social motivations for
codeswitching,= evidence from Africa. Oxford: Oxford
University Press.=
Myers-Scotton, C. = M. (2002a). Contact linguistics, bilingual
encounters and= grammatical outcomes. Oxford: Oxford
University Press.=
Myers-Scotton, C. = M. (2002b). Frequency and intentionality in
(un)marked choice= s in codeswitching: =E2=80=9CThis is a 24-hour
country=E2=80=9D.= International Journal of Bilingualism, 6, 205=E2=80=93
Myers-Scotton, C. = M. (2005). Embedded language elements in
Acholi/English co= deswitching. Language Matters, 36, 2=E2=80=93
Myers-Scotton, C. = M. (2013). New Nairobi Swahili=E2=80=93English
corpus (unpublish= ed). Michigan State University.
Myers-Scotton, C.= M., & Bernsten, J. G. (1995). Acholi=E2=80=93English
corpus (unpublish= ed). University of South Carolina.
Myers-Scotton, C.= M., & Jake, J. L. (1995). Matching lemmas
in a bilingual la= nguage competence and production model:
Evidence from int= rasentential codeswitching. Linguistics,
33, 981=E2= =80=931024.
Myers-Scotton, C.= M., & Jake, J. L. (2009). Universal structure
in code-switching= and bilingual processing and production.
In Bullock & = Toribio (eds.), pp. 336=E2=80=93357.
Okasha, M. (1999)= . Structural constraints on Arabic=E2=80=93English
codeswitching: Tw= o generations. Ph.D. dissertation,
University of Sou= th Carolina.
Pfaff, C. (1979).= Constraints on language mixing: Intrasentential
code-switching an= d borrowing in Spanish/English.
Language, 55, = 291=E2=80=93318.
Poplack, S. (1980= ). Sometimes I=E2=80=99ll start a sentence in Spanish y
termin=C3=B3 e= n espa=C3=B1ol. Linguistics, 18, 581=E2=80=93616.
Sakel, J. (2004).= A grammar of Moset=C3=A9n. Berlin: Mouton de
Simango, S.R. (19= 95). Chichewa=E2=80=93English corpus (unpub-
lished). Universi= ty of South Carolina.
Simango, S. R. (2= 011). When English meets isiXhosa in the
clause: An explor= ation into the grammar of codeswitching.
Southern Afric= an Linguistics & Applied Language Studies,
29, 127=E2= =80=93134.
Sridhar, S.N., &a= mp; Sridhar, K.K. (1980). The syntax and
psycholinguistics= of bilingual code-mixing. Canadian
Journal of Psy= chology, 34, 407=E2=80=93416.
Swigart, L. (1994= ). Cultural creolisation and language use in
post-colonial Afr= ica: The case of Senegal. Africa, 64, 180=E2=80=93
Talmy, L. (2000).= Toward a cognitive semantics (2 vols.).
Cambridge, MA: Th= e MIT Press.
Treffers-Daller, = J. (1994). Mixing two languages: French=E2=80=93Dutch
contact in a c= omparative perspective. Berlin: Mouton de
Treffers-Daller, = J. (2006). French=E2=80=93Brussels Dutch corpus
(unpublished). Un= iversity of Reading.
Vincente, A., &am= p; Ziamari, K. (2008). L=E2=80=99arabe morocain au
contact du fran= =C3=A7ais et de l=E2=80=99espagnol. In S. Proch=C3=A1zka
& V. Ritt-Ben= mimoun (eds.), Between the Atlantic and
Indian Oceans,= Studies on Contemporary Arabic Dialects :
Proceedings of= the 7th AIDA Conference, Neue Beihefte zur
Wiener Zeitsch= rift f=C3=BCr die Kunde des Morgenlandes, Bd. 4
(Vienna), pp. 457= =E2=80=93469.