Meaning and function of dummy auxiliaries in adult acquisition of Dutch as an additional language

Abstract

This article presents the results of experimental data on language production and comprehension. These show that adult learners of Dutch as an additional language, with different language backgrounds, and a L2 proficiency below level A2 (Waystage) of the Common European Framework of Reference for Languages (CEFR; Council of Europe, 2001), use dummy auxiliaries as a structural device and interpret them as semantically vacuous. Proficiency level in the target language, more than language background, seems to determine the occurrence of dummy auxiliaries, and also which dummy auxiliary is used. Participants at a lower level of language acquisition use both dummy auxiliaries zijn (‘be’) and gaan (‘go’), whereas more advanced learners continue using predominantly dummy auxiliary gaan. These findings suggest that both dummy auxiliaries have a trigger function in setting the step from nonfinite utterances, to utterances with dummy auxiliaries carrying morphological information, and finally to utterances in which the morphological information is carried by the finite verb.

Keywords

aspect Berber Tarifiyt dummy auxiliaries Dutch finiteness Moroccan Arabic second language acquisition tense Turkish

I Introduction

1 General introduction

Dummy auxiliaries are non-grammatical forms often attested not only in the learner varieties of adult learners of Dutch as an additional language (DAL) (Blom and De Korte, 2008 and 2011; Starren, 2001; Van de Craats, 2009; Van de Craats and Van Hout, 2010; Verhagen, 2009, 2011, 2013), but also in learner varieties of other target languages such as French and German (Schimke, 2013), and English (Fleta, 2003; Huebner, 1989; Huebner, Carroll, and Perdue, 1992; Zobl, 2002). They have also been observed in child first language (L1) and second language (L2) acquisition of Dutch, English and German (Cornips, 2013; García et al., 2005; Haberzettl, 2003; Hollebrandse et al., 2013; Hulk and Cornips, 2005; Jordens and Dimroth, 2006; Julien et al., 2013; Lalleman, 1986; Tracy, 2002; Zuckerman, 2001), suggesting that the occurrence of dummy auxiliaries may be a more general, rather than a language-specific step in language acquisition.

Use of dummies is often related to verb movement; compare Chomsky’s (1995) Economy Principle. Dummy auxiliaries are assumed to be easier to access than lexical verbs because, due to their high frequency, they are stored in their inflected form in the mental lexicon, from where they can be directly retrieved and inserted in the position of a functional head. Production of lexical verbs in the same position, on the other hand, requires movement, which is considered a more costly operation. Van Kampen (1997) and Zuckerman (2001, 2013) observed that monolingual Dutch children use dummies in main sentences – which require verb movement – and that dummies in subordinate clauses – where verb movement is not required – are very rare. Blom and De Korte (2011), Van de Craats (2009) and Van de Craats and Van Hout (2010) studied, respectively, the learner varieties of Dutch L2 Turkish-speaking children, and adult Turkish- and Moroccan Arabic-speaking DAL learners, and concluded that dummy auxiliaries, more in particular gaan (‘go’) and zijn (‘be’), precede movement of the lexical verb and constitute a structural device that does not require movement and that helps the learner to open a position in the sentence for verbs to move to.

Other researchers, such as Dimroth et al. (2003), Jordens and Dimroth (2006), Starren (2001), and Verhagen (2009) offer a primarily semantic account for the use of zijn + inf and gaan + inf. They claim that (dummy) auxiliaries, just as modals, are carriers of both the morphological properties of finiteness and the illocutionary function of an assertion. In her longitudinal study of language production data from two Turkish and two Moroccan learners of Dutch, Starren (2001) observed that, after an initial phase in which adverbs are used to express tense and aspect, they are replaced by forms of the auxiliary verb zijn, which she terms proto-auxiliaries. She concluded that her learners use those proto-auxiliaries to encode topic time (in pre-subject position) and perfect aspect (in post-subject position).

Zuckerman (2001, 2013) and Verhagen (2013) investigated respectively the comprehension of dummy auxiliary gaan by monolingual children acquiring Dutch and dummy zijn by Moroccan DAL learners. They came to the conclusion that learners, in early stages of their acquisition of Dutch, ascribe imperfective meaning to those two dummy auxiliaries, rather than prospective (gaan) or perfect (zijn).

From the above, it is clear that there is no consensus as to the function and meaning of dummy auxiliaries. The overarching research question that we want to answer in the present study is whether dummy auxiliaries are devoid of meaning and are used by adult DAL learners as a structural device that helps them acquire the proper morphological marking of agreement, as claimed by Van de Craats (2009) and Van de Craats and Van Hout (2010).

This study contributes in three ways to what we know from previous research on this topic. One is that not only Turkish and Moroccan Arabic speakers were the objects of study, but also speakers of Berber Tarifiyt were included. This group of learners has not been systematically distinguished from the Moroccan–Arabic speakers in previous studies. Tarifiyt speakers constitute approximately 60% of the Moroccan population in the Netherlands (El Aissati et al., 2005). The second contribution is that not only the production but also the comprehension of (dummy) auxiliaries in Dutch was studied. The aim of studying both production and comprehension was to provide a deeper insight of the question whether learners assign meaning to dummy auxiliaries or not. To our knowledge only Zuckerman (2001) and Verhagen (2013) have addressed the comprehension of dummy auxiliaries. The third contribution is the diversity in elicitation tasks to trigger the use of dummy auxiliaries in different contexts and to test factors that may inhibit or enhance the occurrence of dummy auxiliaries. In addition, we have drawn a larger sample of learners than other studies on adult DAL learning did so far, with the exception of Verhagen (2009, 2013).

The outline of the article is as follows. We start by describing the relevant markers of temporality in the four languages involved in this study and by presenting the research questions. Section II explains the experimental framework and the methods for data collection and analysis. This is followed by the Section III, in which the results are presented. Section IV discusses the main findings and presents the conclusions.

2 Markers of temporality and the concept of dummy auxiliaries

Investigation of the source of deviant forms in the learner varieties cannot be carried out without considering the possibility that those errors are produced under influence of the languages involved. Van de Craats and Van Hout (2010) have put forward the hypothesis that dummy gaan could be the result of interference from L1 Moroccan Arabic, since that language has a real auxiliary which is, qua form and meaning, very similar to the auxiliary gaan in Dutch. Dummy zijn is assumed by them, by Julien et al. (2013) and by Verhagen (2013) to originate from the Dutch language itself, under the influence of the extensively used copula zijn.

A description of relevant linguistic means for the expression of temporality in the languages involved in the present study is presented in Table 1. The temporal framework of Klein (1994) was adopted to distinguish the aspectual categories of prospective, imperfective, perfect and perfective. Table 1 shows how different the four languages involved are in expressing aspectual distinctions. Turkish uses mainly synthetic forms (Göksel and Kerslake, 2005), whereas the other three languages often make use of analytic forms. In Tarifiyt (E-Rramdani, 2003; El Aissati, 1994, 2001) and Moroccan Arabic (Harrell, 1962; Hoogland, 1996), perfect and perfective aspects are expressed each with a different form, whereas Turkish has only one form for both aspects. Dutch is peculiar in that it has two forms to express perfective aspect, one of them being the same as the form used to express perfect aspect (Klein, 1994). Further, Tarifiyt has different forms to express prospective aspect, and Dutch has multiple ways of expressing imperfective aspect.

Table 1.

Markers of aspect in Dutch, Tarifiyt, Moroccan-Arabic and Turkish.

Basic order	Prospective	Imperfective	Perfect	Perfective
Dutch (SOV)	gaan + inf	zijn + aan het + inf	zijn/hebben + ppart	simple past
		present tense	zijn/hebben + ppart	zijn/hebben + ppart
		posture verbs + inf
Tarifiyt (VSO)	ad + aor traḥ + ad + aor qa + traḥ + ad + aor	qa + imprf	qa + prf	prf
Moroccan-Arabic (SVO)	ġādi + imprf	ka (ta) + imprf	active part	prf
Turkish (SOV)	stem+ -(y)acak/ecek	stem + -yor	stem + -DI	stem + -DI

Notes. aor= aorist; imprf= imperfect; inf = infinitive; O = object; ppart= past participle; prf = perfect; S = subject; V = verb.

Since the present study concerns the acquisition of agreement marking on the finite verb in Dutch, a more detailed description is given below. Dutch main clauses have the special property that the finite verb occupies the second position, irrespective of which constituent occupies the first position. This property is known as Verb Second (V2). In the standard generative analysis the V2 order is derived from an underlying structure in which the verb is in head-final position (SOV) within the Verb Phrase (VP). Through head-to-head movement from verb (V) to Inflection (I), the finite verb ends up in the complementizer (comp or C) position (cf. Den Besten, 1989; Koster, 1975), or in I (AGR/T) (see Zwart, 1997). If a constituent other than the subject is moved to the Spec,C position, the result of this movement is that subject and the finite verb appear in the opposite order. This is often called subject–verb inversion (see Stage 4 in (1)).

An example of the stages various types of learners go through, with dummy auxiliary (Stage 2) and without dummy auxiliary insertion (skipping Stage 2), is provided in (1).

(1) Stage 1: * Jan [_VP naar de kerk lopen]

Jan to the church walk.inf

‘Jan walk to the church.’

Stage 2: [_CP Jan [_C gaat [_VP naar de kerk lopen]]]

Jan go.3sg to the church walk.inf

‘Jan is going to walk to the church.’

Stage 3: [_CP Jan [_C loopt_i [_IP[_VP naar de kerk t_i ]t_i]]] (according to ten Besten/Koster)

Jan walk.3sg to the church

‘Jan walks to the church.’

or [_IP Jan [_I loopt_i [_VP naar de kerk t_i ]]] (according to Zwart)

Jan walk.3sg to the church

‘Jan walks to the church.’

Stage 4: [_CP Hier [_C loopt_i [_IPJan t_i [_VPnaar de kerk t_i ]]]]

Here walk.3sg Jan to the church

‘Here Jan walks to the church.’

Dutch has a two-way system – past and non-past – to grammaticalize tense. Aspect is often expressed through auxiliaries and constructions with a posture verb or with the prepositional phrase aan het V (‘on the V’), where V stands for a verb in infinitival form. The aspectual auxiliary occupies the second position and the lexical verb, in the infinitive, a sentence-final position. Prospective aspect is commonly expressed with the auxiliary gaan in combination with the lexical verb in the infinitive as illustrated in (2).

(2) Jan gaa-t zijn vriend roepen

Jan go-pres.3sg his friend call.inf

‘Jan is going to call his friend.’

The simple present tense, as well as posture verb constructions and constructions with an adjective or preposition can be used to convey imperfective aspect. In (3a–c), examples are given of each of these constructions.

(3) a. Simple present

Moeder bouw-t een toren

Mother build-pres.3sg a tower

‘Mother builds / is building a tower.’

b. Zijn + aan het + INF

Zij is een toren aan het bouw-en

She be.pres.3sg a tower at the build-inf

‘She is (in the process of) building a tower.’

c. Zijn + INF

Moeder is werk-en

Mother be.pres.3sg work-inf

‘Mother is (not here) working.’

Flecken’s study (Flecken, 2011) reveals that the aan het inf construction (3b) is widely used by adult Dutch native speakers. The last construction (3c) is less frequent than the other constructions. It has the very specific meaning of somebody doing something not in the vicinity of the speaker, and is, in general, not used with verbs requiring an object or having a clear endpoint.

Perfect aspect is expressed with the construction zijn (‘be’) / hebben (‘have’) + past participle.¹ Example (4) illustrates this.

(4) a. Peter is aangekomen

Peter be.pres.3sg arrive.ppart

‘Peter has arrived.’

b. Moeder heeft een toren gebouwd

Mother has.pres.3sg a tower build.ppart

‘Mother has built a tower.’

Perfective aspect is expressed by using the simple past, as in example (5), or the construction zijn (‘be’) / hebben (‘have’) + Past Participle as examples (6a) and (6b) illustrate.

(5) De poes at de vis

The cat eat.past.3sg the fish

‘The cat ate the fish.’

(6) a. Ik heb in het park gelopen

I have.pres.1sg in the park walk.ppart

‘I walked in the park.’

b. Ik ben naar het park gelopen

I be.pres.1sg to the park walk.ppart

‘I walked to the park.’

3 Research questions

As stated in the introduction, the main inquiry in this study is whether dummy auxiliaries are indeed devoid of meaning and used by adult DAL learners as a structural step in the acquisition of finiteness, as claimed by Van de Craats (2009) and Van de Craats and Van Hout (2010). In order to answer that overall question, we address the following specific research questions:

1. Do low-proficient adult DAL learners use dummy auxiliaries?

The prediction is that dummy auxiliaries will be found in the present study since they have often been attested in (early) learner varieties of adult DAL learners.

2. Does level of language proficiency in Dutch influence the choice and frequency of dummy auxiliaries?

Previous research has shown that L2 children and adult learners of Germanic languages use very few dummy auxiliaries at the initial stage of language acquisition, and that dummy use increases as learners’ exposure to the target language increases, decreasing again later, while finiteness is being acquired. Huebner (1989), Haberzettl (2003), Van de Craats (2009) and Van de Craats and Van Hout (2010) showed in their studies the emergence, increase and decline of the is/ist- and ga/gaat-patterns in relation to increasing proficiency in the target language. That pattern appeared between a stage in which the lexical verbs were uninflected (in German and Dutch the word order at that stage was mainly SOV) and a stage with inflected verbs (in German and Dutch the word order was SVO). In Van de Craats’ study (2009), participants began to produce is-constructions about nine months after having started receiving formal L2 input, increased their use in the next nine months and, in the case of some participants, decreased in the third period of nine months that followed.

In the light of these findings, the same pattern can be expected in the present study. It is predicted that the number of dummy auxiliaries produced by participants who have a lower level of proficiency will be lower than the number produced by the participants with a higher proficiency level.

3. Does language background determine the choice of dummy auxiliary?

Previous research (Van de Craats, 2009) has shown that Turkish learners of Dutch mainly used the dummy auxiliary zijn, and Moroccan Arabic learners had a preference for the dummy auxiliary gaan (Van de Craats and Van Hout, 2010). On the basis of these findings, it is predicted that the choice of dummy auxiliaries will differ among the three language groups participating in this study. It is further hypothesized that correct aspectual use of the construction gaan + inf will be acquired faster by the Arabic-speaking group than by the other groups, since that language has a construction (ġādi (‘go’) + imprf) that resembles, in form and meaning, the Dutch prospective construction.

4. Do adult DAL learners, in the initial stages of language acquisition, assign meaning to the constructions zijn + inf and gaan + inf?

Given the results of Zuckerman’s (2001) and Verhagen’s (2013) studies, in which it was found that L1 children and Moroccan learners of Dutch ascribe present tense reading to respectively gaan + inf and zijn + inf, our prediction is that the participants in the present study will also include the imperfective aspect when assessing the meaning of these two dummy auxiliary constructions. If dummy auxiliaries are devoid of meaning for learners, no distinction will be made between utterances with dummy auxiliaries and utterances containing the present tense.

II Method

1 Participants

The participants were 40 adult learners of Dutch, five male and 35 female, who came to the Netherlands after their sixteenth year of life. They spoke Moroccan-Arabic, Berber-Tarifiyt, or Turkish as their L1, and were acquiring Dutch at various training centers in cities of the Randstad (the metropolitan region in the western part of the Netherlands), and in cities in the eastern part of the Netherlands.

All participants had a low level of proficiency in Dutch and were attending Dutch classes aimed at attaining levels A1 (Breakthrough) and A2 (Waystage) of the CEFR. Their L2 oral proficiency was rated by experienced language teachers, who assigned them to one of the two groups mentioned above. Assessments based on CEFR are aimed at evaluating communicative rather than grammatical skills. All participants were reported by the teachers as not yet having fully acquired verb inflection and placement in Dutch.

All three participant groups had a low educational level. None of them had attended university and six of them had never attended school. Most of the participants had some knowledge of one or more other languages than their L1 when they started learning Dutch. The languages that they understood and/or spoke to various degrees of proficiency were English, French, Kurdish or Spanish. Seven of the Tarifiyt speakers had also command of spoken or written Arabic. Their command varied from only understanding it (four participants) to speaking it with varied degrees of fluency (five participants). One participant could only read and write it. The social status (low), in the Netherlands, of these three groups of immigrants, as well as their history of migration, are comparable (Crul and Doomernik, 2003; Roes, T, 2008; Scheele et al., 2010). Table 2 gives an overview of the characteristics of the 40 participants divided into language background and proficiency level in Dutch.

Table 2.

Most relevant learner characteristics: Range (minimum–maximum), Mean (M) and SD.

L1	Level		Age			Length of residence (years)			Length of instruction in Dutch (years)
	A1	A2	Range	M	SD	Range	M	SD	Range	M	SD
Tarifiyt	6	4	30–49	38	7.6	0.7–33	13.2	11	0.5–4	1.9	1.4
Arabic	5	5	26–50	32	6.7	1–17	7.0	5.3	0.5–4	1.5	1.4
Turkish	7	13	21–54	37	10	1–28	7.1	6.4	0.7–6	1.7	1.9

2 Elicitation tasks

Table 3 shows the three production experiments and the comprehension experiment that were administered to the participants. This set of experiments, all of them eliciting the third person singular, was designed not only for adult DAL learners, but also for monolingual and bilingual children acquiring Dutch, who took part in a bigger study on this subject.

Table 3.

The experiments in this study.

Experiment	Type of task	Mode	N of items
1	Multiple choice (meaning interpretation)	Comprehension	60
2	Narrative (meaning)	Production	16
3	Completion syntax	Production
	(a) canonical order (SV)		23
	(b) inverted order (XVS)		23
4	Completion morphology	Production
	(a) present tense		34
	(b) past tense		34

Both the comprehension task and the narrative task were designed to test whether aspectual meaning is involved in the constructions zijn + inf and gaan + inf. The tasks consist of looking at film clips, and at three pictures extracted from each film, which are presented on the screen immediately after the film. The first picture shows the image one fraction before the beginning of an action or state, the second picture shows the action/state itself and the third picture shows the end of the action/state.

The completion syntax task was developed to answer the question whether the number of dummy auxiliaries produced increases as the number of syntactic steps required to get an inverted (XVS) order increases. This was the case in Blom and De Korte’s study (2011), but only with respect to children. They produced more dummy auxiliaries in sentences with than without inversion. Adult DAL learners who took part in that same study did not produce sentences with inversion, presumably due to their low level of proficiency in Dutch. In view of those results, the participants in the present study were not expected to realize inversion consistently, since their level of proficiency in Dutch is below A2. Nevertheless, the task was administered to observe their use of dummy auxiliaries in that context. The sentence completion tasks ‘present tense’ and ‘past tense’ were developed to investigate the relationship between morphological skills (inflection) and the use of dummy auxiliaries.

3 Materials

The verbs used in the experiments were divided into four classes based on lexical aspect (Aktionsart), that is, their durative nature, argument structure and the possible presence of an adverbial or an object in the sentence signaling an endpoint to the action or state. Vendler’s (1957) four-way classification was the basis for the categorization of the verbs used in this study. These verbs were also grouped on the basis of morphological regularity and presence of a verbal particle.

Short clips and pictures taken from the television series Pingu were used.² The program DVDx 4.0 was used to rip the DVD Pingu voor altijd (‘Pingu forever’) (Pygos Group & Hit Entertainment Limited, 2010). Afterwards clips were made from the converted DVD using the program Virtual Dub 1.9.

4 Procedures

Participants were tested individually in a quiet room, in the center in which they were studying. The tasks were presented to them in a laptop using the software program E-prime 2.0 (Schneider et al., 2001). Depending on the task, each participant was instructed to match a picture with a stimulus utterance, which they heard via the laptop’s loudspeaker, or to describe an event on a film clip and/or an action depicted in a picture, or a picture sequence, presented to them in the laptop. The E-prime program allowed not only a randomization of the test items in the experiments, but also the recording of the participant’s responses. In the language comprehension task the E-prime’s function to record accuracy of response was used.

We analysed the data using the software programs Microsoft Excel 2007 and IBM SPSS 21. All the utterances produced by the participants were orthographically transcribed. Transcription conventions used were taken from the transcription system of CHAT (Codes for the Human Analysis of Transcripts) (MacWhinney, 2000).

A coding system in which all types of verbal constructions produced by the participants were included was designed for all four experiments. The coding system consists of 10 categories, each representing a certain type of construction and its variants.

III Results

This section provides the results of the four experiments used in this study. They will be presented as follows: first, the analysis of target responses in the comprehension experiment will be reported. This will be followed by the results of the four production tasks. For each experiment the analysis of target responses will be presented first, followed by the analysis of the occurrence of dummy auxiliaries. Although the effects of lexical aspect and morphological characteristics of the verbs on the use of dummy auxiliaries was analysed, the results of those analyses will not be presented, because there were no significant outcomes.

1 Comprehension experiment: Target responses

Reliability tests were carried out to measure the internal consistency of the 60 items that this experiment comprises. In the gaan + inf condition, with the prospective meaning as target³, the reliability of the items is high (Cronbach’s α = .906), and there were no deviant items. In the present tense condition, with the imperfective meaning as target, the number of correct scores is high. Six of the items elicited always the target responses. Twelve items elicited non-target reactions. The reliability for those items was α = .674. In the zijn + inf condition with the imperfective meaning as target as well, there were also six items that always elicited the target responses, and 12 items that sometimes elicited non-target responses. The reliability of those items was not high (α = .556). The lower reliabilities of the present tense and the zijn + inf conditions are brought about by the high scores of the informants. Figure 1 shows that the outcomes have a ceiling effect. Most informants have scores between 90% and 100% of target responses.

Figure 1.

Mean proportions and error bars (+/– 2SE) of target responses in the comprehension experiment per language group and proficiency level for the conditions gaan+ inf, present tense and zijn+inf.

An ANOVA was carried out on the data to test whether there was an effect of proficiency level and first language on the responses obtained in each of the three conditions (gaan + inf, present tense and zijn + inf). An α level of .05 was adopted.

In Figure 1 it can be seen that all three groups of participants produced more target responses in the present tense and the zijn + inf conditions than in the gaan + inf condition, suggesting that comprehension of the prospective meaning of that construction is difficult for all groups.

The ANOVA shows an effect of proficiency level for the gaan + inf condition, F(1, 34) = 13,879, p = .001. Knowledge of the prospective meaning of gaan + inf apparently improves as a function of proficiency. However, the proportion of target responses for all groups and levels (the highest mean proportion being only slightly above .50 (= 50%)) reveals that none of the groups, not even the A2 participants, show a high degree of understanding of the meaning of that construction. There is for this condition no effect of language group, F(2, 34) = .406, p = .669. The interaction effect between ‘language group’ and ‘proficiency level’ was not significant, F(2, 34) = .139, p = .870.

In the ‘present tense’ condition the ANOVA yields a non-significant main effect of group, F(2, 34) = 2.330, p = .116. There is no effect for level either, F(1, 34) = 2.969, p = .094. No interaction effect was found, F(2, 34) = 2.001, p = .151.

Similar findings were obtained for the zijn + inf condition. There were no significant effects for this condition (group, F(2, 34) = 1.126, p = .336, level F(1, 34) = 2.289, p = .139, interaction F(2, 34) = 2.106, p = .137).

The accuracy of target response for the conditions ‘present tense’ and ‘zijn + inf’ is very high (reaching above 90% in both proficiency levels) for all three participant groups as can be seen in Figure 1. This suggests that participants consider both constructions as indicative of imperfective aspect. This conclusion is further supported by the observation that none of the participants hesitated in their choice when hearing the zijn + inf construction.

2 Narrative experiment

a Target responses

Figure 2 gives the mean proportions of target responses⁴ and their standard error bars split out for the three experimental conditions, the three learner groups, and the two proficiency levels. A reliability analysis gives high Cronbach’s alphas for the items in all three conditions; α = .900 for the prospective condition, α = .934 for the imperfective condition, and α = .926 for the perfect condition. These high values reveal that the tests are internally consistent with respect to target responses and that we may use the sum of target responses as an indicator for the test performance of the participants.

Figure 2.

Mean proportions and error bars (+/– 2SE) of target responses for the three conditions prospective, imperfective and perfect of the narrative task, split out for the three language groups and two proficiency levels.

As Figure 2 shows, the proportion of target responses in the prospective condition is fairly low, the A2 Arabic learners having the highest mean, which is just above 40%. An ANOVA analysis shows that the responses in the prospective condition show a significant difference among the three language groups, F(2, 34) = 16.092, p < .000, $η_{p}^{2}$ = .486 and between proficiency levels, F(1, 34) = 8.670, p = .006. $η_{p}^{2}$ = .203. In addition, there is an interaction effect between language group and proficiency level, F(2, 34) = 4.188, p = .024, $η_{p}^{2}$ = .198, the group of Turkish speakers showing extremely low scores (less than 5%) at both proficiency levels. The two other groups perform better (the group of Arabic learners performing the best) and the A2 participants have better results than the A1 participants.

The ANOVA for the imperfective condition yields a no significant main effect of group, F(2, 34) = 2.741, p = .079, $η_{p}^{2}$ = .495, but there is a significant effect of proficiency level, F(1, 34) = 33.385, p < .000, $η_{p}^{2}$ = .139. There is no interaction effect, F(2, 34) = 1.024, p = .370. The Turkish-speaking group produces the smallest percentage of target responses of all three groups. The A1 participants produce 10% target responses and the A2 participants 40%. The other two language groups do not differ much from each other in this condition, the A1 participants producing between 15% and 20% of target responses, and the A2 participants of both language groups, their target responses reaching approximately 75%. The error bars indicate large differences between the individual learners.

In the perfect condition there are significant main effects of both language group (L1), F(2, 34) = 16,464, p < .000, $η_{p}^{2}$ = .492, and proficiency level, F(1, 34) = 18,658, p = .000, $η_{p}^{2}$ = .354. An interaction effect was found between language group and proficiency level in this condition, F(2, 34) = 16,442, p < .000, $η_{p}^{2}$ = .492. The Arabic-speaking group shows a noteworthy improvement. The A1 participants produce less than 5% of target responses, while the A2 participants produce almost 60%. The Berber-speaking group, on the other hand, shows a small increase in the correctness of responses between the two levels. The Turkish group produces less than 5% correct responses at both levels.

b The use of dummy auxiliaries

Although in standard Dutch the constructions gaan + inf and hebben or zijn + ppart are correct to express respectively prospective and perfect aspect, we cannot be sure that the participants really wanted to express those aspects when using them. Despite this note of caution, it is important to realize that the construction gaan + inf is used 102 times in the prospective condition, and the construction hebben/zijn + ppart occurred 131 times in the perfect condition. This can be taken as an indication that at least part of the participants is aware of the prospective and perfect meaning of these constructions.

In order to estimate the extent of the use of dummy auxiliaries and to formulate hypotheses as to their function in the acquisition of finiteness, the question we will now try to answer is ‘How many and which dummy auxiliaries occurred in our data?’ We will start by looking at the narrative task. Table 4 presents an overview of the dummy auxiliaries used in that task and their frequencies of occurrence.

Table 4.

Narrative experiment: Dummy auxiliaries and their frequencies of occurrence in the three experimental conditions.

	zijn (‘to be’)	gaan (‘to go’)	doen (‘to do’)	hebben (‘to have’)
Prospective	36	n/a	1	0
Imperfective	73	45	4	3
Perfect	42	18	4	10
Total	151	63	9	13

As can be seen in Table 4, four different dummy auxiliaries were found in this experiment: zijn, gaan, doen and hebben. What strikes the eye in Table 4 is that, in all conditions, the frequencies of dummy zijn and dummy gaan are much higher than the number of dummy doen and dummy hebben. The difference between the quantity of dummy zijn and the dummy gaan is also noticeable, dummy zijn being used more frequently. The frequency of both dummy auxiliaries is the highest in the imperfective condition, providing evidence that part of the participants do not ascribe prospective nor perfect meaning to them.

The dummy auxiliaries doen and hebben were hardly used by the participants. This corroborates previous findings based on data of monolingual children (Julien et al., 2013) in which these two dummy auxiliaries were only marginally used. For this reason, the results presented henceforth concern only the dummy auxiliaries zijn and gaan, which are extensively used. So, this answers our first research question and confirms the prediction that dummy auxiliaries are used by adult DAL learners.

Two other research questions were whether proficiency level in Dutch and language background affect the use of those dummy auxiliaries. Two repeated measures ANOVA’s were conducted to compare the effect of proficiency level and language background on the frequency of use of the dummy auxiliaries zijn and gaan in the prospective, imperfective and perfect conditions.

The ANOVA for dummy zijn returned a significant effect of condition, F(2, 68) = 3,618, p = .032, $η_{p}^{2}$ = .096. A post hoc test, LSD, revealed a significant difference between conditions 1 (prospective; M = .90) and 2 (imperfective; M = 1.83), with more dummies in the imperfective. The effect of proficiency level was also significant, F(1, 34) = 4.924, p = .037, $η_{p}^{2}$ = .127 (fewer dummies for A2). There was no effect of language group. As for the dummy auxiliary gaan, the ANOVA showed only a significant effect of condition, F(1, 34) = 6.856, p = .013, $η_{p}^{2}$ = .168 (imperfective M = 1.13; perfect M = .45). There was no effect of language proficiency or language group.

So, in the narrative task, the prediction that language proficiency in Dutch influences the use of dummy auxiliaries is confirmed for dummy auxiliary zijn, but not for dummy gaan. The prediction that language background would influence dummy use was not confirmed for none of the dummy auxiliaries.

In the following section an analysis of the target responses in each completion experiment will be presented, followed by an analysis of dummy auxiliary use in these experiments.

3 Completion experiments

For various reasons such as absence on the test day – having to leave earlier, and tiredness – five of the participants did not perform one or more of the completion experiments. We therefore omit data of one Turkish-speaking participant in the ‘completion syntax’, the ‘present tense’ and the ‘past tense’ experiments, and of four participants (two Turkish-speaking, one Arabic-speaking and one Tarifiyt-speaking) in the ‘past tense’ experiment.

4 Target responses

a Completion experiment ‘present tense’

An analysis of items reliability was carried out for the completion task ‘present tense’, which revealed that the items have a high reliability α = .970, and that there are no deviant items. This was followed by an ANOVA to test the overall accuracy of the responses in this task.⁵ The results are illustrated in Figure 3, giving the mean proportions of target responses and their standard error bars split out for the three learner groups, and the two proficiency levels.

Figure 3.

Mean proportions and error bars (+/– 2SE) of target responses per language group and proficiency level in the completion experiment ‘present tense’.

In the completion task ‘present tense’ there are no significant main effects of language group, F(2, 28) = 1.930, p = .161. There was an effect of proficiency level, F(1, 28) = 36.662, p = .000, $η_{p}^{2}$ = .526. A1 participants of all three groups produce less than 10% of correct responses. A2 participants in the Tarifiyt- and Arabic-speaking groups produce approximately 60% correct responses and the Turkish-speaking group lags considerably behind with approximately 30% of correct responses. No interaction effect was found between language group and proficiency level in this condition, F(2, 28) = 2.334, p = .113.

b Completion experiment ‘past tense’

No statistical analysis was performed on the target data of this task, because participants produced only eight target utterances (out of 1,019) with this structure, indicating that this grammatical construction is not yet acquired by the majority of the participants.

c Completion experiment ‘syntax’

An analysis of items reliability was carried out for this task. The items have a high reliability α = .927, and there are no deviant items. Figure 4 shows the results of an ANOVA for the ‘canonical order’ condition. The results give the mean proportions of target responses and their standard error bars split out for the three learner groups, and the two proficiency levels.

Figure 4.

Mean proportions and error bars (+/– 2SE) of target responses in the experiment ‘completion syntax’ per language group and proficiency level for the condition ‘canonical order’.

In the ‘canonical order’ condition there was no significant effect of language group, F(2, 33) = 1.275, p = .293, but there was an effect of proficiency level, F(1, 22) = 24.596, p = .000, $η_{p}^{2}$ = .427. No interaction effect was found between language group and proficiency level, F(2, 33) = 1.343, p = .275.

No statistical analysis was carried out for the ‘inverted order’ condition, because of the 866 utterances produced in this condition, only 41 (4,4%) were correct.⁶ Analysis of the individual performances shows that those utterances with correct inversion were produced by nine participants, equally spread through the language groups (three Turkish speakers, three Tarifiyt speakers and two Arabic speakers). All, except one, were A2 participants and only one participant produced a relatively high number (16 out of 23) of correct sentences with inversion. The other eight participants produced that type of sentences sporadically, indicating that inversion was not easy to realize for both A1 and A2 participants.

5 The use of dummy auxiliaries

Table 5 presents an overview of the dummy auxiliaries used in the completion tasks. As can be seen in the table, the pattern is the same as that already observed for the narrative task: dummy zijn and dummy gaan are used much more often than dummies doen and hebben.

Table 5.

Completion experiments: Total number of dummy auxiliaries.

	zijn (‘to be’)	gaan (‘to go’)	doen (‘to do’)	hebben (‘to have’)
1. Present tense	147	144	6	15
2. Past tense	34	14	1	5
3. Syntax
• Canonical order	71	52	4	3
• Inverted order	61	74	5	4
Total	313	284	16	27

The sum of dummy auxiliaries used in the completion tasks and in the narrative task (see Table 4) gives the following results: a total of 464 occurrences of dummy zijn, 347 of dummy gaan, 25 of dummy doen and 40 of dummy hebben. Because the last two dummy auxiliaries are only sporadically used, they will be excluded from the analyses presented henceforth, which will focus on the extensively used dummy auxiliaries zijn and gaan.

It is striking that the number of dummy auxiliaries in the ‘past tense’ experiment is considerably lower than in the similar task eliciting the ‘present tense’. The construction ging + inf (go.past + inf), which is often used by children (Cornips 2000; De Jong 1999; Hollebrandse et al. 2013), was used only three times out of 14 dummy gaan constructions. The other 11 occurrences of dummy auxiliary gaan were in the present tense. Dummy zijn was used a few times in the past tense.

Table 5 shows further that the numbers of dummy auxiliaries in both conditions of the ‘syntax’ experiment do not differ much. What is interesting here, is that the 135 utterances (out of a total of 866) produced with a dummy auxiliary had three different surface forms. There were eight utterances with a dummy auxiliary verb in second position, that is, with the structure XV_dummySV_lexical inf, e.g. Hier is Pingu lopen (‘Here be.3sg Pingu walk.inf’). So, inversion was produced in those sentences, but with a dummy auxiliary. There were 57 utterances in which a dummy auxiliary was used, but inversion did not take place since the auxiliary verb was placed in the third position. These sentences had a structure of the type XSV_dummyV_lexical inf, e.g. Hier Pingu is lopen (‘Here Pingu be.3sg walk.inf’). There were also 70 utterances with a dummy auxiliary in which the subject of the sentence was omitted and, therefore, it was not possible to determine whether inversion had taken place. It is possible that omission of subject was caused by the difficulty of producing a sentence with an inverted subject, but it is also plausible that this is interference from the L1, since all three languages are pro-drop languages. Leaving out the subject leads to producing utterances like Hier (Pingu) is lopen (‘Here (Pingu) be.3sg walk.inf’), presumably with the underlying structure XSV_dummyV_lexical inf. This leads us to the conclusion that, at these two language proficiency levels, the syntactic structure underlying the dummy constructions with zijn and gaan, in both the canonical and the inverted orders, is the same, that is to say an IP projection, in which the adverbial is adjoined to the IP projection, _IP[ hier [_IP is [_VP lopen]]].

Two ANOVA repeated measures were carried out to compare the effect of proficiency level and language background on the use of dummy zijn and dummy gaan across the four tasks. The analysis regarding dummy auxiliary zijn shows a near significant effect of proficiency level, F(1, 27) = 3.691, p = .074 (HF corrected), $η_{p}^{2}$ = .114 (A1 more dummies). It also shows a task effect F(3, 81) = 3.388, p = .032 (Huynh–Feldt corrected), $η_{p}^{2}$ = .160 (more dummies in the completion task present tense).

Regarding dummy auxiliary gaan, the results show only a main effect of task, F(3, 81) = .855, p = .010 (Huynh–Feldt corrected), $η_{p}^{2}$ = .178 (more dummies in the completion task present tense). Although we can conclude from these results that there is a proficiency level effect for dummy auxiliary zijn, this effect is not the one expected. Instead of the predicted increase, the A2 participants show a decrease in use of that dummy. As for dummy auxiliary gaan there is no effect of proficiency level. This is a relevant finding for which we will present an explanation in the discussion section. Given that, in almost all tasks, a large amount of individual variation is observed, an analysis of the dummy auxiliary use by individual learners is presented in the next subsection.

6 Individual patterns of variation in dummy use

In order to account for individual variation, but still be able to see general patterns, participants were divided into five groups on the basis of their production of dummy auxiliaries:

extreme dummy users, those who use dummy auxiliaries 10 times or more often across all tasks;

occasional dummy users, who use a dummy auxiliary, up to nine times, spread throughout all tasks;

selective dummy users, who use dummy auxiliaries to various degrees, but only in one, two or three tasks;

infrequent dummy users, who use a dummy auxiliary only once across all tasks and;

no dummy users.

Table 6 displays the four types of dummy users and the no dummy users distributed over the two levels of language proficiency.

Table 6.

Use of dummy auxiliaries by five types of dummy users.

	Level	Extreme	Occasional	Selective	Infrequent	No dummy
Dummy zijn	A1	4	2	–	–	–
Dummy zijn	A2	1	3	–	–	–
Dummy gaan	A1	–	–	–	1	–
Dummy gaan	A2	–	–	2	2	–
Mixed	A1	4	1	4	–	–
Mixed	A2	6	4	3	–	–
No dummy	A1	–	–	–	–	2
No dummy	A2	–	–	–	–	1
Total	–	15	10	9	3	3
Turkish	–	9	0	6	2	3
Tarifiyt	–	4	3	3	0	0
Arabic	–	2	7	0	1	0

Table 6 shows that there were 15 extreme dummy users: five of them use dummy auxiliaries as often as 49 to105 times across the five tasks; 10 occasional dummy users; nine selective dummy users; three infrequent dummy users, and three no dummy users. Although five participants have not performed all tasks, their performance on the remaining tasks showed such clear tendencies for a particular type that it was possible to place them in one of the above categories.

Table 6 also shows that there are participants who use only dummy zijn, there are others who use only dummy gaan, and there is also a group that uses both dummy auxiliaries. Participants who use only dummy zijn are the extreme and the occasional dummy users, and most of them are A1 participants. Dummy gaan, when used as the only dummy auxiliary, is used by selective and infrequent dummy users, most of them being A2 participants. Two of the no dummy users are A1 participants, and one is an A2 participant. It is remarkable that two of those three participants are in the very beginning of their language acquisition process (evidenced by the almost exclusive use of nonfinite verbs), and one of them is in a more advanced stage (evidenced by the high level of accuracy in the use of finite verbs and only a few nonfinite verbs). This suggests that participants who use dummies extensively have already discovered that there must be a finite verb in the frontal part of the sentence. The dummy auxiliaries fulfill that function. Participants who do not use dummy auxiliaries can be divided into two groups: those who are not yet trying to figure out how the Dutch verbal system works, and those who have already discovered how it works. Table 7 clarifies these distinctions in stages of development. The 40 participants are classified according to their use of nonfinite verbs, dummy auxiliaries and finite verbs in the 130 items comprised in all three production tasks. In the nonfinite stage, fewer than 10 dummy auxiliaries are used, nonfinite lexical verbs are used more than 40 times, and finite lexical finite verbs are produced fewer than 15 times. In the dummy auxiliary stage, dummy auxiliaries are used more than 10 times across all experiments. At this stage, the number of nonfinite and finite verbs varies considerably, but the tendency is that nonfinite verbs are produced more often than finite verbs. In the finite stage, the dummy auxiliaries decrease – they are used fewer than 10 times across all experiments – and simultaneously the number of finite verbs in initial position is higher.

Table 7.

Three stages in the language development of Turkish, Berber Tarifiyt and Moroccan Arabic adult DAL learners.

Level	Dummy	Nonfinite stage (V_lex sentence-final); dummy auxiliary <10	Dummy stage (V_lex sentence-final); dummy auxiliary >10	Finite stage (V_lex sentence-frontal); dummy auxiliary <10
A1	No dummy	T11 (110/0/8)
		T17 (42/0/1)*
	Predominantly dummy zijn	T6 (95/6/9)	T1 (45/56/15)
		B2 (94/5/12)	T10 (73/30/2)
		B10 (105/6/5)	T19 (85/23/6)
		A5 (70/7/15)	B3 (35/28/25)
		A8 (91/8/7)	B5 (61/29/15)
			A11 (39/105/1)
	Predominantly dummy gaan		B7 (45/80/5)	A7 (33/1/60)
	Predominantly dummy gaan		B8 (87/14/7)
			A1 (33/77/7)
	Both dummy auxiliaries equally often	T21 (116/4/1)
A2	No dummy			T12 (32/0/51)
	Predominantly dummy zijn		T3 (56/36/10)	B1 (35/6/55)
			T7 (96/19/21)	A2 (16/3/48)
			T8 (29/15/59)	A3 (2/4/86)
			T13 (65/12/1)	A9 (3/3/82)
			T18 (16/32/5)
	Predominantly dummy gaan		T9 (31/16/22)	T2 (11/1/79)
			T14 (56/19/22)	T15 (13/1/53)
			B4 (23/60/25)	T16 (31/5/51)
				T20 (4/4/80)
				B6 (18/7/66)
				B9 (2/8/90)
				A4 (26/4/50)
				A6 (10/5/63)
	Both dummy auxiliaries equally often		T5 (51/49/16)
Total		8	18	14

Notes. The figures given in parentheses are: the number of nonfinite verbs, followed by the number of dummy auxiliaries, followed by the number of finite verbs across all experiments. * Data for this participant is missing for three of the five tasks; because he produced almost exclusively nonfinite verbs in the narrative task, we assume that, had he performed all the tasks, he would have produced nonfinite verbs much more often than 40 times. For this reason, this participant was placed in the ‘nonfinite stage’.

Table 7 shows that the nonfinite stage consists solely of A1 participants (8), and that the finite stage consists mainly of A2 participants (13), with the exception of one Arabic-speaking participant. The dummy stage, on the other hand, is composed of both A1 (9) and A2 participants (9). A considerable number of participants had passed the nonfinite stage and were in the dummy stage. Several A2 participants had mastered that stage, even though they still, sporadically, used dummy auxiliaries.

IV Discussion and conclusions

The aim of this study was to address the question as to whether dummy auxiliaries are used by adult DAL learners as a structural syntactic step towards the acquisition of finiteness. Four specific research questions were posed which will be discussed below.

Even though dummy auxiliaries have been attested in different earlier types of language acquisition (see introduction), it was necessary to find out whether the data collected for the present study corroborated this finding. We review each of the research questions:

1. Do low-proficient adult DAL learners use dummy auxiliaries?

The results reveal that all three groups of participants used dummy auxiliaries abundantly, in particular zijn and gaan (see Tables 5 and 6).

2. Does level of language proficiency in Dutch influence the choice and frequency of dummy auxiliaries?

It was predicted that the number of dummy auxiliaries produced by A2 participants would be higher than that produced by A1 participants. This prediction is not confirmed. There is indeed a level effect, for dummy auxiliary zijn. However, instead of an increase, the A2 participants show a decrease of dummy zijn across all tasks. As for dummy auxiliary gaan, no level effect was found. It seems that the majority of participants in the present study is at a more advanced stage than predicted on the basis of their CEFR level and the results of earlier studies. The starting point in other studies (most of them being longitudinal), such as that of Van de Craats (2009), was the very beginning of being exposed to target language instruction. Van de Craats reported that her participants started using dummy auxiliaries after nine months of instruction. In the present study, at the moment of testing, the majority of the participants (31) had been attending Dutch lessons for a longer period of time than those in Van de Craats’ study. It is, therefore, conceivable that they had already gone through the first phases reported in Van de Craats’ study.

There are strong indications that the developmental path these learners follow is the same as that reported in the studies mentioned above. Moreover, the present study introduces a new finding. It seems that there are two phases within the dummy stage: an initial phase in which both dummy zijn and gaan are used, and a subsequent phase in which the occurrence of dummy zijn diminishes while dummy gaan keeps being used. The conclusion is that level of language proficiency in Dutch influences the use and choice of dummy auxiliaries.

Utterances like Pingu is gaat schoppen (‘Pingu be.3sg go.3sg kick’), Zeehond is … gaat eet (‘Seal be.3sg … go.3sg eat.stem’), produced by a Turkish A2 participant (other participants show the same pattern), provide evidence that at the gaan-phase, participants seem to have realized that zijn is no longer allowed in that construction.

3. Does language background determine the choice of dummy auxiliary?

Although the results of the target responses show significant differences among the three language groups in the prospective and perfect conditions of the narrative task, the prediction that language background influences the choice of dummy auxiliary was not confirmed.

All three groups behave similarly, and they all use the same dummy auxiliaries. This indicates that the choice of dummy auxiliaries is influenced by target language input rather than by language background. On the other hand, the prediction that Arabic-speaking learners would perform better than the other two groups on the prospective condition of the narrative experiment was confirmed. It seems that positive transfer helps the Arabic-speaking group to acquire more easily the prospective use of gaan + inf.

4. Do adult DAL learners, in the initial stages of language acquisition, assign meaning to the constructions zijn + inf and gaan + inf?

The results from the comprehension experiment show that all participants match the picture portraying the ongoing action to the construction zijn + inf and often also to the construction gaan + inf. This observation has also been made respectively by Verhagen (2013) and Zuckerman (2001). However, while those two researchers interpret this as evidence that learners ascribe imperfective reading to those constructions, we see these choices as evidence that, in the initial stages of language acquisition, no meaning is given at all to those two constructions. This interpretation is corroborated by the fact that the participants in this study produce both gaan + inf and zijn + inf arbitrarily to describe all three conditions (see Table 4), and often add lexical means to express aspect. Aspect is not yet expressed by grammatical, but rather by lexical means. That explains why gaan + inf may surface in constructions expressing accomplished actions such as Penguin gaat hier klaar met maken (‘Penguin go.3sg here ready with make.inf’).

Additional evidence that no meaning is attached to dummy auxiliaries is provided by the lack of any significant effect of lexical aspect of the verbs tested in the experiments. Dummy auxiliaries are used indiscriminately with all verb classes, regardless of their semantic constraints. The learners in the present study do not comply with the constraints that statives in Dutch do not allow the continuous constructions zijn + aan het + inf and zitten/staan/liggen + te + inf, and that transitives and most intransitives require the use of auxiliary hebben (‘have’) in Dutch. If zijn + inf carried imperfective meaning, and was, as suggested in the literature (see Blom 2003: 70), the precursor of the prepositional infinitival construction zijn + aan het + inf, it would not occur, or would occur less, with stative verbs in the imperfective condition. Similarly, if perfect meaning was attached to it, it would not appear with transitives and intransitives in the perfect condition.

Our conclusion is that dummy auxiliaries are semantically vacuous. They have a purely structural function, in line with Chomsky’s (1995) Economy Principle. They come from the Dutch input learners get, in which the copula zijn and the construction gaan + inf occur frequently. Learners use them as devices to mark the position of the finite verb prior to the movement of the lexical verb. This happens independently from the L1 background of the adult learners. The dummy auxiliaries zijn and gaan carry finiteness features such as person and number, and occur in the same slot that the finite verb will eventually occupy. They may, therefore, be considered predecessors of movement of the lexical verb. They trigger the marking of the syntactic relation between V and I.

Footnotes

Acknowledgements

We are indebted to the makers and the copyright holders of the DVD Pingu voor altijd (Pygos Group & Hit Entertainment Limited, 2010), from which we extracted the clips we used in eliciting the data for this study.

Declaration of Conflicting Interest

The authors declare that there is no conflict of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Blom

(2003) From root infinitive to finite sentence: The acquisition of verbal inflections and auxiliaries. Utrecht: LOT.

Blom

De Korte

(2008) De verwerving van het Nederlands: Dummies en Verb second [The acquisition of Dutch: Dummies and Verb second]. Nederlandse Taalkunde 13: 133–59.

Blom

De Korte

(2011) Dummy auxiliaries in child and adult second language acquisition of Dutch. Lingua 121: 906–19.

Chomsky

(1995). The minimalist program. Cambridge, MA: MIT Press.

Cornips

(2000) The use of gaan + infinitive in narratives of older bilingual children of Moroccan and Turkish descent. In: de Hoop

Van der Wouden

(eds) Linguistics in the Netherlands. Amsterdam/Philadelphia, PA: John Benjamins, 57–67.

Cornips

(2013) Child use of auxiliary + infinitive in Dutch: Acquisition device or reflection of the input. In: Blom

Van de Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 369–94.

Council of Europe (2001) A Common European Framework of Reference for Languages: Learning, teaching, assessment. Cambridge: Cambridge University Press.

Crul

Doomernik

(2003) The Turkish and Moroccan second generation in the Netherlands: Divergent trends between and polarization within the two groups. International Migration Review 37: 1039–64.

De Jong

(1999) Specific language impairment in Dutch: Inflectional morphology and argument structure. PhD thesis. Groningen: GRODIL.

10.

Den Besten

(1989) Studies in West Germanic syntax. Amsterdam: Rodopi.

11.

Dimroth

Gretsch

Jordens

Perdue

Starren

. (2003) Finiteness in Germanic languages: A stage-model for first and second language development. In: Dimroth

Starren

(eds) Information structure, linguistic structure, and the dynamics of acquisition. Amsterdam: Benjamins, 65–93.

12.

El Aissati

(1994) Nessawal Tamazight (Tarifiyt): A basic course book in Berber (Tarifiyt). Nijmegen: University Press.

13.

El Aissati

(2001) Word order change in Berber. In: Ibriszimow

Vossen

(eds) Etudes berbères Frankfurter Afrikanistische Blätter [Berber studies:. The Frankfurt African Studies Bulletin]. Cologne: Rüdiger Köppe, 13: 33–42.

14.

El Aissati

Boumans

Cornips

Dorleijn

Nortier

(2005) Turks- en Marokkaans-Nederlands [Turkish and Moroccan Dutch]. In: Van der Sijs

(ed.) Wereld nederlands: Oude en jonge variëteiten van het Nederlands [World Dutch: Old and new varieties of Dutch]. Den Haag: Sdu Publishers, 149–83.

15.

E-Rramdani

(2003) Acquiring Tarifit–Berber by children in the Netherlands and Morocco. Unpublished PhD thesis, University of Tilburg, Netherlands.

16.

Flecken

(2011) What native speaker judgements tell us about a progressive aspectual marker in Dutch. Linguistics 49: 479–524.

17.

Fleta

(2003) Is-insertion in L2 grammars of English: A step forward between Developmental stages? In: Liceras

Zobl

Goodluck

(eds) Proceedings of the 6th Generative Approaches to Second Language Acquisition conference (GASLA 2002). Somerville, MA: Cascadilla Proceedings Project, 85–96.

18.

García

Del Pilar

Ibarrola

Liceras

(2005) Placeholders in the English interlanguage of bilingual (Basque–Spanish) children. Language Learning 55: 445–89.

19.

Göksel

Kerslake

(2005) Turkish: A comprehensive grammar. New York: Routledge Taylor & Francis Group.

20.

Haberzettl

(2003) ‘Tinkering’ with chunks: Form-oriented strategies and idiosyncratic utterance patterns without functional implications in the IL of Turkish speaking children learning German. In: Dimroth

Starren

(eds) Information structure and the dynamics of language acquisition. Amsterdam: Benjamins, 45–64.

21.

Harrell

(1962) A short reference grammar of Moroccan Arabic. Washington DC: Georgetown University Press.

22.

Hollebrandse

Van Koert

Van Hout

(2013) Semantic dummy verbs in child Dutch. In: Blom

Van de Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 75–100.

23.

Hoogland

(1996) Marokkaans Arabisch: Een cursus voor zelfstudie en klassikaal gebruik [Moroccan Arabic: A course for self-study and classroom use]. Amsterdam: Uitgeverij Bulaaq.

24.

Huebner

(1989) Establishing point of view: The development of coding mechanisms in a second language for the expression of cognitive and perceptual organization. Linguistics 27: 111–43.

25.

Huebner

Carroll

Perdue

(1992) The acquisition of English. In: Klein

Perdue

(eds) Utterance structure: Developing grammars again. Amsterdam: Benjamins, 1–121.

26.

Hulk

Cornips

(2005) Differences and similarities between L2 and (2)L1: Do-support in child Dutch. In: Dekydtspotter

Sprouse

Liljestrand

(eds) Proceedings of the 7th Generative Approaches to Second Language Acquisition conference (GASLA 2004). Somerville, MA: Cascadilla Proceedings Press, 163–77.

27.

Jordens

Dimroth

(2006) Finiteness in children and adults learning Dutch. In: Gagarina

Gülzow

(eds) The acquisition of verbs and their grammar: The effect of particular languages. Dordrecht: Springer, pp.173–200.

28.

Julien

Van de Craats

Van Hout

(2013) There is a dummy ‘is’ in early first language acquisition. In: Blom

Van de Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 101–40.

29.

Klein

(1994) Time in language. London: Routledge.

30.

Koster

(1975) Dutch as an SOV language. Linguistics Analysis 1: 111–36.

31.

Lalleman

(1986) Dutch language proficiency of Turkish children born in the Netherlands. Dordrecht: Floris.

32.

MacWhinney

(2000) The CHILDES project: Tools for analysing talk. Mahwah, NJ: Lawrence Erlbaum.

33.

Pygos Group & Hit Entertainment Limited (2010) Pingu voor altijd! [Pingu forever] [DVD]. Switzerland: Pygos Group and Hit Entertainment Limited.

34.

Roes

(2008) Facts and figures of the Netherlands: Social and cultural trends 1995–2006. The Hague: Sociaal en Cultureel Planbureau (The Netherlands Institute for Social Research), The Netherlands Institute for Social Research.

35.

Scheele

Leseman Mayo

(2010) The home language environment of monolingual and bilingual children and their language proficiency. Applied Psycholinguistics 31: 117–40.

36.

Schimke

(2013) Dummy verbs and the acquisition of verb raising in L2 German and French. In: Blom

Van

Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 307–38.

37.

Schneider

Eschman

Zuccolotto

(2001) E-Prime user’s guide. Pittsburgh, PA: Psychology Software Tools.

38.

Starren

(2001) The second time: The acquisition of temporality in Dutch and French as a second language. Unpublished PhD thesis, Tilburg University, Netherlands.

39.

Tracy

(2002) Growing (causal) roots: All children start out (and may remain) multilingual. Linguistics 40: 653–86.

40.

Van de Craats

(2009) The role of IS in the acquisition of finiteness by adult Turkish learners of Dutch. Studies in Second Language Acquisition 31: 59–92.

41.

Van de Craats

Van Hout

(2010) Dummy auxiliaries in the L2 acquisition of Moroccan learners of Dutch: Form and function. Second Language Research 26: 473–500.

42.

Van Kampen

(1997) First steps in wh-movement. Delft: Eburon.

43.

Vendler

(1957) Verbs and times. The Philosophical Review 66: 143–60.

44.

Verhagen

(2009) Finiteness in Dutch as a second language. Unpublished PhD thesis, Free University of Amsterdam, Netherlands.

45.

Verhagen

(2011) Verb placement in second language acquisition: Experimental evidence for the different behaviour of auxiliary and lexical verbs. Applied Psycholinguistics 32: 821–58.

46.

Verhagen

(2013) From dummy auxiliary to auxiliary in Moroccan adult learners’ production and comprehension of Dutch. In: Blom

Van de Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 281–306.

47.

Zobl

(2002) Multiple subject constructions in Japanese and the development of Agr in L2 English. In: Foster-Cohen

Ruthenberg

Poschen

(eds) Eurosla Yearbook (Volume 2, pp. 29–47). Amsterdam: Benjamins.

48.

Zuckerman

(2001) The acquisition of ‘optional’ movement. Unpublished PhD thesis, University of Groningen, Netherlands.

49.

Zuckerman

(2013) Dummy auxiliaries in Dutch first language acquisition. In: Blom

Van de Craats

Verhagen

(eds) Dummy auxiliaries in first and second language acquisition. Berlin: De Gruyter Mouton, 39–73.

50.

Zwart

CJ-W

(1997) Morphosyntax of verb movement: A minimalist approach to the syntax of Dutch. Studies in Natural Language and Linguistic Theory, volume 39. Dordrecht: Kluwer Academic.