The Interface Hypothesis and the Vulnerability Hypothesis
tested in situations of comparable frequency
Hans A. Wilke
University of Amsterdam
The current paper investigates the predictions made by the Interface Hypothesis (IH) and the Vulnerability Hypothesis (VH) for L2 learners at near-native level when frequency is taken into consideration. Frequency is not formally considered important for either hypothesis in determining which types of grammatical structures are most vulnerable to cross-linguistic influence. The IH poses that syntactic structures that require an interface with Information Structure (IS) are most vulnerable. The VH poses that syntactic structures that occur in variable distributions are most vulnerable.
To test these hypotheses together with frequency added as a factor, two constructions were chosen that occur comparably frequently in the English language: cleft constructions (CC) and relative clauses with a passive and an infinitive (RC). The first one is an example of a syntactic construction that interfaces with IS (externally), and the second a syntactic construction that does not interface externally. Both constructions occur in variable
distribution, but the RC’s are more variable (29,2%) than the CC’s (17.3%). The VH therefore predicts these to be more vulnerable to crosslinguistic interference. Production of these constructions is measured by two sentence completion tasks: one to elicit CC’s, and one to elicit RC’s. Both tests were done by 38 Dutch students of the English language. 37 native speakers of English acted as a control group.
With frequency taken into consideration, the VH’s prediction that more variability leads to a greater difference between L2 speakers and natives is still borne out in the data. There is a significant difference between natives and near-natives regarding production of the RC’s (P=.008), while no significant difference is measured for this in the CC task. This is opposite from what the IH would predict. Incidentally, age proved to be a significant factor in the RC task (P=.03) while It was not observed to be significant in the CC task. These results are discussed in addition to how frequency could be incorporated into these hypotheses to make more sophisticated predictions.
Keywords: Interface Hypothesis, Vulnerability Hypothesis, information structure, second language acquisition, frequency, transfer
INTRODUCTION
How successful individual people will be at acquiring a second language is determined by many factors: motivation, age, cognitive ability, linguistic distance between L1 and L2, access to the L2, quality of instruction, and many more (Loewen, 2015). These are all examples that play an especially crucial part at the first stages of learning, when the learner is still
struggling. They keep having a noticeable influence up until the point when the learner has reached the near-native level (Loewen, 2015). Whereas younger L2 learners still manage to reach the native level (Johnson & Newport, 1989; Krashen, Long & Scarcella, 1979; Patkowski, 1980) adults plateau: they rarely reach native level (Patkowski, 1980; Sorace, 2003). An explanation for this phenomenon is not likely to be found by examining individual speakers since it holds generally across adults who speak an L2 at near-native level. In the current study I will be analysing two hypotheses that explain this phenomenon. According to Sorace, the answer is found in language itself (Belletti, Bennati & Sorace, 2007; Sorace & Filiaci, 2006). She developed the Interface Hypothesis (IH, Sorace, 2011) which – among
other things – attempts to explain where near-native speakers are different from native speakers, and why. It predicts near-natives will most likely be different from natives when it concerns grammatical structures that lie at the interface between syntax and
discourse/pragmatics. Another more recent hypothesis, the Vulnerability Hypothesis (VH, de Prada Pérez, 2018) was developed by de Prada Pérez. It challenges the IH in that it does not assume interface to be a problem for near-natives. It instead proposes that difficulty arises in contexts of variable distribution, where several grammatical constructions are interchangeable to denote the same meaning. The IH has gained a footing in research on bilingualism and second language acquisition (SLA), but it has also received criticism from many different scholars since its first mention (Sorace, 2012). The VH is a response to the IH and attempts to solve some of the problems scholars have with the IH.
The current study will test both hypotheses by comparing native English speakers to Dutch students of English on two different sentence completion tasks. What sets the currents study apart from previous research done on both the IH and the VH is that the grammatical constructions that are compared are evenly matched with respect to frequency and potential transfer. These two factors are considered important in determining which elements of a language will be learned more successfully. For frequency, see R. Ellis (1994) and Hulstijn (2005), for transfer, see Montrul (2014) and Odlin (1989). The results therefore will not only show how well both hypotheses fare in comparison to each other, but also whether adding more factors to the has an effect on their reliability. This might help us better understand how and why second language learners at near-native level are different from natives. In the section below, the IH, VH, frequency and transfer will be discussed in more detail, as well as their relation to linguistic complexity. After that the research questions will be described, followed by the methodology, the results, discussion and conclusion.
BACKGROUND
The Interface Hypothesis
The term ‘interface’ is used to denote that two domains, e.g., syntax and discourse, are working together in order to produce a grammatical construction (Sorace, 2011; Tsimpli & Sorace, 2006). A cleft construction is an example of such a grammatical construction (1):
(1) It is the American Culture that I find interesting.
(Verheijen, Los & de Haan, 2013: 98)
This construction provides additional focus to the constituent ‘American Culture’ by ‘clefting’ it to the first position where it is represented by ‘it’. This is an example of an it-cleft. Other types of clefts are wh-clefts and pseudo clefts. All versions of the cleft constructions are used to assign syntactical focus to a constituent (Lambrecht, 2001). As a grammatical object, the constituent ‘American culture’ cannot receive syntactical focus without using a cleft
constructions. When no cleft construction is used, the sentence would appear as follows (2):
(2) I find the American culture interesting.
Without using a cleft construction, the constituent ‘ American culture’ does not appear in the first position, and does not receive syntactical focus. The cleft construction in example (1) fulfils an information-structural need: assigning focus. To achieve this syntax and information structure are required to interface, whereas the construction in example (2) does not interface with Information Structure (Verheijen et al., 2013).
The IH originally is a theory of bilingual language development (Sorace & Filiaci, 2006). It builds on the work done by Hulk and Müller (2000; Müller, 2000; Müller & Hulk, 2000, 2001) which analyses the role the interface between syntax and pragmatics plays in the L1 acquisition of bilinguals. The IH is first mentioned by Sorace and Filiaci (2006) where it proposes that syntactic constructions that do not interface are fully acquirable by second language learners, but that syntactic constructions with an interface between syntax and another cognitive domain will be problematic. This hypothesis is later made more specific for each domain it applies to: L2 acquisition, L1 attrition and bilingual L1 acquisition (Sorace, 2011). These domains that can be seen as operating outside of grammar can be captured under the umbrella term Information Structure (IS, Chafe, 1976; Krifka, 2008). Lambrecht (1994) says the following about IS:
“Information-structure analysis is centered on the comparison of semantically equivalent but formally and pragmatically divergent sentence pairs (…).”
(Lambrecht 1994: 6)
This observation that IS involves dealing with pragmatically divergent sentence pairs is apparent in residual optionality as well. Residual optionality is the phenomenon that a specific syntactic construction in the target language that interfaces with IS is not produced when it is not available in the L1. It instead is replaced by a construction that does not interface, or a construction that is available in the L1 (Sorace, 2000, 2005; Sorace & Serratrice, 2009). This does not mean that the syntactic construction in question is never produced in the L2, but that there is more variation in when it is used compared to native speakers. Residual optionality is the core of the problem for near-natives according to the IH. The following examples (3 & 4) are taken from Belletti et al. (2006: 667):
(3) L’anziana signora salute la ragazza, quando lei attraversa la strada.
‘The old lady says hello to the girl, when she crosses the road.’
(4) Appena lui chiude la borsa, il fattorino dà il denaro al cassiere.
‘As soon as he closes the bag, the deliveryman gives the money to the cashier.’ In this study, native speakers of Italian were compared to English speakers of Italian at near-native level. They were tested on their interpretation and use of overt subject pronouns and null subjects in situations where their production involved an interface between syntax and IS. When presented with sentences with overt subject pronouns in anaphoric contexts (3 & 4), the native speakers rarely interpreted the personal pronoun lei/lui ‘she/he’ to be coreferential with the subject of the matrix clause (5%). The near-natives interpreted it like this 30% of the time. In the same study, production of overt pronominal subjects in Italian was also analysed by means of a story telling task (spontaneous speech). The natives used overt pronominal subjects 4% of the time, the near-natives 14% of the time. The natives opted for null-subjects more often than the near-natives with English as an L1; a language in which null-subjects do not occur. This is in line with the predictions made by the IH. Sorace further explains this phenomenon by the computational complexity1 of syntactic constructions that interface with
IS (Sorace 2012). Constructions that do not interface with IS require less computation than constructions that do. These constructions are therefore less likely to be the first choice for L2 speakers as they will – consciously or subconsciously – follow the path of least resistance.
1 Computational complexity is a topic deserving its own chapter, but as this would diverge from the topics at hand I will refer readers interested in reading more about this to consult Szymanik (2016). Computability theory is one of the main topics of the book, with computational complexity specifically receiving attention in the second chapter.
The Vulnerability Hypothesis
Multiple scholars have voiced their criticism of the IH. A collection of critiques as well as Sorace’s reply to these can be found in Sorace (2012). Most interestingly for the topic of the current study are those that point out complexity as a factor would be more productive in predicting which constructions are more problematic for second language learners than interface. According to Hopp (2011) and Pires and Rothman (2011) the problem is not that the IH foregoes complexity as a relevant factor, but that it defines complexity by interface only. Pires and Rothman (2011) even question whether the syntax-pragmatics interface (an external interface) actually is the most problematic. They give examples where the syntax-semantics interface (an internal interface) can be viewed as equally if not more problematic. A response to the IH in the form of a novel hypothesis is the Vulnerability Hypothesis (de Prada Pérez, 2018). The VH does not define complexity by interfaces but by the position of a construction along the variability continuum (de Prada Pérez, 2018: 3). On the leftmost side of the variability continuum are categorical distributions. These are forms that occur close to 0% or 100% of the time in a certain context. For example, in French a sentence containing the particle ne introducing negation, always requires there to be negation marker like pas ‘not’ or
personne ‘nobody’. This is an example of categorical distribution. However, a sentence in
French containing such a negation marker (pas, personne, etc.), does not necessarily require the presence of the particle ne (McMahon, 1994: 161–164). The production of the particle ne in this context is optional, therefore more complex, and an example of variable distribution. On the other side of the spectrum are highly variable contexts where a form occurs (near) 50% of the time. These are considered most complex and most problematic for second language learners. Unfortunately, an example of such a construction is not given/found by de Prada Pérez (2018). The variability continuum is shown in Figure 1.
FIGURE1 | Variability continuum, adapted from de Prada Pérez (2018)
The biggest difference between the IH and the VH is that the VH defines complexity less rigidly: there is a spectrum and different constructions occur on different parts of the spectrum regardless of their interface (or lack thereof). The IH defines certain categories (interfaces) complex, and others more simple; there are no proposed differences between structures that belong to the same category of complexity. The key similarity between the two hypotheses – at least for the current study – is that influences outside of interface (for the IH) and outside of the variability continuum (for the VH) are not considered. Leaving them both vulnerable to critique from scholars who wonder why complexity is defined with just one determiner.
Frequency
Frequency is typically short for input frequency when discussed within the realm of SLA (Bybee & Hopper, 2001; Loewen, 2015). Input frequency regarding a specific grammatical construction can mean how often a certain construction is discussed/mentioned in a method used for instructed SLA, or how often a teacher discusses/mentions this construction (N. Ellis,
2002; Gass & Mackey, 2002; Loewen, 2015). This is a very active way of providing input. On a more passive level, input frequency of a specific grammatical construction means, for example, how often this construction appears in a newspaper the language learner is reading in the target language, or how often the people around her/him use this construction in their day to day speech. The part of language-learning that involves instruction (be it by an actual instructor or by the chosen method guiding the learner) will not showcase the same frequency input of target constructions as the more passive methods mentioned above. The frequency of a specific grammatical construction in a newspaper will be a far more accurate representation of how often this construction generally appears in this language (in written form) than the frequency of this construction in a teaching method. There is often a direct relation between input frequency and how well constructions are acquired: more frequent constructions are more likely to be acquired easily. This is true in the case of instructed SLA (explicit learning) (N. Ellis, 2002; N. Ellis & Collins, 2009; R. Ellis, 1994; Hulstijn, 2005) as well as in cases of incidental (implicit) learning (Denhovska, Serratrice & Payne, 2016; Hulstijn, 2005). The Associative-Cognitive CREED model (N. Ellis, 2006) suggests that the process of L2 acquisition is no different from other processes involved with acquiring information in general. If this is the case then – similarly to other information acquiring processes – high-frequency constructions will be learned more easily than constructions that are (relatively) infrequent. In connection to the Associative-Cognitive CREED model this means that constructions and/or elements that co-occur with the presented input will be picked up and remembered more productively by association. This is something that previous research has shown humans to be sensitive to (N. Ellis, 2002; Lieven, 2010; Saffran, 2003; Saffran, Newport, Aslin, Tunick & Barrueco, 1997). Additionally, constructions that are more
complex tend to be less frequent (Haspelmath, 2008). This suggests that constructions that are problematic according to the IH and the VH are likely to be infrequent, and also acquired less well because of this.
Frequency will play a crucial part in the current study. The above mentioned hypotheses make predictions based on the grammatical ‘status’ of constructions. The IH predicts
grammatical constructions requiring an interface between syntax and pragmatics to be more difficult for learners than grammatical constructions that do not require an interface (Sorace, 2011; Sorace & Filiaci, 2006). The frequency of investigated constructions is not considered. This means that any effect of frequency would be interpreted as an effect of interface (or lack thereof). The VH predicts that constructions that are more variable are more susceptible to cross-linguistic influence, whereas constructions with categorical distribution will be more impervious to cross-linguistic influence (de Prada Pérez, 2018). Similarly to the IH, potential effects of frequency are not considered and effects found are attributed to the type of
distribution (variable or categorical). Both the IH and the VH rely on a comparative approach of the data (Tagliamonte, 2002): different subpopulations (natives, learners, bilinguals) are compared as well as different grammatical constructions (with or without an (external) interface, variable/categorical). The present study removes the risk of this potentially incorrect interpretation by comparing natives and near-natives on production of constructions of
comparable frequency.
Transfer
Transfer, or as it also known: linguistic interference or crosslinguistic influence, is used to define the phenomenon where speakers apply knowledge from one language to another language (Odlin, 1989; Weinreich, 1953). Transfer is most often discussed in a context of negative transfer; when a speaker transfers an item from one language into another where it is not acceptable. This usually happens when the speaker’s command of the L2 is not at a very high level (Montrul, 2014). The more proficient a speaker is in the L2, the less likely it is that negative transfer will occur. In the case of the current study, as well as the hypotheses
discussed, the speakers are assumed to have a high proficiency in their L2. It is therefore more likely for positive transfer to occur than negative transfer. In cases of positive transfer, the speaker applies knowledge from her/his L1 to the L2 in situations where this does not cause any problems of acceptability. This situation was briefly touched upon while discussing the IH above. The IH in its definition of residual optionality includes potential transfer as a factor in how likely speakers are to produce a certain construction in their L2. Situations where a speaker does not have the intended construction available in her/his L1 are potentially subject to positive transfer (Sorace, 2000, 2005; Sorace & Serratrice, 2009). Positive transfer is not always visible, but in the case of the IH it plays an important role. Positive transfer enables the speaker to (consciously or subconsciously) avoid constructions in the L2 that are more complex as defined by the IH (Sorace, 2012). The VH incorporates transfer as a factor in a similar way the IH does. Situations that are more complex as defined by the VH: variable contexts, are more likely to be subject to transfer (de Prada Pérez, 2018). Transfer as a factor in this study therefore is to be considered a potential result of linguistic complexity. Both the IH and the VH agree that increased complexity can cause a speaker to draw upon her/his L1 knowledge to produce a grammatical construction in the L2.
THE STUDY
The present study will contribute to existing research done on the IH and the VH. This is done by comparing native speakers of English to Dutch students of the English language who belong to the category of “advanced learners” (Callies, 2009). They are compared on the written production of two different grammatical constructions: it-cleft constructions (CC), and relative clauses with a passive and an infinitive (RC). Both these constructions are more common in the written register than in the spoken register (CC: Lambrecht, 2001; Velleman, Beaver, Destruel, Bumford, Onea & Coppock, 2012; Verheijen et al., 2013; RC: Simon, 2013). Only it-clefts (CC) were considered as other cleft constructions are not necessarily considered typical for the written register (Lambrecht, 2001; Verheijen et al. 2001). CC’s are syntactic constructions that involve an interface between syntax and IS, whereas the RC’s do not involve any external interface. The CC’s and RC’s are both variably distributed, but CC’s are closer to the categorical distribution end of the spectrum than RC’s. As a result, the IH and the VH predict opposite results regarding differences between natives and near-natives on these constructions: the IH predicts the CC’s to be more difficult, the VH the RC’s. The production of the RC’s is not subject to potential transfer from Dutch: this specific
construction does not exist in this same form in Dutch. Because CC’s are few and far between in the Dutch language, a transfer effect from Dutch was not expected (van der Beek, 2003). Unfortunately, an account of exactly how frequently they occur in Dutch was not available at time of the current study. To ensure the production of CC’s by the Dutch students would not be influenced by positive transfer an additional task was created (CCT-NL, see below) which measured the production of CC’s in the Dutch of native Dutch speakers. To account for frequency, Roland, Dick and Elman (2007) was consulted when choosing constructions for this study. They carried out a study on frequency of syntactic constructions in different written corpora. Two consulted corpora were checked for both CC’s and RC’s: The Wall Street Journal Corpus and the Switchboard Corpus. The following frequencies were observed (Table 1):
TABLE1 | Frequency of construction per million NP’s
Wall Street Journal Corpus 406 421
Switchboard Corpus 288 88
CC = Cleft Constructions, RC = Relative Clauses (with an infinitive and a passive)
The difference in frequency is due to the nature of these corpora: Wall Street Journal Corpus is a corpus consisting of written texts. Switchboard Corpus is a collection of telephone conversations. Because the current study focusses on the written register, the frequencies observed in the Wall Street Journal Corpus – which are very close to equal – are most
relevant. Accounting for frequency is not something that has been done before in either IH or VH-related research. The results will provide new insights into how these hypotheses could benefit from considering frequency. All experimental research done in the current study was approved by the ethics committee of the Faculty of Humanities (FGW).
Research Questions and Predictions
The study aims to answer the following questions:
(i) Do the predictions of the Interface Hypothesis and the Vulnerability Hypothesis hold when frequency is taken into consideration?
To test the effect of frequency in combination with both hypotheses, the constructions compared will be of comparable frequency in the L2. Neither the IH nor the VH take frequency into consideration when making predictions, but frequency is generally considered a productive predictor in L2 acquisition. Because of this, the current study predicts that frequency will cause the predictions of both hypotheses to no longer hold. Both the IH and the VH have been tested, for example, by comparing the production of overt subject pronouns and null subjects in null subject languages, by natives and near-natives (IH, Belletti et al., 2006, VH, de Prada Pérez, 2018). There is a big difference between the frequency of occurrence of overt subject pronouns and null subjects in all languages tested (Italian, Spanish, Catalan). This could have skewed the results.
(ii) How can the Interface Hypothesis and the Vulnerability Hypothesis be improved to make more reliable predictions?
Under the assumption that frequency plays a role (as is predicted) and the IH and/or the VH prove to be less reliable when it is taken into consideration, the hypothesis/es affected require adjustment. Suggestions as to how this could be done will be made depending on the results.
MATERIALS AND PROCEDURE Participants
For the main tasks, there were a total of 75. The native English speakers (n=37) were recruited via different channels: Facebook, social circle and the University of Amsterdam. They are attending, or have a degree in a higher education. The near-natives (Dutch participants, n=38) were recruited from two Dutch universities – Radboud University
(Nijmegen) and Utrecht University – where they were participating in a BA English Language program at the time of testing. All of the participants were asked to disclose their style of English, only participants who judged their English as only British were considered for the statistical analysis. A total of 20 participants were excluded from the results because of this: 13 native English speakers and seven near-natives. Age mean, standard deviation, and range for the included participants are given in Table 2. Additionally, for the CCT-NL there was a total of 20 Dutch participants. These participants were all recruited through my social circle. Participants were only considered if they grew up speaking only Dutch. Age mean, standard
deviation, and range are given in Table 3. Two participants were excluded from the results: one was disqualified for not following the instructions properly, the other for being raised bilingually (Dutch and English).
TABLE2 | Means, standard deviations and range for main task participant age
Mean (SD) Range
Age overall (n=55) 27 (11) 18-54
Age English natives (n=24) 36 (12) 18-54
Age Dutch students of English (n=31) 20 (2) 18-29 TABLE3 | Means, standard deviations and range for secondary task participant age
Mean (SD) Range
Age Dutch natives (n=18) 37 (16) 22-70
Measurement Instruments
Three different sentence completion tasks were designed: the Cleft Construction Task (CCT), the Relative Clause (with an infinitive and a passive) Task (RCT), and the CCT-NL, which is the CCT translated into Dutch. All these tasks were sentence completion tasks consisting of seven items and were designed to elicit a response in the written register. The items contained a contextual element consisting of a paragraph providing background information needed to complete the sentence completion elements. The CCT and the RCT were included in a
statistical analysis, for the CCT-NL only descriptive statistics were carried out. Each task will be discussed in more detail below with examples.
For the primary tasks (the CCT and the RCT), age was added to the statistical analysis as a control variable. This was done because age is controlled for in the experiment: only adults speakers are of interest to the current study.
CCT – Cleft Construction Task
The items I used for this task have been adapted from Bisnath, Wilke and Wolterbeek (2017). The task consists of seven items. Callies (2009: 123) notes that cleft constructions are likely to be used in cases of clarification, correction or misunderstandings. A sentence completion task was constructed to elicit cleft constructions. The context given with this task describes a mistake that was made, and the participant was asked to correct this mistake. The sentence completion element of the task was structured in such a way that the beginning of the
sentence – the corrective element – was left out, and the part to be corrected was given as the ending of the sentence, making it possible to answer with a cleft construction. See example
(5):
(5) You are the editor of a newspaper. One of your journalists made a mistake that must
be corrected publicly. He wrote that George R. Martin stole the book, but Salman Rushdie was the actual thief. What does the correction say?
(...), not George R. Martin.
The most straightforward (least stylized) answer to this item would be an answer like: ‘Salman Rushdie stole the book, not George R. Martin’. Because the participant is asked to correct a mistake, it was expected that answers with a cleft construction, like ‘It was Salman Rushdie, not George R. Martin’ would be elicited as well to emphasize the corrective nature of the sentence. A complete list of the items is provided in Appendix 1.
RCT – Relative Clause Task
The items I used for this task have also been adapted from Bisnath et al. (2017). The task consists of seven items. In order to elicit a relative clause with an infinitive and a passive, it
was necessary to give the beginning of the relative clause containing a superlative or an ordinal, and have the participant complete the ending of the relative clause. The items all contain a paragraph with a context to be able to complete the sentence following that. Unlike the CCT, which was built around ‘correcting mistakes’, there was no such theme for this task. See example (6):
(6) You are writing a travel guide. In this guide you describe the beautiful houses in
Amsterdam that were built alongside the canal. You add a picture of the oldest house. Under the picture you write:
This was the first house (...).
The two expected possible answers for this task are ‘This was the first house built / that was built alongside the canal’ and ‘This was the first house to be built alongside the canal’ (Simon, 2013). Of course there are variations of these answers, but syntactically there are no other options that are different in construction than the two examples given above. The paragraph providing background information cannot be interpreted to allude to either of the possible constructions. A complete list of the items can be found in Appendix 2.
CCT-NL – Dutch Cleft Construction Task
The items in this task were all translations from the CCT and also consists of seven items. The translations were kept as close to the original items as possible. Example (3) is the Dutch version of example (7):
(7) Je werkt als redacteur van een krant. Een van je journalisten heeft een fout gemaakt
die publiekelijk gecorrigeerd moet worden. De journalist in kwestie schreef dat een boek door George R. Martin was gestolen, maar de daadwerkelijke dader was Salman Rushdie. Wat staat in de correctie?
(…), niet George R. Martin.
The possible answers to this task are similar to the ones given for the CCT: Salman Rushdie
heeft het boek gestolen, niet George R. Martin ‘Salman Rushdie stole the book, not George R.
Martin, and Het was Salman Rushdie, niet George R. Martin ‘It was Salman Rushdie, not George R. Martin’. However, in Dutch, cleft constructions are far more marked than they are in English, as well as far less frequent (van der Beek, 2003). Because of this, it was expected that answers with a cleft construction would occur seldomly. All items from this task can be found in Appendix 3.
Fillers
Although technically not a measurement instrument, the fillers play a crucial part in making the aim of the questionnaire less transparent to the participants. The CCT and the RCT are both clearly distinguishable by the given part of the sentence completion element, and the CCT also by the corrective theme. Because of this, fillers were added to make the pattern more opaque. The fillers consisted of a context paragraph and a sentence completion element with the middle part of the sentence left out, as is shown in example (8):
(8) Because of a production error, a certain rice cooker model S808-A is being recalled. It
might catch fire while being used. What is written in the recall notice?
Due to a production error, the S808-A model is (...), please return it to your retailer.
The fillers were also translated to Dutch and used in the Dutch questionnaire. The English fillers can be found in Appendix 4, the Dutch fillers in Appendix 5.
Questionnaire design
Two separate questionnaires were designed; one in English and one in Dutch. The English questionnaire contained the CCT (7 items), the RCT (7 items), and fillers which were adapted from Bisnath et al. (2017) (6 items), amounting to 20 items in total. In the previous study – where the current study adapted the tasks from – the questionnaire contained 30 items (Bisnath et al., 2017). This turned out to have an impact on how many participants finished the questionnaire completely. To prevent the same thing happening, the amount of items was reduced. The questionnaire was presented to participants randomly in one of four orders, with equal distribution. The questionnaire was created in four different orders to prevent accidental priming, and to make sure that all items received equal attention overall in the case that participants might become less committed as they reached the end of the questionnaire. There was no way to test if this was actually happening as the participants filled in the questionnaire online on their own device without assistance. Before the questionnaire started, the
participants were asked to read the ethics statement, and to consent to the terms. Afterwards, the participants were asked to answer the following questions:
(i) What is your native language?
(ii) Which language(s) did you speak at home growing up? (iii) I consider my English to be…
- American - British
- Other (please specify) (iv) What is your age?
(v) Are you a student?
if they answered ‘yes’ on question (v), they would also have to answer question (vi) (vi) What do you study?
Following these questions was the main body of the questionnaire: the CCT, RCT, and fillers in one of four orders. After finishing these, the participants were asked to guess the topic of the questionnaire. If participants figured out what was expected of them they would be excluded. In Bisnath et al. (2018) – where the methodology was similar – some participants had to be disqualified because of this, in the current study this was not the case.
The Dutch questionnaire consisted of the CCT-NL and fillers, which were translations of the fillers used in the English questionnaire. The questionnaire was also preceded by the ethics statement, which was followed by three questions.
(i) Wat is je moedertaal?
(ii) Welke talen werden er thuis gesproken toen je opgroeide?
(iii) Wat is je leeftijd?
These questions are literal translations of questions (i), (ii) and (iv) respectively from the English questionnaire.
RESULTS
Descriptive Statistics
Means and standard deviations for how many times the intended construction was elicited in the CCT and the RCT are given in Table 4. Overall means and standard deviations are given
as well as per group of participants (English natives and Dutch students of English). These means are shown as percentages for the natives and the Dutch students in Table 5 to illustrate the difference, or lack thereof, in performance between the different groups.
TABLE4 | Means and standard deviations for CCT and RCT
Mean Standard Deviation
CCT overall 1.18 2.63
CCT English natives 1.21 2.65
CCT Dutch students of English 1.16 2.61
RCT overall 2.53 3.37
RCT English natives 2.04 3.19
RCT Dutch students of English 2.90 3.46
CCT = Cleft Construction Task, RCT = Relative Clause (with an infinitive and a passive) Task
TABLE5 | Mean percentage of CC in the CCT and RC in the RCT for the native speakers and the Dutch students
Natives Dutch students
CC 17.3% 16.6%
RC 29.2% 41.5%
CC = Cleft Constructions, RC = Relative Clauses (with an infinitive and a passive)
Mean and standard deviation for how many times a cleft construction was elicited for the CCT-NL task are given in Table 6. How this translates to percentages is shown in Table 7.
TABLE6 | Mean and standard deviation for CCT-NL
Mean Standard Deviation
CCT-NL 0.17 0.71
CCT-NL = Dutch Cleft Construction Task
TABLE7 | Mean percentage of CC in the CCT-NL
CC 2.4%
CC = Cleft Constructions
Mixed Models Analyses
For the statistical analyses Native Language (Dutch or English, representing the Dutch students of English and the native speakers of English) and Age were included as fixed effects. The outcome of the CCT was used as a reference level for the first analysis (Table 8). The outcome of the RCT was used as a reference level for the second analysis (Table 9). The analyses were conducted using the lme4 package (Bates, Maechler, Bolker & Walker, 2015) in R (R Development Core Team, 2016). In both models, Age was centred to an average of zero, and participant and task were added as random effects.
The first model (Table 8) shows no significant effects of any of the predictors. The second model (Table 9) shows a significant effect of Native Language on the outcome of the RCT, with a confidence interval running from –6.74…–1.10 and a p-value of 0.008. Age also has a significant effect on the outcome of the RCT with a confidence interval running from 0.03…0.40 and a p-value of 0.030.
TABLE8 | Effects – CCT as reference
Effect Estimate Std. Error Z-value P-value
Native Language 0.952 1.381 0.690 0.490
Age ……-0.064 0.093 ……-0.689 0.491
CCT = Cleft Construction Task
TABLE9 | Effects – RCT as reference
Effect Estimate Std. Error Z-value P-value
Native Language ……-3.654 1.381 ……-2.646 …0.008 **
Age 0.197 0.091 2.168 … 0.030 *
Native Language x Age ……-0.305 0.181 ……-1.682 0.093
RCT= Relative Clause (with an infinitive and a passive) Task DISCUSSION
The main goal of this study was to test the predictions made by the IH: syntactic constructions interfacing with IS will be more difficult for learners than purely syntactic constructions, and the VH: variable distributions will be more problematic than categorical distributions. These predictions were tested while also taking into consideration frequency. In the case of the current study, strictly categorical distributions were not considered. Instead, two situations of variable distributions with different degrees of variability were compared. The two different tests performed were not compared directly, the observed differences between natives and near-natives were considered separately. To make sure transfer did not play a part in the production of the CC’s by the Dutch students of English, an additional test was carried out (CCT-NL). The results of this test showed that Dutch natives are highly unlikely to produce cleft constructions in Dutch, in situations similar to the ones the Dutch students of English were presented with in English. Cleft constructions were produced only 2,4% of the time, and this was done by only one participant. Statistical analyses show that Dutch learners only perform significantly differently from the native speakers on the test eliciting the purely syntactical construction (RCT, P=.008) that was chosen for this study. Age also had a significant effect in the RCT (P=.03); which shows that the older participants were more likely to produce the target construction than the younger participants. There is no significant difference between the Dutch learners and the native speakers on the test designed to elicit a syntactical construction interfacing with IS (CCT). In fact, on this test there is only a 0,7% difference between how many times the intended construction was produced by the native speakers (17.3%) and by the Dutch learners (16.6%).
As was predicted these results are not in line with the IH’s predictions, for which the opposite would be expected. This outcome suggests that frequency plays a role in determining which constructions will be learned better by speakers at, or reaching the near-native level. When the IH was previously tested by comparing native and near-native speakers’ on their interpretation and production of overt personal pronouns and null subjects in Italian, it
showed a significant difference in performance between these two groups (Belleti et al, 2007). This is accredited to the nature of these elements: they involve an interface between syntax and IS. However, is it not the case that – perhaps incidentally – the situations in which the near-natives differ from the natives, are those which occur infrequently? If that is the case – which it seems to be according to Belleti et al. (2007) – would it not be premature to solely assign this result to the predictions made by the IH? Especially if infrequent constructions tend to be more complex (Haspelmath, 2008). Similarly, it would certainly be premature to take the results from this study and run with them without verifying them in different lingual contexts, e.g., different source and target language and different grammatical constructions. They do suggest that there is room for sophistication in the study of the IH when it comes to defining complexity, and determining the culprit that causes this complexity (interface or frequency).
It was predicted as well that frequency would have an impact on the reliability of the predictions made by the VH. This prediction is not borne out in the study. The more variably distributed construction (RC, 29.2%) is shown to be more problematic for the Dutch students
of English than the CC, which is a less variably distributed construction. The CC might even be considered more categorically distributed than variably with an observed production of 17.3% by the native speakers. This solidifies the validity of the predictions made by the VH being in line with the results. Frequency has not impacted these results and there might be a simple explanation for this. In contrast to the IH, the VH has internalised frequency in a way in the variability continuum. The variability of distributions of constructions is measured by the frequency of their relative occurrence. This relative frequency does not necessarily have to be similar to overall frequency, but if this is the case in the current study, it would explain why overall frequency did not have any impact.
The results show that the IH could benefit from including frequency in its claims somehow. If relative frequency as is considered by the VH is a reliable enough predictor, it could be added as a continuous variable to the categorical division as claimed by the IH (Graph 1):
GRAPH 1 | The Interface Hypothesis and the Vulnerability Hypothesis combined
The white area represents syntactic constructions that are not complex according to the VH (close to 0% or 100% on the variability continuum: categorical distributions). The light grey area represents syntactic constructions that are somewhat complex according to the VH (between categorically and variably distributed). The dark grey area represents syntactic constructions that are complex according to the VH (close to 50%, variably distributed). When this is combined with the IH, the numbers (1) represent where constructions would be located that are not complex: categorically distributed constructions that do not interface externally. The number (3) represents the most complex constructions: variably distributed with an external interface. The “in between” constructions are more difficult to assign a degree of complexity. According to the VH, the number (2A) in the dark grey area would represent more complex constructions, according to the IH the number (2B) in the light grey area would represent more complex constructions. When overall frequency is considered, the results found in the current study indicate that the VH makes a more reliable claim and constructions in the (2A) area (RC’s) would be more complex than those in the (2B) area
B
(CC’s). However, these same results also indicate that overall frequency is important to consider, especially when the IH is considered. Another problem with the above Graph (1) is that even though the IH currently defines complexity by interface, Sorace (2012) has already mentioned that she believes that interface conditions could be “graded according to their computational complexity and their dependence on extra-linguistic factors.” This would result in a far less rigid distinction between interface conditions. How this would be represented is not clarified, but it does indicate that other factors outside of interface should be considered when determining complexity. The current study alludes to the fact that frequency is one of those factors, more research is needed to confirm this to be true.
The results of the current study suggest that there is more to be learned by researching the effect of frequency when it concerns the L2 capability near-natives. However, there are several elements in the methodology of the current study that should be improved upon. Firstly, the sentence completion task when it concerns the CC’s showed high variability between items. Some items elicited far more CC’s than others. The items that were most productive in eliciting CC’s were those in which the clefted constituent would be a subject or an object. In the case of adjuncts, CC’s were rarely produced. Because this was true for both the English natives and the Dutch students of English it did not skew the results found by the statistical analysis, but it could have had an impact on the descriptive results. If CC’s are less or more variably distributed based on the grammatical role of the clefted constituent, it means that they cannot be directly compared to the RC’s when making generalizations about the variability of distribution. The RC’s were produced in response to all items, and there is no variation as to which grammatical roles occur in the sentences containing an RC. For future research, it would be important that the sentence completion tasks consist of items that are all productive in eliciting the targeted constructions. CC’s might not be the best to look at as there is not only variability in which grammatical constituents can be clefted, but also in which cleft construction can/will be used. In order to find better constructions to use, it is imperative that corpus research be done to account for frequency. There is currently very little research available that provides us with frequency numbers. If there is more data available, there will be more grammatical constructions of comparable frequency available to choose from.
A potential issue with the statistical analysis is that only the targeted constructions were considered: CC’s and RC’s. With variability of distribution being a factor of importance, it would have been interesting to see how the alternatives to the targeted constructions were distributed, as well as how the English natives differed from the Dutch students of English on these. The reason why this was not for the current study is because there was a focus on frequency. No data regarding frequency of the alternative constructions were available at the time of the current study, but the assumption is that they are very frequent. The most used alternative to CC’s was a non-clefted SVO sentence, for the RC’s this was a construction where the word ‘that’ was used instead of the ‘to be’ construction (‘The last tiger that was
seen / to be seen.’). There is also no reason to believe that these constructions are difficult for
near-natives, which is why they were difficult to include in the narrative of the current study. Lastly, there is no evidence that the Dutch students of English should be considered near-native speakers of English. While these students are considered (very) advanced in Verheijen et al. (2013) they are never referred to as near-native. Additionally, in this same study they found there to be differences between students of different years. Students in the final (third) year of their BA produced written language closer to that produced by native speakers of English than first or second year students. This implies that the Dutch students of English in the current study from various years of BA English programmes should not be considered near-native but (very) advanced. Whether this difference has an impact on the results is not clear and it should be accounted for in future research.
A re-evaluation of the work that has been done in the past years on the IH from the perspective of the VH and/or by considering frequency as well might help us understand better what actually is complex to the near-native speaker. This knowledge in turn could bring us closer to helping near-natives cross the bridge between near-native level and native level.
REFERENCES
Baguley, T. (2009). Standardized or simple effect size: what should be reported? British
Journal of Psychology 100, 603–617.
Bates, D., Mächler, M., Bolker, B., & Walker, S. (2014). Fitting linear mixed–effects models using lme4. Journal of Statistical Software 67(1), 1–48.
Beek, L. van der (2003). The Dutch it–cleft constructions. Butt, M., & Holloway King, T. (Eds.), The Proceedings of the LFG'03 Conference: CSLI Publications: 22–42.
Belletti, A., Bennati, E. & Sorace, A. (2007). Theoretical and developmental issues in the syntax of subjects: Evidence from near–native Italian. Natural Language and Linguistic
Theory 25, 657–689.
Biber, D., Johannsson, S., Leech, G., Conrad, S., & Finnegan, E. (1999). The Longman
grammar of spoken and written English. Harlow: Longman.
Bisnath, F., Wilke, H. & Wolterbeek, J. (2017). Are syntactic constructions harder to learn if
they require an interface? Unpublished manuscript, University of Amsterdam,
Amsterdam.
Bybee, J. & Hopper, P. (Eds.). (2001). Frequency and the emergence of linguistics structure. Amsterdam: Benjamins.
Callies, M. (2009). Information highlighting in advanced learner English: The syntax
pragmatics interface in second language acquisition. Amsterdam: John Benjamins.
Chafe, W. L. (1976). Givenness, contrastiveness, definiteness, subjects, topics and point of view. In C. N. Li (Ed.), Subject and topic (pp. 27–55). New York: Academic Press. Denhovska, N., Serratrice, L. & Payne, J. (2016). Acquisition of second language grammar
under incidental learning conditions: The role of frequency and working memory.
Language Learning 66(1), 159–190.
Ellis, N. (2002). Frequency effects in Language Processing: A review with implications for theories of implicit and explicit language acquisition. SSLA 23, 143–188.
Ellis, N. (2006). Cognitive perspectives on SLA: The associative-cognitive CREED. AILA
Review 19. 100–121.
Ellis, N. & Collins, L. (2009). Input and second language acquisition: the roles of
frequency, form, and function – introduction to the special issue. The Modern Language
Journal 93(3), 329–335.
Ellis, R. (1994). The study of second language acquisition. Oxford: Oxford University Press. Gass, S., & Mackey, A. (2002). Frequency effects and second language acquisition:
A complex picture? Studies in Second Language Acquisition 24, 249–260.
Gass, S., & Selinker, L. (2001). Second language acquisition: An introductory course (2nd ed.). Mahwas, NJ: Erlbaum.
Haspelmath, M. (2008). Frequency vs. iconicity in explaining grammatical asymmetries.
Cognitive linguistics 19(1), 1–33.
Hopp, H. (2011). Extended patterns and computational complexity. Linguistic Approaches
to Bilingualism 1(1), 43–47.
Hulk, A. & Müller, N. (2000). Crosslinguistic influence at the interface between syntax and pragmatics. Bilingualism: Language and Cognition 3(3), 227–244.
Hulstijn, J. (2005). Theoretical and empirical issues in the study of implicit and explicit second-language learning. Studies in Second Language Acquisition 27, 129–140.
Johnson, J. S., Newport, E. L. (1989). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language.
Cognitive Psychology 21, 60–99.
Krashen, D., Long, M. A. & Scarcella, R. C. (1979). Age, rate and eventual attainment in second language acquisition. TESOL Quarterly 13(4), 573–582.
Krifka, M. (2008). Basic notions of information structure. Acta Linguistica Hungarica 55 (3–4), 243–276.
Lambrecht, K. (1994). Information structure and sentence form: A theory of topic, focus, and
the mental representations of discourse referents. Cambridge: Cambridge University
Press.
Lambrecht, K. (2001) A framework for the analysis of cleft constructions. Linguistics 39(3), 463–516.
Lieven, E. (2010). Input and first language acquisition: Evaluating the role of frequency.
Lingua 120. 2546–2556.
Loewen, S. (2015). Introduction to Instructed Second Language Acquisition. New York: NY: Routledge.
McMahon, M. S. (1994). Understanding language change. Cambridge: Cambridge University Press.
Montrul, S. (2014). Interlanguage, transfer and fossilization: Beyond second language
acquisition. In: Z. Han & E. Tarone (Eds.), Interlanguage: Forty years later (pp. 75–104).
Amsterdam: John Benjamins.
Müller, N. (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. EURSOLA Yearbook 8, 52–78.
Müller, N., & Hulk, A. (2000). Crosslinguistic influence in bilingual children: Object
omissions and root infinitives. In C. Howell, S. A. Fish & T. Keith-Lucas, Proceedings of
the 24th Annual Boston University Conference on Language Development, 546–557.
Somerville MA: Cascadilla Press.
Müller, N., & Hulk, A. (2001). Corsslinguistic influence in bilingual language acquisition: Italian and French as recipient languages. Bilingualism: Language and Cognition 4(1), 1–21.
Odlin, T. (1989). Language Transfer: Cross-linguistic influence in language learning. New York: Cambridge University Press.
Patkowski, M. S. (1980). The sensitive period for the acquisition of syntax in a second language. Language Learning 30(2), 449–472 .
Pires, A., & Rothman, J. (2011). An integrated perspective on comparative bilingual
differences: Beyond the interface problem? Linguistic Approaches to Bilingualism 1(1), 74–78.
Prada Perez, A. de (2018). Theoretical implications of research on bilingual subject production: The Vulnerability Hypothesis. International Journal of Bilingualism.doi: 10.1177/1367009618763141
Qualtrics. (2017). http://www.qualtrics.com
R Core team. (2016). R: A language and environment for statistical computing. Vienna: R foundation for statistical computing.
Roland, D., Dick, F. & Elman, J.L. (2007). Frequency of basic English grammatical structures: A corpus analysis. Journal of Memory and Language 57, 348–379. Saffran, J. (2003) Statistical language learning: Mechanisms and constraints. Current
Directions in Psychological Science 12. 110–114.
Saffran, J., Newport, E., Aslin, R., Tunick, R., & Barrueco, S. (1997). Incidental language learning: Listening (and learning) out of the corner of your ear. Psychological Science 8. 101–105
Simon, P. (2013). The grammaring guide to English grammar. Retrieved from http://www.amazon.com
Sorace, A. (2000). Syntactic optionality in non-native grammars. Second Language Research
16, 93–102.
Sorace, A. (2003). Near-nativeness. In C. Doughty & M. Long (Eds.), Handbook of second
language acquisition (pp. 130–151). Oxford, UK: Blackwell Publishers.
Sorace, A. (2005). Selective optionality in language development. In L. Cornips & K. P. Corrigan (Eds.), Syntax and variation. Reconciling the biological and the social (pp. 55– 80). Amsterdam: John Benjamins.
Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic
Approaches to Bilingualism 1(1), 1–33.
Sorace, A. & Filiaci, F. (2006). Anaphora resolution in near-native speakers of Italian. Second
Language Research 22(3), 339–368.
Sorace, A. & Serratrice, L. (2009). Internal and external interfaces in bilingual language development: Beyond structural overlap. International Journal of Bilingualism 13(2), 195–210.
Szymanik, J. (2016). Quantifiers and cognition: Logical and computational perspectives. Berlin: Springer
Tagliamonte, S. (2002). Comparative Sociolinguistics. In J. K. Chambers, P. Trudgill & N. Schilling-Estes (Eds.), The handbook of language variation and change (pp. 729–763). Oxford, UK: Blackwell Publishers.
Tsimpli, I. & Sorace, A. (2006). Differentiating interfaces: L2 performance in syntax-semantics and syntax–discourse phenomena. Proceedings of the 30th Annual Boston
University Conference on Language Development (pp. 653–664). Somerville, MA:
Cascadilla Press.
Velleman, D., Beaver, D., Destruel, E., Bumford, D., Onea, E., & Coppock, L. (2012). It–clefts are IT (inquiry terminating) constructions. Proceedings of SALT 22: 441–460. Verheijen, L., Los, B. & Haan, de P. (2013) Information structure: The final hurdle? The
development of syntactic structures in (very) advanced Dutch EFL writing. Dutch Journal
of Applied Linguistics 2(1), 92–107.
Weinreich, U. (1953). Languages in contact: Findings and problems. New York: Linguistic Circle of New York.
Appendix 1 – CCT
CC1
There is an accident between a car and a bike at a traffic light on your street in your neighbourhood. The driver was responsible; however, the local newspaper reports
that it was the cyclist. You write a letter to the newspaper correcting them. Because of space constraints you must use one sentence. What do you write?
(...), not the cyclist.
CC2
The official graduation party of the class of 2018 is to be held at Cafe de Gaeper; however, on the advertisement it is printed Cafe de Slaper. A one-sentence notice is sent by the university to staff and students making the correction. What is written?
(...), not Cafe de Slaper.
CC3
You are writing an email to a respected professor from a university you are hoping to attend. He runs two projects: one in Ghana and one in Guyana. You made a mistake and said you wanted to go to Ghana. You now want to specify that you meant
Guyana instead. What do you write, using one sentence?
(...), not Ghana.
CC4
You work at a bank. One day €50,000.00 goes missing from an ATM. Your boss thinks that your department made a mistake, but really it was the security company. You must write a formal report explaining this. How do you summarise what
happened in one sentence?
(...), not our department.
CC5
A group of exchange students from Portugal is arriving to the university this
afternoon. On the TV screen in the entry hall a message reads: ‘Welcome, students
from Spain!’. You write a short message to the IT-department explaining they made a
mistake. How do you conclude this message?
(…), not Spain.
CC6
You are the editor of a newspaper. One of your journalists made a mistake that must be corrected publicly. He wrote that George R. Martin stole the book, but Salman Rushdie was the actual thief. What does the correction say?
(...), not George R. Martin.
CC7
You are writing a letter to the head of your university. You were recently accused of stealing a wallet. You did not do it but you know that Hans Steiner did. What single sentence do you write to make it clear who stole the wallet?
(...), not me.
Appendix 2 – RCT
RC1
Although computers existed for quite some time in the U.S. already, they were
sold in the Netherlands before any other kinds. It was sold for home and office use. This made it special, because earlier computers were only used for research. The IBM website writes:
The IBM personal computer was the first computer (...).
RC2
A very rare kind of tiger lives in the forests of the Amazon. Thirty years ago, the last one was seen in the north part of the Amazon. You read about this tiger in a book:
This tiger was the last one (...).
RC3
Before 1991, it was not forbidden for a man to rape a woman he was married to. Mike Smith was convicted for raping his wife in 1992. This was the first time a sentence like this was carried out. The newspaper reports:
Mike Smith was the first man (...)
RC4
Famous artworks are sold for high prices nowadays. The painting Salvator Mundi by Leonardo da Vinci was the most expensive one, and was sold for 400 million dollars. You read an article about the sale. What does it say?
Salvator Mundi was the most expensive painting (...).
RC5
A huge earthquake has destroyed a Mexican village. After days of searching, no survivors are found anymore. However, after a week, a small 6 year old boy, Frankie Muñez, is found alive in one of the collapsed houses. What does the police report say?
Frankie Muñez (6) was the last survivor (...).
RC6
Harvey Milk was elected in 1977 as the first homosexual politician. He was elected as city councillor of San Francisco, where his statue also is. What does it say on the plaque?
Harvey Milk, the first gay politician (...).
RC7
You are writing a travel guide. In this you describe the beautiful houses in Amsterdam that were built alongside the canal. You add a picture of the oldest house. Under the picture you write:
This was the first house (...).
Appendix 3 – CCT-NL
CC-NL1
Er is een ongeluk tussen een auto en een fiets bij een verkeerslicht in jouw buurt. De bestuurder van de auto was verantwoordelijk, maar de lokale krant schrijft dat de
fietser fout zat. Je schrijft een brief aan de krant om hen te corrigeren. Wegens ruimtegebrek kan je dit in maar één zin doen, wat schrijf je?
(…), niet de fietser.
CC-NL2
Het afstudeerfeest van de klas van 2018 wordt gehouden in het café ‘De Gaeper’; echter, op een flyer staat gedrukt dat het gehouden wordt in het café ‘De Slaper’. Een correctie bestaande uit één zin wordt geschreven door de universiteit en verspreid onder de studenten. Wat staat hier op geschreven?
(…), niet in café De Slaper.
CC-NL3
Je schrijft een e-mail aan een professor die werkzaam is aan een universiteit waar jij graag zou studeren. Hij is met twee projecten bezig: een in Ghana, en een in
Guyana. Je hebt per ongeluk in jouw e-mail geschreven dat je mee wilt werken aan het project in Ghana. Nu wil je dit rechtzetten. Wat schrijf je, in één zin?
(…), niet Ghana.
CC-NL4
Je werkt voor een bank. Op een dag blijkt 50.000 euro vermist te zijn uit de voorraad. De baas denkt dat jouw afdeling verantwoordelijk is voor de fout, maar de fout is gemaakt door de waardetransport. Je moet hier formeel melding van maken. Hoe vat je in één zin samen wat er gebeurd is?
(…), niet onze afdeling.
CC-NL5
Een groep studenten uit Portugal zal deze middag op de universiteit aankomen. OP het TV scherm in de hal is een bericht te zien: ‘Welkom, studenten uit Spanje!’. Je schrijft een kort bericht aan de IT-afdeling om uit te leggen dat ze een fout hebben gemaakt. Hoe sluit je dit bericht af?
(…), niet Spanje.
CC-NL6
Je werkt als redacteur van een krant. Een van je journalisten heeft een fout gemaakt die publiekelijk gecorrigeerd moet worden. De journalist in kwestie schreef dat een boek was gestolen door George R. Martin, maar de daadwerkelijke dader was Salman Rushdie. Wat staat in de correctie?
(…), niet George R. Martin.
CC-NL7
Je schrijft een brief aan de directie van je universiteit. Je bent onlangs beschuldigd van het stelen van een portemonnee. Jij hebt dit niet gedaan, je weet dat Hans Steiner degene is die de portemonnee gestolen heeft. Wat schrijf je, in één zin, om duidelijk te maken wie de dader is?
(…), ik niet.
Appendix 4 – English fillers
EF1
You are writing an essay about brown-headed spider monkeys and black-headed spider monkeys that you observed at a conservation site. The brown-headed
monkeys were aggressive, while black-headed ones were docile; however, the literature reports the opposite. What is the first sentence you write in a paragraph contrasting your observations with the literature?
I noted that, in contrast to the literature, (...) were aggressive, not the black-headed ones.
EF2
You are reading an online newspaper article about a masked gunman who shot twenty people in New York City today. What does the one-sentence summary of the article say?
Today, twenty people were killed (…) in New York City.
EF3
You are asked to create flyers for a live music event in the Vondelpark in Amsterdam. A maximum of 500 people are allowed to attend the event at the same time. The organisation wants you to emphasize this. What do you write on the flyer?
No more than 500 people (...) at the event.
EF4
You receive an official notice from your university that the location of the promotion ceremony for Max van der Waal, which was going to be held at the university library, will now be held at the university auditorium. What does it say in one sentence?
Max van der Waal’s promotion ceremony (...) at the university auditorium, not the university library.
EF5
You are writing an essay for a science class. You read an article published by the Centers for Disease Control and Prevention saying that pneumonia, ear infections, sinus infections, meningitis and bacteremia are caused by Streptococcus
pneumoniae bacteria. How do you summarise this information in one sentence?
These bacteria (...) different types of infections, among them pneumonia.
EF6
Because of a production error, a certain rice cooker model S808-A is being recalled. It might catch fire while being used. What is written in the recall notice?
Due to a production error, the S808-A model is (...), please return it to your retailer.
Appendix 5 – Dutch fillers
DF1
Je schrijft een essay over bruinkop slingeraapjes en zwartkop slingeraapjes. Je hebt deze geobserveerd in een beschermd gebied. De bruinkop slingeraapjes waren
agressief, terwijl de zwartkop slingeraapjes juist erg rustig waren. In de literatuur die je gelezen hebt staat echter precies het tegenovergestelde. Wat is de eerste zin van je paragraaf waarin je dit contrast met de literatuur benoemt?
Ik heb geobserveerd dat, in tegenstelling tot wat in de huidige literatuur beschreven staat, (…), niet de zwartkop slingeraapjes.
DF2
Je schrijft een artikel voor een online krant over een gemaskerde gewapende man die vandaag twintig mensen in New York heeft neergeschoten. Wat is de
samenvatting van dit artikel in één zin?
Vandaag zijn twintig mensen neergeschoten (…) in New York.
DF3
Er is je gevraagd om flyers te ontwerpen voor een muzikaal evenement in het
Vondelpark in Amsterdam. Er mogen niet meer dan 500 gasten aanwezig zijn bij het evenement. De organisatie wil dat dit duidelijk gecommuniceerd wordt. Wat zet je in de flyer?
Maximaal 500 mensen (…) tot het evenement.
DF4
Je ontvangt een officieel bericht van je universiteit over de locatie van de promotieceremonie van Max van der Waal. Deze zou plaatsvinden in de universiteitsbibliotheek, maar er is besloten om het te verplaatsen naar het auditorium. Hoe staat dit beschreven in het bericht?
Max van der Waals promotieceremonie (…) in het auditorium, niet in de universiteitsbibliotheek.
DF5
Je schrijft een opstel voor een college bacteriologie. Je leest een artikel waarin gesteld wordt dat longontsteking, oorontsteking, en hersenvliesontsteking allen veroorzaakt worden door dezelfde soort bacteriën. Hoe vat je deze informatie samen in één zin?
Deze bacteriën (…) verschillende soorten infecties, waaronder longontsteking.
DF6
Door een productiefout wordt een zeker model rijstkoker, model S808-A, herroepen. De rijstkoker zou vlam kunnen vatten tijdens gebruik. Wat staat er in het bericht van de producent over dit probleem?
Vanwege een productiefout, is het model S808-A (…), breng het alstublieft terug naar de winkel waar u deze gekocht heeft.
