Lexical aspects of standard dialect bilingualism

Applied Psycholinguistics 15 (1994), 447-473Printed in the United States of America

Lexical aspects of standard dialectbilingualism

MIRJAM WOUTERSEN, ALBERT COX, BERT WELTENS, andKEES DE BOTUniversity ofNijmegen

ADDRESS FOR CORRESPONDENCEMirjam Woutersen, University of Nijmegen, Erasmusplein 1, 6525 HT Nijmegen

ABSTRACTWeinreich (1953) distinguished three types of bilingualism: the compound, the coordinate,and the subordinative. In this article, we use his partition to describe the effects of a smalltypological distance on the organization of the bilingual lexicon. In order to do so, tworelatively closely related varieties were used, standard Dutch and the dialect of Maastricht.Subjects had to carry out an auditory lexical decision task using the repetition priming para-digm. Stimuli under investigation were cognates and noncognates. There were two age groups(13 and 17 years old) and two language backgrounds (standard Dutch and Maastricht dialect).The results indicated no differences depending on age. With regard to language background,no interlingual repetition priming was found for the dialect speakers. However, in contrastwith earlier findings on visual repetition priming, there were interlingual repetition effects notonly for the cognates, but also for the noncognates when the standard speakers were con-cerned. Therefore, we concluded that, at least in the auditory modality, the dialect speakers inquestion are coordinate bilinguals and the standard speakers are subordinative bilinguals.Finally, it is shown that Weinreich's model in his pure form leads to unexplainable processes inlanguage acquisition. For that reason, his distinctions are incorporated into the lexico-semanticmodel of Levelt (1989).

This article presents the results of a bilingual lexical decision task usingthe repetition priming paradigm. Up until now, research using bilingualrepetition tasks concentrated on more or less related language pairs (e.g.,Dutch-English, English-Spanish). The present experiment is part of alarger research program that investigates the effects of typological distanceon the organization of the bilingual lexicon. In this program, comparisonsare made between language pairs with a large typological distance (Dutch-Turkish, English-Welsh), language pairs with a relatively small typologicaldistance (Dutch-English), and language pairs with a very small typologicaldistance (standard Dutch-Dutch Dialect).

In this article, the effects of a very small typological distance will beexamined. We made use of standard Dutch and the dialect of Maastricht.The Maastricht dialect was chosen because it is typologically more distantfrom the standard language than other Dutch dialects. This means that the

© 1994 Cambridge University Press 0142-7164/94 $5.00 + .00

Applied Psycholinguistics 15:4 448Woutersen et al.: Lexical aspects of standard dialect bilingualism

LI L2 LI L2 LI L2

OO 0 O SIGNIFIEDS

O O O O O - O SIGNIFIERS

COORDINATE COMPOUND SUBORDINATTVE

Figure 1. Bilingual signs according to Weinreich.

two varieties differ not only at the phonological level, but also at the lexicallevel - the level under investigation here.

Maastricht is a city in the south of the Netherlands with a dialect thathas been very well preserved. Its language situation is very suitable forinvestigating the bilingual lexicon, since the dialect is still actively used by alarge proportion of speakers in all generations and in all levels of society.Unlike most other city dialects, it has a relatively high prestige within thespeech community (Miinstermann, 1989; Miinstermann & Hagen, 1986).As a consequence, even people brought up in the standard language learnhow to speak the dialect; it becomes their second language (L2). On theother hand, all people brought up in the Maastricht dialect acquire standardDutch at an early age; this is the language used in education, by the Dutchmedia, and by most official bodies. Consequently, for the native speakersof the dialect, the standard language has the status of L2.

Over the last 40 years, the most commonly cited approach to the problemof lexical organization in the bilingual lexicon has been the compoundversus coordinate distinction (Weinreich, 1953). We use this theory as aframework to explain the repetition effects found with the bilingual speak-ers tested in our experiment.

In order to arrive at a precise understanding of the theory - one that isnot blurred by the interpretations made by other authors - we went back toWeinreich's original work. Weinreich discussed whether the two linguisticsystems present in a bilingual speaker are coexistent or merged into a singlesystem. Although he explicitly mentioned the phonemic and semantic sys-tems, it is not entirely clear which linguistic systems Weinreich was referringto. Of special interest is the attention Weinreich paid to the sign. The signis a Saussurian notion referring to the combination of a unit of expression(the so-called signifier) and a unit of content (also known as the signified).According to Weinreich, bilingual signs can be coordinate, compound, orsubordinative (see Figure 1). A sign is coordinate when it has two signifierslinked to two signifieds, a sign is compound when it has two signifiers with

Applied Psycholinguistics 15:4 .. . 449-Woutersen et al.: Lexical aspects of standard dialect bilingualism

only one signified, and a sign is subordinative when the signifier of the L2is indirectly linked to its signified, that is, through a signifier of the firstlanguage (LI).

Ervin and Osgood (1954) also distinguished between compound and co-ordinate bilingualism. According to them, these types of bilingualism arethe result of the context in which the two languages are learned. As a matterof fact, Weinreich triggered this way of looking at bilingualism when henoted that subordinative bilingualism is likely to apply when a new lan-guage is learned with the help of another. According to Ervin and Osgood,compound bilingualism is a development typical of foreign language learn-ing in the school situation and also of the acquisition by a child who growsup in a home where two languages are spoken more or less interchangeablyby the same people and in the same situations. They have presented coordi-nate bilingualism as a characteristic development of learning two languagesin two different situations, for instance, at home versus at school, or as theresult of second language learning by immersing oneself in the living cultureof another language community, thereby relying on translation as little aspossible.

In several publications (e.g., Grosjean, 1982; Romaine, 1989), objec-tions have been raised against the distinction between coordinate and com-pound bilingualism that Ervin and Osgood made. One of the objectionsraised was that Ervin and Osgood placed too much emphasis on the lexicon,while Weinreich's distinction related to all linguistic systems. Another ob-jection was that the types of bilingualism cannot be of a rigid nature,since the existence of totally coordinate bilinguals is questionable; that is, aperson's type of bilingualism should be a matter of degree.

These objections are not entirely justified because Weinreich only re-ferred to the notions of compound and coordinate in relation to the natureof signs, entities situated within the lexicon. This justifies the emphasis ofErvin and Osgood on the lexicon. Furthermore, Ervin and Osgood wereaware of the fact that the types of bilingualism should not be interpretedtoo categorically: "For any semantic area we would expect speakers of morethan one language to distribute themselves along a continuum from a purecompound system to a pure coordinate system" (p. 141). In fact, Weinreichgave the impetus for distinguishing types of bilingualism in the same way asdistinguishing signs. However, he also warned against drawing too strict adistinction between coordinate and compound bilingualism: "It would ap-pear offhand that a person's or group's bilingualism need not be entirely oftype A [coordinate] or B [compound], since some signs of the languagesmay be compounded while others are not" (p. 10). Recently, de Groot(1993) also claimed that the lexicon of a bilingual is not organized aseither a compound, a coordinate, or a subordinative system, but rather as amixture of those three types. Other problems were noted by Green (1993),who discussed some shortcomings of the coordinate versus compound dis-tinction when it is applied as a model of speech production and perception.Despite all the discussions about the validity of the model for differentpurposes, we hope to show that it is still a useful tool in discussing the


results of lexical decision tasks that probe the organization of the bilingualmental lexicon.

Before doing so, we would like to discuss some general aspects of theparadigm that is primarily used for testing hypotheses about the mentallexicon, lexical decision, and repetition priming - the particular techniquewe used. The reaction times (RTs) obtained in a lexical decision paradigmcan be influenced by manipulating the stimuli. For instance, embedding thestimuli in a related context may lead to shorter RTs (e.g., syntactic, seman-tic, or associative priming). Shorter RTs will generally occur when thetarget stimulus (e.g., sister) is preceded by a related prime (e.g., brother)instead of an unrelated prime (e.g., tree). Neely (1991) presented a reviewof findings and theories using the priming paradigm.

The technique we employed is called repetition priming. Here, the samestimulus is presented for a second time. Normally, RTs are shorter on thesecond presentation of a stimulus. In a monolingual experiment, this meansthat the same mental representation of a lexical entry is accessed twice(intralingual repetition). However, repetition priming can also be used inbilingual tasks. Instead of presenting the same word twice, translationequivalents are used at the second presentation (interlingual repetition).For instance, the Dutch word broer can have a facilitating effect on therecognition of its English translation equivalent, brother, presented later onin the same experiment. Naturally, instead of two languages, a standardlanguage and its dialect can be used: the Dutch standard word veter (shoelace) may or may not have a facilitating effect on the recognition of itsMaastricht dialect translation equivalent, riijstartel. The presence (or ab-sence) and strength of an interlingual repetition effect will have conse-quences for a model of lexical organization in bilinguals. A strong interlin-gual repetition effect, for instance, could indicate a (partially) sharedconceptual representation.

Actually, a more accurate way of referring to the priming in this experi-ment is to call it classical repetition priming. In a classical, delayed repeti-tion priming experiment, the period between the first and second presenta-tions of a stimulus is usually more than 10 minutes. In an immediatepriming experiment, the stimuli are presented instantly after another. DeGroot and Nas (1991) claimed that, in a classical repetition experiment, thepriming effect is episodical rather than lexical in nature. However, Kirsner,Lalor, and Hird (1993) and Monsell, Matthews, and Miller (1992) arguedthat delayed priming is of a linguistic nature - a view we will also adopt inthe rest of this article.

As stated before, in our repetition priming experiment we wanted toinvestigate the effects of a very small typological distance on the organiza-tion of the bilingual lexicon. Consequently, we used standard and dialectstimuli. Furthermore, we investigated three factors: intra- versus interlin-gual repetition priming, proficiency, and level of cognateness.

The variable of intra- versus interlingual priming is of interest because wewanted to distinguish between the various types of bilingualism proposedby Weinreich. In earlier studies using the classical repetition paradigm,


intralingual repetition priming was found, regardless of word type or profi-ciency level (de Bot, Cox, Ralston, Schaufeli & Weltens, 1993; Kerkman,1984; Kerkman & de Bot, 1989; Kirsner, Brown, Abrol, Chadka, &Sharma, 1980; Kirsner, Smith, Lockhart, & King, 1984; Scarborough, Ge-rard, & Cortese, 1984). Monsell et al. (1992) also found intralingual repeti-tion priming in a naming experiment where they first presented definitions,which were repeated after several minutes by pictures. In fact, one couldstate that intralingual repetition priming is a prerequisite for interlingualrepetition priming to appear.

Interlingual priming depends on the relations between the signified (con-cept) and the signifier (word form). Only when there is a relation betweenthe concept and the word forms of the two languages will interlingualpriming occur. When there is an indirect relation between a concept and aword form, there will be more priming in the indirect direction than in thedirect direction because, in the indirect direction, there is more to be gainedby priming.

For the three types of bilingualism distinguished by Weinreich, the fol-lowing predictions about intra- and interlingual repetition effects can bemade:

1. for all types, there will be intralingual repetition priming;2. for the coordinate type of bilingual, there will be no interlingual repeti-

tion priming;3. for the compound type of bilingual, there will be equal interlingual repeti-

tion priming from LI to L2 and from L2 to LI; and4. for the subordinative type of bilingual, there will be more interlingual

repetition priming from LI to L2 than from L2 to LI.

Proficiency is an interesting variable because it is expected to have aneffect on the links between the elements of the two languages and, conse-quently, on the type of bilingualism. Recently, the role of proficiency onthe organization of the bilingual lexicon was modeled by Kroll (1993), whoproposed a model based on Weinreich's compound bilingual. In this modelthere were conceptual links fromboth LI and L2 to the conceptual systemand vice versa. In addition, there were lexical links in two directions be-tween LI and L2. Furthermore, Kroll proposed different connectionstrengths between the various systems, an idea based on connectionist mod-els (see, e.g., Dell, 1986). Kroll stated that the lexical link from L2 to LIwill be stronger than the link from LI to L2 since, for every L2 word, therewill be a mapping in the LI, whereas only few LI words will have mappingsin the L2. As a consequence, the L2 word will be a sort of prototype forseveral LI words.

Paradis (1981, 1987) also proposed different connection strengths be-tween the various elements in the mental lexicon. His subset hypothesisclaimed that the strength of the relations between elements will depend onthe degree to which they are used together. Therefore, elements from onelanguage will generally be more strongly related with each other than with


elements from other languages. A set of strongly related elements is calleda subset. Furthermore, according to the subset hypothesis, the strengthof the relations between elements of two languages may vary depending onthe way in which they are being used together; for example, relationscan be stronger in the direction L1-L2 than in the direction L2-L1 or viceversa.

Both Kroll and Paradis implicitly assumed that the amount of activationspreading from one node to another is a function of the connection strengthof the link between these two nodes. However, with respect to activationspreading, the difference in connection strengths between the two nodes inquestion is not the only important factor. Balota and Lorch (1986) statedthat the amount of activation spreading from a given node is also a functionof the connection strength of this link relative to the connection strengths ofall the other links emanating from the node. In terms of the bilingual lexicon,this means that one has to take into account not only the strength of the con-nection between an element in the LI and an element in the L2, but also thestrengths of the connections between these two elements, on the one hand,and all other elements connected to a given element, on the other.

Proficiency level was operationalized in two ways: (1) by distinguishingtwo language background groups, dialect speakers (LI speakers of dialect,L2 speakers of the standard language) and standard speakers (LI speakersof the standard language, L2 speakers of the dialect); and (2) by using twoage groups, 13-year-olds and 17-year-olds. We expected that a two-wayoperationalization like this would be more likely to reveal the underlyingfactors influencing proficiency than a one-way protocol.

Of the two language background groups, the dialect speakers were sup-posed to be the better L2 speakers: they generally use their L2 (the standardlanguage) in more situations than the standard speakers use theirs (thedialect). We assumed that the dialect speakers were of the coordinate type,while the standard speakers could be of either the compound or the subordi-native type. In Appendix 1 a more detailed description of the dialect andstandard language situation is given.

The older subjects were supposed to be the better L2 speakers of the twoage groups: they have had a longer period of experience in speaking andlistening to their L2. However, differences in proficiency related to age areknown to form a continuum. Thus, the experiment was designed to showwhether there were clear-cut differences in proficiency between 13-year-oldand 17-year-old speakers.

The variable of cognateness is of interest since interlingual repetitionpriming is nearly always found for cognates (i.e., words with a more or lesssimilar form and meaning in both language varieties), but hardly ever fornoncognates (i.e., words with a similar meaning but a dissimilar form,traditionally called translation equivalents). Cristoffanini, Kirsner, and Mi-lech (1986) and Monsell et al. (1992), for instance, found interlingualpriming for cognates in a classical repetition priming experiment. Kerkman(1984) and Kerkman and de Bot (1989) found no interlingual repetitioneffect for cognates, but this result was only obtained for (near-)native


subjects; when the subjects were of an intermediate level, these researchersdid find interlingual repetition priming.

In general, no interlingual repetition effects were found for noncognates(Kerkman, 1984; Kerkman & de Bot, 1989; Monsell et al., 1992). However,most research using bilingual repetition tasks was based on visual wordrecognition. Only de Bot et al. used an auditory version of the classicalrepetition paradigm, and they were the only ones to report interlingualrepetition priming for noncognates; they claimed that this result may bedue to different processing strategies for different modalities. In our experi-ment we were forced to use the auditory version of the repetition primingparadigm because, when using languages that do not have a written formsuch as a dialect, visual word recognition cannot be used. In spite of thefact that the technique used here was the same one used by de Bot et al.,there was no a priori reason for assuming that interlingual repetition prim-ing for noncognates would occur. Therefore, we predicted no interlingualrepetition effect for the noncognates in our experiment.

In order'to arrive at hypotheses that could account for the repetitionpriming of the proficiency groups in the experiment, we combined thepredictions based on Weinreich's three types of bilingualism with the find-ings of earlier studies on cognateness. The formulation of the hypotheseswas also based on the assumption that connections between two elementswould grow stronger as a result of their being used together. Furthermore,we assumed that, in the language one knows better, the RTs would besmaller, which would result in smaller repetition effects because therewould be less to be gained by priming. This led to the following hypotheses.

1. There will be intralingual repetition priming.2 a. There will be interlingual repetition priming for cognates if the subjects

are beginning or intermediate bilinguals.2 b. There will be no interlingual repetition priming for cognates if the

subjects are near-native bilinguals.3. There will be no interlingual priming for noncognates.4. Interlingual priming will vary depending on age.5 a. For the standard speakers, there will be asymmetrical repetition priming

between dialect and standard language if the subjects are subordinatebilinguals.

5 b. For the standard speakers, there will be an equal amount of repetitionpriming between dialect and standard language if the subjects are com-pound bilinguals.

6. For the dialect speakers, there will be no interlingual repetition priming,indicating a coordinate relationship.

METHOD

Materials

A total of 80 standard-dialect word pairs, divided into 40 cognate and 40 non-cognate pairs, were selected as experimental items. In order to find suitable


word pairs, two rating tasks were used. Both rating tasks were performed bysix bilingual speakers. At the beginning of each task, the judges received writ-ten instructions and were allowed to ask questions if anything was unclear.The words to be judged were presented on audio-tape.

In the first rating task, a judgment had to be made about 274 standard-dialect word pairs. The two members of a word pair always had the samemeaning, but varied phonologically from virtually identical to radicallydifferent.' To put it differently, the word pairs varied from cognate tononcognate. The judges had to place the word pairs on a 7-point scalevarying from (1) very dissimilar to (7) very similar. The ratings (i.e.,composite ratings with subjects as a random factor) were highly consistent,Rk(r) = .97. Contrary to our expectations, the distribution of the ratingswas bimodal. This indicates that the speakers perceived most word pairs asbeing either similar or dissimilar.

In the second rating task, the judges had to decide whether the 75 dialectmembers of noncognate pairs would be known by dialect speaking adoles-cents. Here, the judges had to place the word pairs on a 5-point scaleranging from (1) 0 percent to (5) 100 percent. Reasonably consistent rat-ings were found Rk(r) = .84.

Words were selected by taking the 40 cognate and 40 noncognate wordsat the extremes of the distribution of the first rating task. Then, the dialectnoncognate members were compared to the dialect noncognates of the sec-ond rating task. Words that were judged to be used by less than 75% ofdialect-speaking adolescents (i.e., words with a mean rating lower than 4.0)were replaced by new noncognate words taken from the extreme of thedistribution of the first rating task.

To reduce the saliency of the repetition, a total of 160 filler words wereadded to the experimental items, 80 standard words and 80 dialect words.Furthermore, 160 pseudo-words were constructed for each language varietyto obtain an equal number of potential yes and no responses. The pseudo-words were derived from real words by changing one or more phonemes indifferent positions in the word, while obeying the morphophonetic rules ofthe language variety in question. (The experimental stimuli are listed inAppendix 2.)

Apparatus

The experiment was run on an Apple Macintosh SE computer. For digitiz-ing the stimuli, a Farallon Macrecorder was used in combination with theSoundEdit software package. The Experiment Control System (ECS), de-veloped at Carnegie-Mellon University (MacWhinney & Clynes, 1990), wasused to carry out the experimental task. ECS is capable of working with thesoundfile outputs from Farallon's SoundEdit without the use of any furtherconversion software or loss of data. The CrossModal subprogram wasused; in CrossModal mode, the ECS program can be programmed to giveauditory output only, while allowing the experimenter to monitor the sub-ject's progress.

Applied PsychOlinguistics 15:4 455Woutersen et al.°. Lexical aspects of standard dialect bilingualism

Subjects

All subjects (N = 64) were students attending a Maastricht comprehensiveschool. Four groups were formed on the basis of school year and nativelanguage type. Half of the students were in the first year of secondaryeducation and the other half were in the fifth. Most of the first-year stu-dents were 13 years old, but some of them were 14: these subjects comprisedthe younger group. The age of the students in the fifth year varied from 16to 19 years, although most of them were 17 years old: these subjects com-prised the older group. Both the younger and older group included 16 nativespeakers of the dialect and 16 native speakers of the standard language. Inthis way, four groups of 16 students were obtained.

Design

The independent variables in the research design consisted of two between-subjects factors, age and language background, and three within-subjectsfactors, word type, repetition, and stimulus language. The between-subjectsfactors were crossed with each other. They were implemented by using fourgroups of students who differed in age (13 or 17 years old) and nativelanguage (standard or dialect).

The first within-subjects factor, word type, was implemented by usingtwo sets of stimuli, the cognates and the noncognates. The second within-subjects factor, repetition, was differentiated in intra- and interlingual repe-tition. To determine an intralingual repetition effect, the RTs measured onthe second presentation of a word (preceded by a presentation of the sameword in the same language variety) were contrasted with the RTs measuredin the baseline condition (the first presentation of a word in a given lan-guage variety). To determine an interlingual repetition effect, the RTs mea-sured on the second presentation of a word (preceded by a presentation ofits translation equivalent) were compared to the RTs measured in the base-line condition. In this design, the factor repetition was crossed with thethird within-subjects factor, stimulus language, and with the factor wordtype. This means that there were four intralingual repetition conditions,two for standard (i.e., cognates and noncognates) and two for dialect. Wealso obtained four interlingual repetition conditions, two for each languagevariety and two for each word type. In addition, there were four baselineconditions. In this way, a total of twelve conditions was obtained.

In order to distribute the word pairs over the conditions, we had to solvetwo problems. The first problem concerned the fact that the baseline andrepeated presentations in a given condition were separated by a variablenumber of blocks. The second problem concerned an unequal distributionof different word lengths over the various conditions. We handled theseproblems by using a very complicated design in which all blocks and wordswere rotated over the subjects.2


Procedure

The stimuli were presented through Sennheiser HD 400 headphones. Dur-ing each session, only one subject and the experimenter were present. Atthe beginning of each session the subject listened to an instruction read bythe experimenter. There were separate instructions for the younger andolder groups, because the original instruction, which was used for the 17-year-olds, turned out to be too difficult for the 13-year-olds. The subjectswere instructed to respond as quickly and accurately as possible. Bothinstructions were in the standard language, which was the normal means ofcommunication at school. After listening to the instructions, subjects weregiven the opportunity to ask questions. Then, two practice blocks werestarted. These blocks, one in standard and one in dialect, consisted of 25stimuli each. The standard stimuli were pronounced by a man and thedialect stimuli were pronounced by a woman; this was done because thecontrasts between the cognates of both varieties are sometimes so small thatan extra cue is needed to distinguish the different varieties. By using stimulipronounced by persons of different sexes, the subjects could link the voiceof a particular sex to a particular language variety. After the practiceblocks, the subjects were given another opportunity to ask questions, andthe real experiment started, consisting of four blocks of 160 stimuli.

The procedure during the practice and experimental blocks was the same:at the beginning of each block, the experimenter informed the subject whichlanguage variety would be used. Then, the subject reacted to one block ofstimuli. After a break of a few minutes (in which the experimenter selecteda new block), the next block was started. The entire session lasted approxi-mately 1 hour. The subjects were paid for their participation.

RESULTS

Since the hypotheses were explicitly formulated in advance, we were able totest each hypothesis by using a priori comparisons. In order to do so, weused the univariate option within the SPSS MANOVA procedure. For eachhypothesis, the relevant contrasts were defined. Due to the fact that thenoncognate dialect stimuli were not a sample but an almost exhaustiveselective of all noncognate words in the Maastricht dialect, it was not possi-ble to generalize over items. Therefore, only an analysis over subjects (F,)will be reported.3 Thep values that are less than 1% will be reported assignificant, while the p values between 1% and 5% will be reported asindicating a tendency.

Before analyzing the hypotheses, the baseline conditions of the intra- andinterlingual conditions were compared. No significant differences betweenthe intralingual and the interlingual baseline condition were found, F,(l,63) < 1. Therefore, these two baselines were combined and formed thebaseline of the comparisons in the rest of the analyses.

Hypothesis 1, which stated that intralingual repetition priming wouldoccur, was confirmed by our analysis, F,(l, 60) = 45.6, p < .01. Table 1


Table 1. Mean RTs in ms (SD; error percentages) in the intralingual condition

Stimulus language Condition

Prime Target Baseline Repeated Priming

Standard Standard 963 (121; 8Vo) 920 (145; 6%) 43Dialect Dialect 1,210 (182; 1297b) 1,111 (164; 7%) 100

Table 2. Mean RTs in ms fSD; error percentages) of thestandard speaking subjects in the interlingual condition

Target stimulus

CognateBaselineRepeatedPrimingNoncognateBaselineRepeatedPriming

Standard

935 (134; 9%)907 (168; 7%)28

953 (130; 9%)965(176;8%)

- 1 2

1,2511,160

91

1,180, 1,124

56

Dialect

(237; 14%)(250; 10%)

(195; 15%)-(188; 20%)

shows that the RTs in the baseline condition were significantly slower thanthe RTs in the repeated condition. Furthermore, there was a significantrepetition effect for both the standard and the dialect targets, F,(l, 60) =17.8,/? < .01, and/^(l, 60) = 44.6,p < .01, respectively.

In Hypothesis 2a, it was stated that, for beginning or intermediate bilin-guals (standard speakers), there would be interlingual priming for cog-nates. This hypothesis was confirmed by our analysis, F{{1, 30) = 13.3,p < .01. Conversely, in Hypothesis 2b, for near-native bilinguals (dialectspeakers), no interlingual repetition priming for cognates was expected.This hypothesis was also confirmed by our analysis, -F,( 1, 30) = 1.71, n.s.

Hypothesis 3 predicted that there would be no interlingual repetitioneffects for noncognate stimuli. This hypothesis was not supported sincethere was a tendency for repetition priming to occur in this condition, F,( 1,60) = 4.44,/? < .05.

Hypothesis 4 stated that the interlingual repetition effect might vary de-pending on age. However, the analysis showed no difference between theyounger and older students, F,(l, 60) = 3.6, n.s.

Hypothesis 5a stated that, if the standard speakers were of the subordina-tive type, there would be larger interlingual repetition effects for the dialectstimuli than for the standard stimuli. For this hypothesis there tended to bea difference, F,(l, 30) = 5.8, p < .05. Table 2 shows that, for the stan-dard speakers, the effect from the standard language on the dialect (91 ms


Table 3. Mean RTs in ms fSD; error percentages) of thedialect speaking subjects in the interlingual condition

Word

CognateBaselineRepeatedPrimingNoncognateBaselineRepeatedPriming

Target stimulus

Standard

965 (126; 6%)934 (162; 4%)

31

998 (116; 7%)988 (156; 6%)

9

Dialect

1,261 (207; 13%)1,232(242; 11%)

28

1,151 (139; 5%)1,114 (142; 7%)

37

for the cognates and 56 ms for the noncognates) was larger than the effectfrom the L2 on the LI (28 ms for the cognates and -12 ms for the noncog-nates).

In Hypothesis 5b, it was claimed that, if the standard speakers were ofthe compound type, there would be an interlingual repetition effect, regard-less of stimulus language. As shown in Hypothesis 5a, this hypothesis wasnot confirmed by our analysis. For the standard speakers, repetition prim-ing was significant when the targets were dialect cognates, F,(l, 30) =13.6, p < .01, but not when the targets were standard cognates, F,(l, 30)= 2.3, n.s. For the noncognate targets in the dialect, a tendency towardsinterlingual repetition priming was found, F,(l, 30) = 4.8,p < .05. How-ever, there was no interlingual repetition priming for the noncognate targetsin the standard language, F,(l, 30) < 1.

In Hypothesis 6, it was claimed that there would be no interlingual repeti-tion effect for the dialect speakers. This hypothesis was confirmed by ouranalysis, Fx(\, 30) = 3.6, n.s. The results for the dialect speakers are repre-sented in Table 3.

In order to investigate whether the effects found in the F, analysis wereconsistent within the set of items we used, we also carried out an F2 analysiswith only 50% of the stimuli. Overall, the effects found in this analysiswere smaller than in the F, analysis, which was probably due to the limitednumber of items. The only difference we found was an inexplicable interac-tion with age. In sum, the results of this analysis led us to the conclusionthat the effects found in the F, analysis were valid.

Furthermore, because of the rather slow RTs on the dialect stimuli com-pared to the standard ones, an additional analysis on word length (in ms)was carried out. Again we used the univariate option within the SPSSMANOVA procedure. The results indicated that the standard stimuli weresignificantly shorter than the dialect stimuli, F(l, 158) = 171.6, p < .01.The mean word lengths for the standard and dialect stimuli were 542 ms(SD = 107) and 825 ms (SD = 160), respectively.


DISCUSSION

The results clearly showed intralingual repetition priming. This means thatthe precondition for interlingual repetition priming to appear was fulfilled.

Concerning interlingual repetition, there was no interaction with age.This indicates that ages 13 and 17 were not clearly separated on the profi-ciency continuum in this particular context. Furthermore, for the dialectspeakers, no interlingual repetition effect was found. Following ourhypotheses, this means that the dialect speakers were coordinate-type bilin-guals. For the standard speakers there was a difference in the directionalityof priming: namely, there was more priming in the L2 than in the LI.This means that, following our hypotheses, the standard speakers weresubordinate bilinguals.

Although we are aware of the fact that all words in the mental lexicon ofa bilingual person need not be represented in a unitary way (de Groot,1993; Ervin & Osgood, 1954; Grosjean, 1982; Romaine, 1989; Weinreich,1953), we believe that most word representations in the mental lexicon ofa bilingual due to exposure to the L2 tend to evolve by developing andstrengthening links in roughly the same way. Therefore, we are able tospeak of types of bilinguals parallel to the organization of most of theirword representations.

More explicit modeling of how the three types of bilingualism developcan be done on the basis of activation-spreading mechanisms. Let us recallthat the amount of activation spreading from a given node to another nodeis a function of (a) the connection strength of the link between those twonodes and (b) the connection strength of that link relative to the connectionstrengths of all the other links emanating from that node (Balota & Lorch,1986). Furthermore, we will assume that the strength of a link has a decayrate, which can only be stopped by usage of the link in question. In otherwords, links will become weaker when they are not used.

Generally, in the mental lexicon of a bilingual, three types of links can befound: there are intralingual links (i.e., links between concepts and betweenword forms in a given language), there are interlingual links (i.e., linksbetween LI concepts and L2 concepts or between LI word forms and L2word forms), and there are links which - for lack of a better term - maybe called subintralingual (i.e., links between concepts and word forms in agiven language). When we apply these insights to Weinreich's distinctions,the following acquisition model is obtained.

In a subordinative bilingual, there will be intra- and subintralingual linksin the LI, since this is the language these bilinguals have used for theirentire lives. In addition, interlingual links are developed between wordforms of the LI and word forms of the L2 (see Figure 2). As a consequence,activation spreading is possible in the intra-, inter-, and the subintralingualdirections. This means that the activation that spreads into the interlingualdirection has to compete with the activation spreading in the other twodirections.

When a bilingual is becoming compound, subintralingual links in the L2

Applied Psycholinguistics 15:4Woutersen et al.: Lexical aspects of standard dialect bilingualism

460

L2 LI LI

o—osubintra

L2

CONCEPTS

subintra

V - ^ inter ^ - ^ intra ^— inter ^--^WORD FORMS

o CONCEPTS

WORD FORMSinter intra inter

Figure 2. Towards subordinative bilingualism.

strong link

weak link

are developed. (Note that the concepts of two words with the same meaningin LI and L2 stay the same.) At the same time, the interlingual linksbetween the word forms will become weaker because they are used lessfrequently. Therefore, most activation will spread into the intra- and subin-tralingual directions. At first, the subintralingual links in the L2 will be lessstrong than in the LI, since the L2 words are less frequently used together.However, due to more frequent usage, the subintralingual L2 links willgrow stronger until, finally, they will have the same strength as the subin-tralingual LI links, and the bilingual will be completely compound type (seeFigure 3).

When a person becomes increasingly proficient, intralingual links be-tween concepts and between word forms in the L2 are developed. Then,after frequent usage of the L2, the intra- and subintralingual links in the L2will grow stronger and the interlingual links will become weaker. In orderto accommodate this finding (i.e., no interlingual repetition priming), the


461

L2 LI

subintra

LI L2

O Q CONCEPTS

subintra

inter ^—' Iiitra ^ inter ^~-WORD FORMS

inter

CONCEPTS

WORD FORMSinter ultra inter

strong link

weak link

Figure 3. Towards compound bilingualism.

model forces us to assume that, eventually, the intra- and subintralinguallinks will become so strong that the concepts will split - a process for whichwe cannot provide an intrinsic reason. Now, such a person would be acoordinate bilingual with separate concepts for the LI and L2 (see Figure4). As a consequence, activation spreading is no longer possible from L2word forms via concepts to LI word forms or vice versa. In both languagesthe most important directions of activation spreading will be the intra- andsubintralingual.

Thus, a crucial characteristic of the original Weinreich model is thatthere are separate concepts in different languages, at least for coordinatebilinguals. However, there is no intrinsic reason for splitting concepts. Fur-thermore, we think that one can argue that there are no language-specificconcepts at all. Of course, it is reasonable to assume that the conceptualsystem contains sets of concepts that are only used for saying things in aparticular language. But this does not mean that those concepts can only be

L2 LI LIinfra

L2

intra CONCEPTS

subini

^—' intra V - ' inter v - 'WORD FORMS

intra

inter ^—' iintra

intra

intra

\y"""inter v. / ktra V - ^ inter

intra

CONCEPTS

WORD FORMS

CONCEPTS

WORD FORMS

strong link

weak link

Figure 4. Towards coordinate bilingualism.


expressed in that language. For example, for a Dutch speaker, the word cabmay be connected to a concept including the features "New York," "yel-low," and "Chevrolet," simply because that is where he or she has used cabsmost often. Thus, there is a concept somewhere in the speaker's mindfor which there is only an English word. This does not make the conceptlanguage-specific. If this speaker wants to explain in Dutch to colleagueshow to get from JFK Airport to the Empire State Building in a taxi, theentity about which he or she is talking is a cab. That is the starting concept.The speaker will probably mention something about the specifics of taxis inNew York in order to convey the main characteristics of this concept. Inother words, the specific associations and experiences that a speaker haswith a concept can be conveyed by a verbal explanation. In this way, theconcept can be grasped by the listener.

Furthermore, it could be argued that the fact some words have no trans-lation equivalent in the other language proves that some concepts are lan-guage-specific. Every language has a few words that are notoriously diffi-cult to translate. A well-known example of this is the Dutch noungezelligheid, which has to do with coziness, family togetherness, and be-longing; the adjective gezellig means something like 'chummy'. The point isthat the Dutch may be the only speakers to combine these elements into asingle concept. However, it would be possible to give an expose on Dutchfamily life and the role of gezelligheid that would enable a speaker ofEnglish to develop a concept for gezelligheid, no matter what verbal Englishlabel was connected to it. In short, we think that concepts are not language-specific: for some words, there is a one-to-one relation between conceptand word in a specific language, yet for the same concept, there may not bea single word in another language.

Therefore, it has recently been argued that one needs a third level toaccount for language processing in bilinguals (de Bot, 1992; de Bot &Schreuder, 1993; Poulisse, 1993). These researchers have based their pro-posals on the lexico-semantic model of Levelt (1989). (Although this modelis originally a production model, it can also be used to explain perceptionphenomena.) In the Levelt model, three levels are distinguished: the con-ceptual level, the lemma level, and the lexeme level (see Figure 5). At theconceptual level, meaning configurations are represented; at the lemmalevel, semantic and syntactic information is stored; and at the lexeme level,morphological and phonological information are represented. Together,lemmas and lexemes form lexical items. De Bot (1992), de Bot andSchreuder (1993), and Poulisse (1993) claimed that concepts are language-independent, while lemmas are language-specific.

When we implement the distinctions of Weinreich into the Levelt model,the organizational differences are located at the lemma and lexeme levels -not at the conceptual level. Thus, a coordinate bilingual has one concept,two lemmas, and two lexemes; a compound bilingual has one concept, onelemma, and two lexemes; and a subordinate bilingual has one concept,one lemma, an LI lexeme, and an L2 lexeme. This has the following impli-cations for bilingual modeling.

Applied.Psycholinguistics 15:4 464Woutersen et al.: Lexical aspects of standard dialect bilingualism

CONCEPTS

LEMMAS

LEXEMES

strong link

Figure 5. The lexico-semantic model of Levelt.

U LI L2

CONCEPTS

o-oLEMMAS

LEXEMES

strong link

weak link

Figure 6. Towards subordinate bilingualism.

In the first stage, links between LI and L2 are developed at the lexemelevel (see Figure 6). We can speak of subordinative bilingualism. In thesecond stage, links are developed between the (LI) lemmas and the L2lexemes (see Figure 7). This leads to compound bilingualism. Finally, in thethird stage, the lemmas will split. As opposed to the splitting of the con-cepts, the splitting of the lemmas can be explained by the fact that thesyntactic characteristics of the two items are different. Consequently, wecan speak of coordinate bilinguals (see Figure 8).

However, for the coordinate bilinguals, there is a complication at thelemma level. On the basis of Figure 8, one would expect an interlingualrepetition effect because there are links between the LI lemma level, the


465

CONCEPTS

LEMMAS

LEXEMES

strong link

weak link

Figure 7. Towards compound bilingualism.

LEXEMES

Figure 8. Towards coordinate bilingualism.

CONCEPTS

LEMMAS

LEXEMES

strong link

weak link

Applied Psycholinguistics 15:4 466Woutersen et al'.': Lexical aspects of standard dialect bilingualism

CONCEPTS

LEMMAS

LEXEMES

strong link

weak link

inhibitory link

Figure 9. Coordinate bilingualism with inhibitory link.

conceptual level, and the L2 lemma level. The fact that there is no interlin-gual repetition priming for coordinate bilinguals can be explained by aninhibitory link at the lemma level (see Figure 9). This inhibitory link pre-vents the activation from the conceptual level to result in a repetition effect(see also de Bot et al., 1993).

To conclude, psycholinguistic models of the bilingual lexicon should gobeyond Weinreich's model and take into account more recent, empiricallybased proposals for (monolingual) language processing: in particular, bydistinguishing concepts, lemmas, and lexemes and by adding the dynamicsof connectionist models.

APPENDIX 1

SOME REMARKSON THE LANGUAGE SITUATION IN MAASTRICHT

The differences between the Maastricht dialect and standard Dutch are relativelylarge. As a result of its location, with the German border and the French-languageborder (i.e., a French-speaking part of Belgium) only some 20 miles away, Maas-tricht has always had a strong influx from both of these neighboring languages. Inthe lexicon, these influences are very much apparent in words such as the following:

kraank (/kra:rjk/), German krank, as opposed to Dutch ziek (/zi:k/), meaning 'sick';schink (/ink/), German Schinken, as opposed to Dutch ham (/ham/), meaning 'ham';tas (/tas/), from French tasse, as opposed to Dutch kopje, diminutive form of kop

(/kop/), meaning 'cup';versjkt (vSRJet/), from French fourchette, as opposed to Dutch vork (/voRRdk/),

meaning 'fork'.


467

However, the number of these radically different words is very limited. Althoughmany words differ substantially, they mostly derive from the same stem. At thesame time, the list of cognate stimuli presented in Appendix 2 rightly suggests thatboth varieties share a large number of words that are hardly indistinguishable. Evenin cases where there is a spelling difference (excuus-ekscuus; ceintuur-sentuur), theactual stimuli sound very similar. We did, in fact, carry out a discrimination experi-ment with words of this type, and we found that most native speakers of the dialectcannot identify them as belonging to one or the other variety, at least when pre-sented in isolation. This is the reason why we decided to use speakers of differentsexes for the two varieties, in addition to presenting the stimuli in blocks from thesame variety.

As for the language situation in Maastricht, the town is renowned in sociolinguis-tic circles for the widespread use of the dialect. It is spoken in all strata of societyand in all oral domains except the school; there, all "official" interactions takesplace in standard Dutch, while the dialect is used in all informal interactions. Thereis, in fact, a notable range of styles within the dialect. Unlike other dialects, whichare marked as "informal" and "lower class," the Maastricht dialect has both such aninformal, lower class accent and a formal, more refined one. However, the dialectis hardly ever written.

The typical situation, then, is for children to be raised in the dialect if theirparents are dialect speakers. These children are gradually confronted with the stan-dard language through the media and through visitors and peers who speak standardDutch. For most mixed marriages and monolingual standard Dutch families, thechildren's situation is basically the reverse: they are raised in the standard language,but are gradually exposed to the dialect when the children start to play outside thehouse or go to a nursery school. In general, all dialect speakers master the standardlanguage at a very early age, although most of them retain a noticeable southernDutch accent for life. Standard speakers usually completely master the dialect re-ceptively, but their productive skills vary substantially: they may range from elemen-tary to fully proficient, largely depending on their personal language situation.

APPENDIX

STIMULI

Practice,Standard

bochelboodschapdamgalggansgilvangen

2

Cognates,Standard

aapaardeadderaltaarbaasbakbang

Cognates,Dialect

aapaardeadderaltaarbaasbakbang

Noncognates,Standard

aalbessenbaksteenbiddenbordbottenglijbaanham

Noncognates,Dialect

wiemelebrikbeijetelleurkneuksjievelsjink


468

Practice,Standard

vcrgietwoordzaak

Practice,Dialect

bokdochtergaafhuifkeerbeurstelkelderkrissie16tsjvrechwdnder

Fillers,Standard

Cognates,Standard

beeldbeurtbevenblikbrckcnbrokceintuur

dadeldendokterdriftigdunexcuusezelfakkelfietsflesframboosgalgavegebouwgedraggeldgesprekgeweergladgolfgrensmistaaktikkenverlegenziel

Cognates,Dialect

beeldbeurtbeveblikbrekebroksentuur

dadeldendokterdriftigdunekskuusezelfakkelfietsflesframboosgalgavegebouwgedraggeldgesprekgeweergladgollefgrensmistaaktikkeverlegeziel

Fillers,Dialect

Noncognates,Standard

hardhelemaalhuilenjurkklaarklokhuiskopje

krabbenkrantkruisbeskruiwagenkwaadlaarslucifermooinaaldniezennooitoomplaspreiprulruziescheidingschommelsnorspeldspuugsturenvorkvuilnisbakziekzoenen

Noncognates,Dialect

helganskrietekleidveerdigkitsjtaskretsegezetkrdnselkroukaarkoetstievelzwegelsjoennaojeneestenoetsnoonkpoolpoorkldmmelruizingsjeigelsjoggelknievelspangspeijsjikkeversjetdrekbakkraankpune

aambeeldaardbeiabdijafgrondautobandjebeen

bijbijbelbindenbisschopblauwboekboordje

antwoordaosemartiesbagaasjballekbatteraofbed

beerbekkerberregblaankbrandbrelbroed


469

Fillers,Standard

bootbreedbroekcitroendagdeegdekseldijkdikwijlsdorpdraaddrempelduifdurfeeuwernstfatsoenganggewoongraangrappenmakergreppelhelhooihurkenklapknijpenkoekoelkomkoorkorstkraai

Pseudo-words,Standard

aafaalbekkenaarpeaftrondagemalfaaramfoort

kraankruimelladenleeuwlerenleverenloodloonmensmesmuisnichtjeoogooroudpaardpetrugsausschurenslasprongstemtapkastuurvoerenvriendvriezenvuilzilverzingenzuigenzweep

ankeltapperaukobabbezaarbakkelbeelbekkel

Fillers,Dialect

br6mmelbroordampdekdolfijndollekdoorsdoum

. drokkedroppeleerappeleilandfellemflinkfruitgaasgasgleujegoejegriesgr6ndhierh6ndhummeiesjoedkaajtkaosketeerknienkorteletleedlieg

belebeslatbeulaarbeusbiekbiekelbiffen

lieningluipmellekmerretmeulemiewmoermojermoknuijtsorregensrieteroukesjawsjeepsjeetesjoklaatsniestasiestrutaaktoentoerttouwversnerrekeweurswoenewoer.woordebookzaajtziezoerzwaon

bijkbingelenbin-enbluisblukblunkboffel

bokerboofborkbrankbrekbrepenbrieken


470

Pseudo-wordsStandard

brietbrimmelbrolbrukbuifbuitnbuizelburkdaapdakgoordalmdasteldaveldeteldeuldevendochteldokserdolfoendoondraasdraftigdrakdreeddrullenduipdwoom

Pseudo-words,Dialect

aajkaankelabrijjaostelartielartosawwerieberekberrefbiffeblakbliewbluuk

etelfalgfaroenfarrelfichtflaasfrantfrepfroodfuimfuitgalkgalpergarmgepsglapgolgorkgroelgruipelgruitenguienguifhespelheukhichtenhintel

boeldboffelbopkebrepaajbrezebriekebrierbroulbruikbuukbuuldbuzeldakel

hoikhuukjuifkangkistelklamerkleizenknekerknisserknotelkolfijnkraafkrangkrempelkretelenkromelkwaspkwoestlefterlekellorplorstmeegmeepoeperoerensomvak

defdetdieksdokserdorrekdraonsdreiddrottelfiekfitoeflaapflietelfosser

ooppeempinzelplaamplinkpraspraunpruirastelrazijnrindenrolenrolkscherschieksmarensmuisspatensprokerstagersteepstolkstrapstreulstromstupsuif

fotsfoumfreelfrugamgekragglotsgrendgrinshaorstelheurthierelehietel

suksundertasptekkenteultiekeltolftuikvrunterwakkelwapelwapswasselwegeerwengerweukwijgwoekworwustwuutzaapzarmzeldzintenzurkenzwoog

hotsjhuijkeriefiekrepinklingeinslienitterpinjaggeljarpinjorskepperaarketkinnezik


471

Pseudo-words,Dialect

kissetrekklabbedieklakasklatsjelkleuverkloupkniekelkdmferkorrepelkoutepritsjkraspelkriemelkroefelkuulpelaaslaordsjelezaarliemerlieremeloorslorres

loutelmagoejeremavemegoemersjeemesteermiekermogkelmortintnaotsjneuntefnig6wnistelnittelnupsielnuustelokkelepiepokskeoileropplatsjeortsje

oukpatangperrekelpijjzelplaansplegelpleksiepplitskeploetpottelpraanprakuulpraogelprikkedasierawwelerekijjnrelispretapriekelerielriesjp

sartijjsentuuksjandongsjmuussjosterdsjprelskootsliesmilsnoksoppelsoufstriestaanjeltadtakketiestaortelteuktoptrangtrastel

trekstaktrijsuitelumweesviegowvijjvoekelvraamvraotelvreuswaalekwanswaperwetsjwiepwintelewokworrelijjnzaafzokkerzuul

ACKNOWLEDGMENTSThis research was reported by the Netherlands Organisation for Scientific Research(NWO). The authors wish to thank Rob Schreuder for his helpful comments andErik Schils for his statistical assistance.

NOTES1. It is not claimed here that so-called translation equivalents are completely equiv-

alent in meaning. In the research described here, a bilingual dictionary (Ende-pols, 1955) was used as a criterion for the selection of translation equivalents.

2. Further details about the design are available by writing the authors.3. As an anonymous reviewer pointed out, the absence of an F2 analysis has

implications for generalizing to other dialects. We are aware of this fact. In thisarticle, we only want to make claims about the Maastricht dialect.

REFERENCESBalota, D., & Lorch, R. (1986). Depth of automatic spreading activation: Mediated priming

effects in pronunciation but not in lexical decision. Journal of Experimental Psychol-ogy: Learning, Memory and Cognition, 12(1), 336-345.

de Bot, K. (1992). A bilingual production model: Levelt's speaking model adapted. AppliedLinguistics, 13(1), 1-24.

Applied Psycholinguistics 15:4 .. . 472Woutersen et al.: Lexical aspects of standard dialect bilingualism

de Bot, K., Cox, A., Ralston, S., Schaufeli, A., & Weltens, B. (1993). Lexical processing inbilingual*. Manuscript submitted for publication.

de Bot, K., & Schreuder, R. (1993). Word production and the bilingual lexicon. In R.Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 191-214). Amsterdam:Benjamins.

Cristoffanini, P., Kirsner, K., & Milech, D. (1986). Bilingual lexical representation: The statusof Spanish-English cognates. Quarterly Journal of Experimental Psychology, 38A,367-393.

Dell, G. (1986). A spreading-activation theory of retrieval in sentence production. Psychologi-cal Review, 93, 283-321.

Endepols, H. (1955). Diksjenaer van 't Mestreechs [Dictionary of the Maastricht dialect].Maastricht: Gadet.

Ervin, S., & Osgood, C. (1954). Second language learning and bilingualism. Journal ofAbnormal and Social Psychology, 4P(suppl.), 139-146.

Green, D. (1993). Towards a model of L2 comprehension and production. In R. Schreuder &B. Weltens (Eds.), The bilingual lexicon (pp. 249-277). Amsterdam: Benjamins.

de Groot, A. (1993). Word-type effects in bilingual processing tasks: Support for a mixedrepresentational system. In R. Schreuder & B. Weltens (Eds.), The bilingual lexicon(pp. 27-51). Amsterdam: Benjamins.

de Groot, A., & Nas, G. (1991). Lexical representation of cognates and noncognates incompound bilinguals. Journal of Memory and Language, 30{ 1), 90-123.

Grosjean, F. (1982). Life with two languages. Cambridge, MA: Harvard University Press.Kerkman, J. (1984). Woordherkenning in twee talen [Word recognition in two languages]. In

A. Thomassen, L. Noordman, & P. Eling (Eds.), Het Leesproces (pp. 139-152). Lisse:Swets & Zeitlinger.

Kerkman, J., & de Bot, K. (1989). De organisatie van het tweetalig lexicon [The organizationof the bilingual lexicon ]. Toegepaste Taalwetenschap in A rtikelen, 34, 115-121.

Kirsner, K., Brown, H., Abrol, S., Chadka, N., & Sharma, N. (1980). Bilingualism andlexical representation. Quarterly Journal of Experimental Psychology, 32, 585-594.

Kirsner, K., Lalor, E., & Hird, K. (1993). The bilingual lexicon: Exercise, meaning andmorphology. In R. Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 215-248). Amsterdam: Benjamins.

Kirsner, K., Smith, M., Lockhart, R., & King, M. (1984). The bilingual lexicon: Language-specific units in an integrated network. Journal of Verbal Learning and Verbal Behav-ior, 23, 519-539.

Kroll, J. (1993). Accessing conceptual representations for words in a second language. InR. Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 53-81). Amsterdam:Benjamins.

Levelt, W. (1989). Speaking: From intention to articulation. Cambridge, MA.: MIT Press.MacWhinney, B., & Clynes, D. (1990). Experimental Control System. Carnegie Mellon Uni-

versity, Pittsburgh, PA.Monsell, S., Matthews, G., & Miller, D. (1992). Repetition of lexicalisation across languages:

A further test of the locus of priming. Quarterly Journal of Experimental Psychology,44A(4), 763-783.

Miinstermann, H. (1989). Dialect loss in Maastricht: Attitudes, functions and structures.In K. Deprez (Ed.), Language and intergroup relations (pp. 99-128). Dordrecht:Foris.

Miinstermann, H., & Hagen, T. (1986). Functional and structural aspects of dialect loss: Aresearch plan and some first results. In B. Weltens, K. de Bot, & T. van Els, (Eds.),Language attrition in progress (pp. 75-96). Dordrecht: Foris.

Neely, J. (1991). Semantic priming effects in visual word recognition: A selective review ofcurrent findings and theories. In D. Besner & G. Humphreys (Eds.), Basic processes inreading: Visual word recognition (pp. 264-336). Hillsdale, NJ: Erlbaum.

Paradis, M. (1981). Neurolinguistic organisation of a bilingual's two languages. In J. Cope-land & P. Davis (Eds.), The seventh Lacus forum (pp. 420-431). Columbia, SC:Hornbeam Press.

(1987). The assessment of bilingual aphasia. Hillsdale, NJ: Erlbaum


Poulisse, N. (1993). A theoretical account of lexical communication strategies. In R.Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 157-189). Amsterdam:Benjamins.

Romaine, S. (1989). Bilingualism. New York: Basil Blackwell.Scarborough, D., Gerard, L., & Cortese, C. (1984). Independence of lexical access in bilingual

word recognition. Journal of Verbal Learning and Verbal Behavior, 23, 84-99.Weinreich, U. (1953). Languages in contact: Findings and problems. New York: Linguistic

Circle.

Lexical aspects of standard dialect bilingualism

Documents