Sage Journals: Discover world-class research

Abstract

Aims and objectives:

The aim of this study is to investigate language contact from a usage-based perspective. In particular, we test the explanatory power of a schematicity continuum, one of the central assumptions within usage-based approaches, in regard to these contact effects. In doing so, we aim to better understand how schematicity influences language transfer as well as learn more about the cognitive mechanism of schematicity itself.

Design:

The empirical focus of this study is on native German speakers living in the Netherlands who tend to experience a lot of language transfer from their second language Dutch to their native language German. The experiment consists of a production task containing constructions that differ in regard to their potential level of schematicity (i.e., lexically specific, partially schematic, fully schematic) and in the type of transfer that they might trigger (i.e., covert and overt transfer).

Data and analysis:

We compare production data of native German speakers living in the Netherlands (N = 60) with those of a control group of speakers not in contact with Dutch (N = 60), to establish whether the language use of these two groups differs. In addition, we analyze both speaker and item variation for the different constructions.

Findings/conclusions:

The speakers with Dutch contact show a number of contact effects, such as covert and overt transfer as well as hypercorrection regarding the use of certain constructions.

Originality:

The variation analysis reveals individual differences in how speakers use the different constructions. Some of these differences can be attributed to their mental levels of schematicity.

Significance/implications:

The study shows that transfer depends on schematicity. In doing so, it provides direct evidence for different levels of schematicity in speakers’ mental representations, demonstrating the usefulness of studying cognitive mechanisms within a language contact setting.

Keywords

Language transfer usage-based approach schematicity continuum covert and overt transfer speaker and item variation

Introduction

Research in the field of language contact has documented a vast variety of language contact settings and phenomena. In regard to the settings, researchers have, for instance, looked at languages that have been in contact for hundreds of years (e.g., Heine & Kuteva, 2003, 2005) as well as languages that have just come into contact (e.g., M. S. Schmid & Köpke, 2017), contact languages with varying degrees of typological closeness, different reasons for the emergence of contact such as migration or colonization (see Appel & Muysken, 2006, pp. 5–6 for an overview), language contact among first-generation immigrants (e.g., Ribbert & Kuiken, 2010), later generation immigrants (e.g., Backus et al., 2011), and even returnees (Flores, 2010), and many, many, more factors that can all shape their respective contact settings. In regard to the contact phenomena, they have investigated changes at a lexical level such as words or fixed collocations (e.g., Treffers-Daller, 2012) and structural changes such as the emergence of new word order patterns that are shared across the contact languages (e.g., Clyne, 2003). They have looked at differences between overt and covert change (i.e., changes that result in a frequency increase of an existing pattern versus changes that result in a new pattern; Mougeon et al., 2005) and changes that result in increased convergence like the examples so far as well as changes that instead result in more divergence across the contact languages as speakers actively try to differentiate between them (see Kühl & Braunmüller, 2014 for an overview).

This variety of contact settings and contact phenomena is both one of the field’s greatest opportunities and one of its greatest challenges. It is an opportunity because it makes it possible to tease apart which settings result in which contact phenomena, and thereby helps us understand why and how languages change in situations of language contact; it is a challenge because there are so many factors to account for when trying to integrate the insights from the different contact settings and phenomena. Clearly, what is needed is an account that allows us to combine research findings and to identify general patterns, while also allowing for enough room to account for the idiosyncrasies of each individual contact setting. In their book Cognitive Contact Linguistics, Zenner et al. (2019) argue that Cognitive Linguistics, in particular the usage-based approach, might be able to provide such a unifying account.

In short, the usage-based approach assumes that different linguistic phenomena — for example, the way people acquire language, the way they process and store language, the way language changes and evolves — are all based on actual language usage. For instance, Bybee (2006, p. 711) defines grammar as the “cognitive organization of one’s experiences with language” and Tomasello (2000, p. 62) argues that our linguistic skills are “the result of accumulated experiences with language across the totality of usage events.” These usage events are assumed to be processed and stored using general cognitive principles, that is, principles that do not only apply to language but are used in other cognitive domains as well, such as entrenchment, categorization, and generalization (e.g., Bybee, 2011; Ibbotson, 2013; Tomasello, 2000; see section “Schematicity continuum” for a more detailed description of these principles). When looking at contact phenomena from such a usage-based perspective, many phenomena that have previously been investigated in separate strands within the research on language contact can now be integrated (Zenner et al., 2019), demonstrating the unifying potential of this approach.

For example, from a usage-based perspective, it becomes possible to directly relate lexical and structural language change. This is because instead of treating the lexicon and syntax as two separate modules, the usage-based approach assumes them to form a schematicity continuum (see section “Schematicity continuum”) ranging from lexically specific items to partially schematic constructions to fully schematic patterns. In this study, we aim to test the explanatory power of this conceptualization in regard to language contact phenomena. To do that, we collect and analyze data on language transfer regarding a variety of different constructions that (1) potentially fall on different stages of the usage-based schematicity continuum (i.e., lexically specific, partially schematic, fully schematic) and (2) also differ in regard to the distinction between covert and overt transfer, which is an important distinction in research on language contact (see section “Covert and overt transfer”). This combination allows us to investigate transfer patterns in a wide variety of constructions. The empirical focus of this study is on native German speakers living in the Netherlands, who have been shown to experience extensive transfer from their second language Dutch to their native language German (Brons-Albert, 1992, 1994; Ribbert & Kuiken, 2010), making this group well-suited for the comparison of transfer patterns across different construction types that we propose here. Besides, this setting allows us to shift the focus of research on language transfer, which predominantly looks at language transfer from speakers’ first to their second language, to also include other language contact situations in which this transfer occurs from speakers’ second to their first language.

Importantly, the overall aim of this study is to investigate language transfer from a usage-based perspective and thereby, hopefully, to learn more about both language transfer and the usage-based principles. This is in line with Zenner et al. (2019, p. 4), who define the aim of Cognitive Contact Linguistics as follows:

to explore how the guiding principles of Cognitive Linguistics apply to the bi- or multilingual mind in its dynamic bi- and multilingual environment, how this feeds back to our general understanding of these guiding principles, and how we can as a result better grasp how the interaction between cognition and context results in contact-induced variation and change.

In regard to this study, this means that we do not only want to test to what extent the usage-based principles are able to account for the transfer patterns across the different construction types, but that we also want to turn this around and reflect on what these transfer patterns in turn can tell us about the usage-based principles and how that can ultimately help us better understand contact-induced language change.

Schematicity continuum

According to the usage-based approach, lexicon and syntax do not represent separate modules, but rather form a so-called schematicity continuum. This continuum is assumed to be the result of a number of cognitive principles: first, speakers entrench words and constructions that they encounter, with words and constructions that they encounter frequently becoming more and more entrenched. This means that speakers can easily activate them (Blumenthal-Dramé, 2016; Bybee, 2011; H.-J. Schmid, 2016) and therefore are also more likely to select them in subsequent language production (Bybee, 2011; De Smet, 2016; Langacker, 1987, 2016). When storing these linguistic exemplars, speakers store exemplars that they perceive as similar closely together (Bybee, 2011; Croft, 2000; Goldberg, 1995; Langacker, 1987, 2016), resulting in categories that speakers might start to generalize over (Bybee, 2011; Ibbotson, 2013; H.-J. Schmid, 2016; Tomasello, 2000b). This results in constructions with varying degrees of schematicity, ranging from lexically specific constructions (e.g., I don’t know, I don’t think, do you want, and and I said; Biber et al., 1999) to partially schematic constructions (e.g., the X is that as in the thing/fact/point is that; H.-J. Schmid, 2007) to fully schematic constructions (e.g., the SVO [subject-verb-object] word order in English main clauses; Tomasello, 2000b).

To capture these different levels of schematicity and study their effects on language transfer, this study uses a two-step approach: first, we selected constructions that we think are likely to be mentally stored as lexically specific items (i.e., light verb constructions [lvcs], section “Lexically specific lvcs”), partially schematic constructions (i.e., um-constructions, section “Partially schematic um-constructions”), and fully schematic patterns (i.e., word order patterns, section “Fully schematic word order patterns”). We want to emphasize here that this selection was nothing more than an informed guess. As Dąbrowska (2020, p. 215) argues: “the generalizations that we as linguists might make to describe the patterns found in the utterances produced by members of a particular speech community do not necessarily correspond to the generalizations that individual members of that community have acquired,” meaning that it is simply not possible to determine the mental level of schematicity just by looking at the construction alone. Nevertheless, we feel that selecting constructions in this way (i.e., selecting constructions that allow for generalizations to different degrees, see section “Stimuli”) is still our best option when attempting to create a stimuli set with constructions at varying levels of schematicity. We then analyzed the data in a way that allowed us to test whether this initial classification of schematicity is also reflected in the speakers’ transfer patterns. Specifically, we show in this paper that a detailed analysis of individual variation (i.e., speaker and item variation), which is a central concept within usage-based theories in general (Dąbrowska, 2020; Verhagen et al., 2018, 2020), can tell us a lot about whether speakers process instances of a given construction in similar or different ways, which in turn provides a first indication regarding the construction’s mental level of schematicity.¹

Covert and overt transfer

For all construction types, we further distinguished between constructions whose transfer results in covert transfer and whose transfer results in overt transfer. This is a distinction that is commonly made in research on contact-induced language change (e.g., Mougeon et al., 2005; Treffers-Daller, 2012; Treffers-Daller & Sakel, 2012; also referred to as system-preserving and system-altering transfer, Aikhenvald, 2003). In the case of covert transfer, there are two competing constructions in language A, one of which has a parallel form in the contact language B. As a result, transfer from language B might result in speakers using the construction with a parallel form rather than the other one in language A, which, if the transfer occurs frequently enough, results in speakers over-using that construction at the expense of the alternative construction (Heine & Kuteva, 2003; Mougeon et al., 2005; Treffers-Daller & Mougeon, 2005). For example, in German it is possible to use both the light verb construction Einfluss nehmen “to take influence” and Einfluss ausüben “to practice influence”. The corresponding construction in Dutch is invloed uitoefenen “to practice influence”. Transfer from Dutch to German might therefore lead native German speakers living in the Netherlands to use the combination Einfluss ausüben more frequently than Einfluss nehmen compared to speakers not in contact with Dutch. In the case of overt transfer, transfer leads to speakers using a new, previously unattested construction (Mougeon et al., 2005), for instance, using the combination *Wecker setzen “to set an alarm” instead of Wecker stellen “to put an alarm” in German due to transfer of the Dutch combination wekker zetten “to set an alarm”. This distinction between covert and overt transfer can also be applied to more schematic constructions (see sections “Partially schematic um-constructions” and “Fully schematic word order patterns”).

Method

Design

The study consists of a production task in which participants are asked to produce instances of a number of different constructions that they might transfer from their Dutch to their German. These constructions differ in their potential level of schematicity (lexically specific, partially schematic, fully schematic) and in the type of transfer that they might trigger (covert transfer, overt transfer). To analyze the results, we first compared the production patterns of native German speakers living in the Netherlands to those of a control group of native German speakers not in contact with Dutch. This comparison can show to what extent usage patterns are different between the two groups, likely due to their contact with Dutch for the speakers living in the Netherlands. In a second step, we focused on both speaker and item variation as a way to learn more about the transfer patterns for the different construction types, especially in regard to what these patterns can reveal about the constructions’ mental levels of schematicity (see section “Schematicity Continuum”). The study was approved by the Research Ethics and Data Management Committee of Tilburg School of Humanities and Digital Sciences. The study was preregistered at https://osf.io/92ngu.

Participants

In total, 120 participants (16.7% male) took part in the experiment. They were either native German speakers living in the Netherlands or native German speakers without any contact with Dutch. On average, they were 36.17 years old (SD = 14.05), ranging from 19 to 73 years. The speakers without Dutch contact (N = 60) were living in Germany at the moment of the study. They were not raised bilingually and did not speak Dutch, so any potential influence from Dutch on their German was highly unlikely. The speakers with Dutch contact (N = 60) were living in the Netherlands. They were not raised bilingually, had immigrated to the Netherlands as adults (mean age of immigration: 27.00 years, SD = 6.18, ranging from 19 to 48 years), and had been living in the Netherlands for at least 5 years (M = 18.02, SD = 10.78, ranging from 5 to 49 years).

Stimuli

Table 1 presents an overview of the selected stimuli. In total, we selected 90 stimuli (30 stimuli per construction type divided into 15 stimuli that might trigger covert transfer and 15 stimuli that might trigger overt transfer; see Appendix 1).² The stimuli were split into two lists, which participants were randomly assigned to. The construction types are explained in more detail in the following sections.

Table 1.

Stimuli categories of covert and overt transfer based on the continuum of schematicity.

	Lexically specific	Partially schematic	Fully schematic
covert	lvc pairings with similar and different light verbs	Consecutive clauses with um	Placement of prepositional phrases in the postfield position
overt	lvc pairings with different light verbs	Attributive clauses with um	Word order of verbal end groups in subordinate clauses

lvc: light verb construction.

Lexically specific lvcs

lvcs (e.g., to set an alarm, to take a walk) consist of a semantically bleached verb and a noun phrase, which carries most of the construction’s semantic meaning (e.g., in a construction like to take a walk, most of the meaning is encoded in the noun phrase a walk and not in the verb take; Butt, 2010; Wittenberg, Jackendoff, et al., 2014; Wittenberg, Paczynski, et al., 2014). Importantly here, not any light verb can be combined with any noun phrase to form an lvc (e.g., *make a walk, *do a decision), which means that speakers have to mentally store these constructions as lexically specific units to produce the conventional combination of noun phrase and light verb (Basarić & Petrič, 2015; Wittenberg, Jackendoff, et al., 2014). This makes these constructions lexically specific.

Transfer of lvcs can result in both overt and covert transfer. For the lvcs potentially triggering overt transfer, we selected German–Dutch lvc pairs that differ in their light verbs (e.g., Wecker stellen—wekker zetten). In these cases, transfer may lead to speakers using a new, previously unattested construction in German (*Wecker setzen). For the lvcs triggering covert transfer, we selected lvcs that have two synonymous constructions in German (e.g., Einfluss nehmen “to take influence”, Einfluss ausüben “to practice influence”). One of the constructions has a parallel form in Dutch (e.g., Einfluss ausüben is similar to the Dutch invloed uitoefenen “to practice influence”). Transfer of that construction from Dutch to German thus results in speakers using the construction with a parallel form rather than the other lvc (see Table 1), which, if the transfer occurs frequently enough, results in speakers using the construction (e.g., Einfluss ausüben) with a higher frequency than speakers not in contact with Dutch.

Partially schematic um-constructions

The partially schematic constructions (um) that were selected for this study all make use of the complementizer om “to” in Dutch. These sentences (i.e., attributive clauses see (1) and consecutive clauses see (2)³) are constructions that contain both lexically fixed and open slots, which makes them potentially partially schematic. To illustrate, sentences (1) and (2) show the sentences’ potential schematic representations. In these representations, square brackets [] are placed around the open slots in the construction to indicate that this position may be filled by different items.

(1) Attributive clause

A. lexical German Peter hat versucht *um Anna anzurufen Dutch Peter heeft geprobeerd om Anna te bellen Peter has tried Anna to call “Peter has tried to call Anna.” B. schematic German [Subject] [versuchen] *um [to do something] Dutch [Subject] [proberen] om [to do something]

(2) Consecutive clause

A. lexical German Anna war alt genug (um) in eine eigene Wohnung zu ziehen Dutch Anna was oud genoeg om in een eigen woning te verhuizen Anna was old enough into an own appartment to move “Anna was old enough to move into her own apartment.” B. schematic German [Subject] [genug X sein] (um) [to do something] Dutch [Subject] [genoeg X zijn] om [to do something]

In these sentences, the complementizer om is used in Dutch, but the sentences differ in regard to whether the German equivalent um is used in the German sentences. In the case of the attributive clauses as in (1), the use of um is ungrammatical in German (DudenOnline, n.d.; Ribbert & Kuiken, 2010). Transfer from Dutch, which results in speakers using um in these sentences, thus leads to a new construction that is not used in standard German. We therefore use these attributive clauses as stimuli in the overt transfer condition. In the covert transfer condition, we selected consecutive sentences as stimuli. These sentences as in (2) follow a main clause containing an expression of degree (enough X, too much X, not enough X, too little X) and express the direct consequence of this main clause (Ten Cate et al., 1998). Consecutive clauses start with the complementizer om in Dutch (e-ANS, n.d.) and it is also possible—but not obligatory—to use this complementizer in German (DudenOnline, n.d.). This means that there are two possible German constructions, one (i.e., the construction with um) shares the form with the Dutch equivalent, the other one (i.e., the construction without um) does not. Transfer from Dutch might therefore lead to speakers using the construction with um more frequently than speakers not in contact with Dutch.

Fully schematic word order patterns

The constructions that were selected as the fully schematic construction type (order) are different word order patterns. For these constructions, all slots in the constructions are open (i.e., there is no restriction on the specific subjects, verbs, prepositional phrases, etc. that can be inserted in the slots), which makes these patterns potentially fully schematic.

In the overt transfer condition, we selected the word order patterns in verbal end groups in Dutch and German subordinate clauses, because these patterns differ in Dutch and German (Coussé, 2003; Swerts & van Wijk, 2005). When there are two verbal elements in the clause (as in (3) the auxiliary have and the verb read), both the word order [verb + auxiliary] and [auxiliary + verb] are possible in Dutch, but only the first word order is possible in German. Transfer from Dutch to German therefore may lead to speakers using the order [auxiliary verb] (e.g., *hat gelesen) in German, even though that is a word order not used in standard German.

(3) Nachdem Anna den Brief gelesen hat

Nadat Anna de letter gelezen/heeft heeft/gelezen

After Anna the letter read/had had/read

[Conjunction] [Subject] [Object] [verb participle]/[finite verb] [finite verb]/[verb participle]

“After Anna had read the letter”

The word order pattern in the covert transfer condition is the placement of prepositional phrases in the main clause. Prepositional phrases can be placed before the verb particle as in sentence (4), which is referred to as the middlefield position, or after the verb as in sentence (5), which is referred to as the postfield position. Both middlefield and postfield position are used in Dutch and German, but research has shown that prepositional phrases in Dutch occur more often in the postfield than in German (De Sutter & Van de Velde, 2007; Van Oost et al., 2016). Transfer of this pattern from Dutch to German (i.e., the use of the postfield position) therefore does not lead to a new construction in German, but rather to an increase of the construction with parallel Dutch form, in this case, an increase in the use of the postfield position.

(4) Ich habe den Mann in der Stadt gesehen. (Fitch, 2011, p. 372)

I have the man in the city seen

[Subject] [finite verb] [Object] middlefield PP [verb participle]

“I saw the man in the city.”

(5) Ich habe den Mann gesehen in der Stadt. (Fitch, 2011, p. 372)

I have the man seen in the city

[Subject][finite verb] [Object] [verb participle] postfield PP

“I saw the man in the city.”

Procedure

Participants first signed a consent form and provided demographic information (i.e., age, gender, native language[s]). They then participated in the production task, which took around 25 minutes to complete. For the lvcs, participants were presented with a sentence that contained one of the lvcs (see sentence (6) for an example; parts that participants had to fill in are underlined). The light verbs were replaced with lines, equally long in every test item, and participants were instructed to fill in a verb. To test the participants’ use of the um-constructions and the different word order patterns, they were provided with two sentence parts and were instructed to combine them into one sentence. The first part of the new sentence was provided (see sentence (7) and (8) for an example; parts that participants had to fill in are underlined). The order in which the stimuli were presented to the participants was individually randomized.

(6) Anna hat Peter das Versprechen gegeben/*getan, ihm bei seinem Umzug zu helfen. Anna promised Peter ( gave/did the promise to Peter) to help him with his move.

(7) Was hat Peter versucht? Anna anrufen. Peter hat versucht, (*um) Anna anzurufen. Peter tried to call Anna.

(8) Anna liest den Brief. Danach geht sie einkaufen. Nachdem Anna den Brief gelesen hat/*hat gelesen , geht sie einkaufen. After Anna read (read had/*had read) the letter, she goes shopping.

Data analysis

Data exclusion

In line with our preregistration, we excluded constructions that (a) had a percentage of alternative responses higher than 25% (lvc: N =9; um: N = 0, order: N = 0). In addition, we asked the native German speakers living in Germany to judge the constructions for their acceptability in German, and excluded any constructions that were not judged as expected based on their condition. That means that we excluded constructions in the covert transfer condition when the construction with and/or without parallel Dutch form was judged as unacceptable by more than 25% of the speakers without Dutch contact (lvc: N = 3; um: N = 4; order: N = 7), and in the overt transfer condition when the constructions with parallel Dutch form was judged as acceptable by more than 25% of the speakers without Dutch contact (lvc: N = 1; um: N = 1; order: N = 7⁴).

There were three possible responses in the production task: the Dutch-like construction (e.g., Wecker stellen), the non-Dutch-like construction (e.g., *Wecker setzen), or an alternative response. For the purpose of the analyses, we excluded all alternative responses, resulting in a binary variable (Dutch-like construction used, Dutch-like construction not used).⁵

Analysis extent of transfer

The first analysis focused on potential differences between conditions (i.e., the between-participant variable speaker group: participants with Dutch contact, participants without Dutch contact, and the within-participants variables related to the construction types: lexically specific lvcs, partially schematic um-constructions, fully schematic word order patterns; covert and overt transfer) in terms of the overall amount of transfer. This was done by building a conditional inference tree predicting speakers’ production choices.⁶ This statistical method is well-equipped for dealing with cases of (complete) separation as is the case in our data (Tagliamonte & Baayen, 2012).⁷ To build a tree, data are repeatedly split based on the best binary split of levels of the independent variables, that is, based on the split that explains most of the variation in the outcome variable. The data keep being split until no further split predicts a significant amount of the remaining variation (Tagliamonte & Baayen, 2012). The alpha-value for the splits was set to 0.1 to be able to explore trends in the data.

Analysis speaker and item variation

Further inference trees were built to explore both speaker and item variation within each condition. To do that, the speaker and item variance was calculated as the squared difference between the average percentage of using the Dutch-like construction (e.g., *Wecker setzen) for each condition and the percentage of each individual speaker/item. As such, the speaker/item variance provides a measure of the overall variance: the more speakers/items deviated from each other regarding their use of the Dutch-like constructions, the higher the speaker/item variance for that construction. In addition, we plotted the results for the individual speakers and items to explore their variation.

Results

Analysis extent of transfer

Figure 1 plots the extent of speakers using the Dutch-like construction based on the speaker group (participants with Dutch contact, participants without Dutch contact), the transfer type (overt transfer, covert transfer), and the construction type (lvc, um, order). The analysis showed that speakers—both with and without Dutch contact—were more likely to use the Dutch-like constructions in the covert transfer condition, when the Dutch-like constructions exist in standard German (e.g., Einfluss ausüben) as opposed to the overt transfer condition when the constructions do not exist (e.g., *Wecker setzen).

Figure 1.

Production choices across the different conditions. The inferential tree predicts the speakers’ production choices (Dutch-like construction used, Dutch-like construction not used) based on the speaker group (participants with Dutch contact, participants without Dutch contact), the transfer type (overt transfer, covert transfer), and the construction type (lvc, um, order).

Within the covert transfer condition, the following pattern emerges: compared to the speakers without Dutch contact, the speakers with Dutch contact over-used the Dutch-like lvcs (e.g., more frequent use of Einfluss ausüben instead of Einfluss nehmen), they showed an overall trend to under-use the Dutch-like um-constructions (i.e., less frequent use of um in the construction alt genug (um) in which the use of um is optional, p = .09), and they showed no differences in their use regarding the order patterns (i.e., no difference in the frequency of using the postfield position).

Within the overt transfer condition, speakers with Dutch contact generally used more Dutch-like constructions than the speakers without Dutch contact. This difference was only present for the lvcs and the um-constructions; for the order patterns there was no difference as both groups never used the Dutch-like word order (i.e., never using the pattern [auxiliary verb] such as *hat gelesen). There was no difference in the overall amount of transfer between the lvcs and um-constructions (see Figure 1).

Analysis speaker and item variation

Figure 2 plots the extent of speaker variation within the different constructions, Figure 3 plots the corresponding item variation. For both speaker and item variation, there was more variation in the covert than in the overt transfer condition. In the covert condition, both the Dutch-like (e.g., Einfluss ausüben) and the non-Dutch-like construction (e.g., Einfluss nehmen) exist in standard German and the results showed that speakers also made use of these different options, hence the higher level of variation as compared with the overt transfer condition. For instance, for the lvcs, most speakers — both with Dutch contact and without Dutch contact — used the Dutch-like lvcs in around 70% of the cases, but there were also speakers who used these constructions much more or much less frequently, with percentages ranging from 0% to 100%. Regarding the lvc item variation, most items increased in the frequency of the Dutch-like lvc for the speakers with Dutch contact. For example, 59.3% of the speakers with Dutch contact used the combination Einfluss ausüben “to practice influence” instead of Einfluss nehmen “to take influence” (compared to 27.6% of the speakers without Dutch contact, frequency increase: +31.7%), and 82.1% of the speakers with Dutch contact used the combination Eid ablegen “to put down an oath” instead of Eid leisten “to pay an oath” (compared to 50.0% of the speakers without Dutch contact, frequency increase: +32.1%). However, note that there are also items for which the frequency of using the Dutch-like lvc actually decreased for the speakers with Dutch contact (see Figure 3; Anweisungen geben “to give instructions” instead of Anweisungen erteilen ‘to grant instructions’: -12.9%, Erlaubnis geben “to give permission” instead of Erlaubnis erteilen ‘to grant permission’: -4.1%, Rat geben “to give advice” instead of Rat erteilen ‘to grant advice’: -2.7%). We come back to these different patterns and what they reveal about speakers’ mental level of schematicity in the discussion section of this paper.

Figure 2.

Speaker variation across the different conditions. The inferential tree predicts the speaker variance based on the speaker group (participants with Dutch contact, participants without Dutch contact), the transfer type (overt transfer, covert transfer), and the construction type (lvc, um, order). The six bottom graphs plot the distribution for the speakers in the overt (left) and covert transfer condition (right) for the lvcs, um-constructions and word order patterns.

Figure 3.

Item variation across the different conditions. The inferential tree predicts the item variance based on the speaker group (participants with Dutch contact, participants without Dutch contact), the transfer type (overt transfer, covert transfer), and the construction type (lvc, um, order). The six bottom graphs plot the distribution for the items in the overt (left) and covert transfer condition (right) for the lvcs, um-constructions and word order patterns.

Regarding the covert um-constructions, the analysis comparing the overall extent of transfer (section “Analysis extent of transfer”) revealed that speakers with Dutch contact tend to under-use the constructions with um compared with the speakers without Dutch contact. A closer look at the speaker variation revealed that there actually is a large group of speakers with Dutch contact that never used um in these constructions (see Figure 2, bottom middle graph). This group of non-users explains the overall under-use, as most other speakers used um as frequently as the speakers without Dutch contact or even more frequently (e.g., note the increase in speakers who used um in 100% of the cases).

This pattern (i.e., avoidance in all cases, usage in the majority of the cases) also has consequences regarding the um item variation: speakers with Dutch contact generally showed less item variation than the speakers without Dutch contact (p = .07, see Figure 2). Specifically, the usage of um decreased for the items that were frequently used with um by the speakers without Dutch contact and their usage of um increased for the items that were frequently being used without um by the speakers without Dutch contact. For example, the construction alt genug, (um), which can be used both with and without the complementizer um in standard German, is used with um by almost all the speakers with Dutch contact and without Dutch contact, with the exception of the speakers with Dutch contact who avoid the use of um in all cases (frequency decrease: −27.6%). On the contrary, the construction klug genug, (um) is used with um by only 7.4% of the speakers without Dutch contact. This percentage increases to 20.0% for the speakers with Dutch contact (frequency increase: +12.6%), as many of the speakers with Dutch contact used um in the majority or even in all cases. Overall, this pattern leads to a leveling across the different instances (and therefore a decreased item variation) for the speakers with Dutch contact.

Regarding the overt transfer condition, the speakers without Dutch contact almost never used the Dutch-like construction. The speakers with Dutch contact, however, did use them and we see a number of interesting patterns emerge regarding their speaker and item variation. For the lvcs, there was relatively little speaker variation (see Figure 2), as most speakers with Dutch contact did not produce any or only few Dutch-like forms. Regarding the lvc item variation, the Dutch-like form was never used for most items (see Figure 3), but there were also some Dutch-like forms that were produced very frequently by the speakers with Dutch contact (e.g., *Grenzen stellen “to put boundaries” instead of Grenzen setzen “to set boundaries” was used by 38.5% of the speakers with Dutch contact and *Initiative nehmen “to take initiative” instead of Initiative ergreifen “to seize initiative” by 29.6%). As a result, the item variation is slightly higher for the speakers with Dutch contact than for the speakers without Dutch contact, who almost never used the forms (p = .08, see Figure 3). The opposite is true for the overt um-constructions: there was a lot of speaker variation and relatively little item variation for these constructions. As shown in Figure 2, the speakers with Dutch contact greatly differed in the extent to which they used um in these constructions, with percentages ranging from 0.0% to 80.0% (note again that the use of um in these constructions is ungrammatical in standard German). The new usage of um in these constructions seems to affect all instances to a similar degree as—very different from the pattern for the overt lvcs—items consistently increased in usage (see Figure 3).

Discussion

In this study, we compared different constructions (i.e., lexically specific lvcs, partially schematic um-constructions, and fully schematic word order patterns) regarding their transfer patterns in the case of covert and overt transfer. We compared these constructions in terms of their overall percentage of transfer and in terms of their speaker and item variation. The aim of this approach was to study language transfer from a usage-based perspective and thereby to learn more about both the transfer itself as well as about the usage-based principles (see also Zenner et al., 2019).

Regarding the first aim, that is, to learn more about language transfer by studying it from a usage-based perspective, our results strongly suggest that transfer does indeed depend on a construction’s level of schematicity. To illustrate this, we first focus on the results in the overt transfer condition, which revealed interesting differences across construction types that might be attributed to schematicity. For the lvcs, there was relatively little speaker variation and a lot of item variation, with many Dutch-like lvcs being never used and others by up to 38.5% of the speakers with Dutch contact (*Grenzen stellen “to put boundaries” instead of Grenzen setzen “to set boundaries”). Apparently, there was something in the representation of these highly transferred lvcs—and not in the ones of the other lvcs—that triggered the transfer (see Barking, 2022, for a follow-up analysis that suggests that perceived similarity between lvcs plays a decisive role here). This in turn suggests that the lvcs are, as we had also previously hypothesized, stored as lexically specific items and, as a result, that speakers transfer these constructions on an idiosyncratic basis. The um-constructions, on the contrary, show a lot of speaker variation and relatively little item variation (i.e., the exact opposite of the lvc pattern). Some speakers never used um in these constructions (i.e., when it is ungrammatical in standard German), whereas others used um in many or even in the majority of the cases. Besides, the use of um increased across all instances (see Figure 3). Overall, this suggests that the um-constructions are stored on a (partially) schematic level and that many speakers do not only show transfer for some of its instances—as they do for the lvcs—but rather for the pattern as a whole. This transfer therefore also seems to be much more pervasive in a speakers’ language use (e.g., the highest percentage of um transfer was 80.0% compared to only 40.0% for the lvcs), showing that, once transferred, the use of um in these new constructions can quickly spread through a speaker’s language use. For the word order pattern, we observed no transfer and therefore also no speaker and item variation, potentially because the word order pattern is so frequent and therefore so entrenched that it preempts the use of any alternatives.

In the covert transfer condition, we observe a similar pattern, in that there are differences between the speakers with Dutch contact and without Dutch contact regarding the lvcs and um-constructions but not regarding the word order pattern. Crucially, for the um-constructions, speakers with Dutch contact showed a trend to under-use um in these constructions (see Figure 1), which is opposite to how this construction is used in Dutch. This overall under-use seems to be the result of a relatively large group of speakers with Dutch contact who avoided the use of um in all cases (see Figure 2). A possible explanation is that these speakers realized that using um similarly to how om is used in Dutch can lead to unconventional constructions in German (i.e., the constructions in the overt transfer condition). They might therefore have decided to avoid the use of um to avoid any potential influences from Dutch. This process is also referred to as hypercorrection and has frequently been reported in settings in which speakers—consciously or unconsciously—feel that they have to meet a certain language standard (e.g., Hubers et al., 2020; Kupisch, 2014). For instance, Hubers et al. (2020) showed that some native Dutch speakers avoid the use of the indirect object pronoun hun “them” in Dutch, even in cases when it should be used, likely to avoid the use of hun as a subject pronoun, which is widespread but strongly disapproved of in Dutch society. Just as the German-Dutch bilinguals investigated in this study, they thus likely want to avoid the supposedly incorrect use of um/hun and, potentially because they do not know the exact rules, over-generalize the avoidance in one context (i.e., attributive clauses for um, subject position for hun) in cases when um/hun can or even should be used. Interestingly, our data suggest that something similar is also happening for the covert lvcs: while there is an increase in use for the Dutch-like lvcs for almost all items, there are also some items that decreased in frequency, all of them with the light verb geven-geben “to give” (Anweisungen geben instead of Anweisungen erteilen: -12.9%, Erlaubnis geben instead of Erlaubnis erteilen: -4.1%, Rat geben instead of Rat erteilen: -2.7%). Just as for the um-constructions, the speakers who did not use the light verb geben, might potentially have noticed a pattern, namely that geben and geven are often used in similar constructions, and therefore decided to avoid using geben in these constructions as a way to avoid transfer from Dutch.

When it comes to the second aim of the study — reflecting on the question what the transfer patterns can tell us about the usage-based principles — the results of this study show the importance of individual variation as a way to both clarify patterns and to reveal new patterns regarding speakers’ language processing. They also show that, when investigating the concept of schematicity, it is necessary to view the different levels of schematicity as a continuum and to account for individual variation regarding these levels. The following paragraphs discuss each of these three points.

First, our results showed that individual variation can be used to clarify patterns. In particular, there were several cases in which the overall usage pattern of a construction and the usage pattern for the individual speakers and items clearly differed from each other. For example, in the overt transfer condition, there was no difference in the overall extent of transfer between the lvcs and the um-constructions, but we did observe clear differences regarding their variation patterns, with more speaker variation for the um-constructions than for the lvcs and more item variation for the lvcs than for the um-constructions. Another example is the transfer in the covert transfer condition, in which the speakers with Dutch contact under-used um compared to the speakers without Dutch contact. A closer look at the speaker variation revealed the complexities of this pattern: while many speakers with Dutch contact indeed drastically under-used um in these constructions (i.e., they never used um at all), others actually did use it just as frequently as or even more frequently than the speakers without Dutch contact. These examples illustrate the benefits of analyzing both speaker and item variation, as this analysis allows for a much more nuanced picture of language transfer than the overall usage pattern can provide.

Second, this study demonstrates how individual variation can be used as a tool to reveal new patterns regarding speakers’ language processing and production. Specifically, in this study, we used individual variation as a way to investigate a construction’s level of schematicity. To do that, we first selected constructions that were likely to differ in regard to their schematicity level and then verified this initial classification by looking at both speaker and item variation in the constructions’ transfer patterns. Overall, using this approach revealed interesting differences regarding how constructions are stored and therefore also transferred (e.g., lvc: high item variation, low speaker variation, indicating that these constructions are likely to be stored as lexically specific items and therefore transferred on an idiosyncratic basis; um: low item variation, high speaker variation, indicating that these constructions are likely to be stored as partially schematic constructions resulting in pattern-based transfer). Generally, evidence for such different schematicity levels is hard to find in production data (see, for example, Quick et al. (2021, p. 6) who argue that “which of the three types of unit [lexically specific, partially schematic, fully schematic] are instantiated in a speaker’s actual utterance is, in general, impossible to say”). The fact that this study is able to find this evidence demonstrates the benefits of studying schematicity within a language contact setting, in which new constructions at different levels of schematicity enter a language, as well as the benefits of approaching individual variation in this new way, that is, in a way that does not view variation as merely noise that hinders research, but instead as a tool that can actually help us investigate speakers’ language use.

Third, our analysis illustrates the need for a continuous view of schematicity, as proposed by the usage-based approach. For example, as argued above, the patterns in the data showed that the um-constructions are likely to be stored as partially schematic constructions, however, this does not necessarily have to apply to all speakers. Instead, some speakers used only a few of the new um-constructions and in a follow-up study we showed that this also applies to their grammaticality judgments suggesting that these speakers might have transferred these constructions on a more idiosyncratic level, similar to the pattern for the lexically specific lvcs. From a usage-based perspective, this is actually to be expected: a schematic representation is assumed to be something that speakers extract from their language input themselves by categorizing and generalizing over stored exemplars (Bybee, 2011; Ibbotson, 2013, see also section “Schematicity continuum”). As a result, it is perfectly plausible that speakers arrive at the same pattern on the basis of different lower-level generalizations, and it is even possible that some speakers do not represent the most schematic patterns at all. Instead, they may also rely on stored lexical items or lower-level generalizations to structure their language output (see Dąbrowska, 2004, 2012, 2020), such as speakers transferring the construction *versuchen um but not any other overt um-constructions to their German.

Interestingly, this mismatch between a construction’s general level of schematicity and its representation in each individual speaker’s mental representation also seems to be happening the other way around: as argued above, transfer patterns suggest that the lvcs are generally stored as lexically specific items. This also makes sense conceptually: speakers have to store each noun-light verb pairing as a fixed unit (Basarić & Petrič, 2015; Wittenberg, Jackendoff, et al., 2014), otherwise speakers would constantly use unconventional combinations such as *make a walk or *do a decision. However, speakers still seem to find patterns in these constructions and start to generalize based on these patterns (e.g., the pattern that geven-geben “to give” are used similarly in Dutch and German). Overall, these findings clearly demonstrate the complexities of schematicity, with speakers making both fewer and more generalizations than expected. As a result, it becomes crucial for future research to carefully assess a construction’s level of schematicity and to account for variation between both speakers and items.

Conclusion

Our results demonstrate that usage-based concepts such as the schematicity continuum as the result of a number of basic cognitive mechanisms (e.g., entrenchment, categorization, generalization, see section “Schematicity Continuum”) can account for a wide range of contact phenomena, for example, covert and overt language transfer (e.g., the over-use of Dutch-like existing and non-existing lvcs such as Einfluss ausüben and *Wecker setzen), hypercorrection (e.g., the avoidance of um in constructions in which um can be used in standard German), or the stability of certain constructions (e.g., no contact effects in the overt order condition, potentially because the German word order was highly entrenched for all speakers). Of course, the fact that language contact can be explained by a set of basic mechanisms certainly does not mean that we see it as a simple process. On the contrary, there can be many complex interactions between the cognitive mechanisms—as also illustrated by this study—resulting in very different contact effects. As such, the results of this study demonstrate the benefits of studying the cognitive mechanisms in language contact settings, which due to their dynamic nature can reveal patterns that otherwise might remain hidden.

Footnotes

Appendix 1 Authors’ note

One of the authors of this article is a member of the Editorial board of the journal and did not participate in the peer-review process.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This publication is part of the project “Two Languages in One Mind” (with project number PGW.18.019) which is financed by the Dutch Research Council (NWO).

ORCID iD

Marie Barking

Notes

Author biographies

Marie Barking is a PhD student at the Department of Communication and Cognition and at the Department of Culture Studies at Tilburg University. Her research focuses on bilingual speakers’ language use, particularly the phenomenon of language transfer, and what that can reveal about the cognitive and social factors that shape speakers’ mental language representations.

Maria Mos is an associate professor at the Department of Communication and Cognition at Tilburg University. Her research focuses on linguistic representations in dynamic situations such as language learning, and on the question how communicative behavior informs us about people’s linguistic representations.

Ad Backus is a full professor of Linguistics and Sociolinguistics at Tilburg University. His research focuses on various aspects of bilingualism, and on the integration of a usage-based approach and the empirical study of language contact. Prominent topics in his research have been codeswitching and contact-induced language change.

References

Aikhenvald

A. Y.

(2003). Mechanisms of change in areal diffusion: New morphology and language contact. Journal of Linguistics, 39(1), 1–29. https://doi.org/10.1017/S0022226702001937

Appel

Muysken

(2006). Language contact and bilingualism. Amsterdam University Press.

Barking

Backus

Mos

(2022). Similarity in language transfer–investigating transfer of light verb constructions from Dutch to German. Journal of Language Contact, 5, 198–239.

Backus

Doĝruöz

S. A.

Heine

(2011). Salient stages in contact-induced grammatical change: Evidence from synchronic vs. diachronic contact situations. Language Sciences, 33(5), 738–752. https://doi.org/10.1016/j.langsci.2011.02.004

Basarić

Petrič

(2015). Funktionsverbgefüge als Einheiten im mentalen Lexikon [light verb constructions as units in the mental lexicon]. Jezikoslovje, 3(2–3), 169–186.

Biber

Johannson

Leech

Conrad

Finegan

(1999). Longman grammar of spoken and written English. Pearson Educated.

Blumenthal-Dramé

(2016). Entrenchment from a psycholinguistic and neurolinguistic perspective. In Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge (pp. 129–152). https://doi.org/10.1515/9783110341423-007

Brons-Albert

(1992). Verlust der Muttersprache in fremdsprachiger Umgebung [loss of the native language in a foreign-language environment]. Info DaF, 19(3), 315–325. https://doi.org/10.1515/infodaf-1992-190302

Brons-Albert

(1994). Interferenzfehler in der Muttersprache von in den Niederlanden lebenden Deutschen [Interference errors in the native language by Germans living in the Netherlands]. In Spillner

(Ed.), Nachbarsprachen in Europa (pp. 96–104). Peter Lang.

10.

Butt

(2010). The light verb jungle: Still hacking away. In Amberber

Baker

Harvey

(Eds.), Complex predicates: Cross-linguistic perspectives on event structure (pp. 48–78). Cambridge University Press. https://doi.org/10.1017/CBO9780511712234.004

11.

Bybee

J. L.

(2006). From usage to grammar: The mind’s response to repetition. Language, 82(4), 711–733. https://doi.org/10.1353/lan.2006.0186

12.

Bybee

J. L.

(2011). Language, usage, and cognition. Cambridge University Press.

13.

Clyne

(2003). Dynamics of language contact: English and immigrant languages. Cambridge University Press. https://doi.org/10.1017/CBO9780511606526

14.

Coussé

(2003). Volgordevariatie en herinterpretatie in de tweeledige werkwoordelijke eindgroep in de bijzin [Word order variation and re-interpretation in two-part verbal final phrases in dependent clauses]. Taal en Tongval, 55(2), 138–156.

15.

Croft

(2000). Explaining language change: An evolutionary approach. Pearson Education.

16.

Dąbrowska

(2004). Rules or schemas? Evidence from Polish. Language and Cognitive Processes, 19(2), 225–271. https://doi.org/10.1080/01690960344000170

17.

Dąbrowska

(2012). Different speakers, different grammars—Individual differences in native language attainment. Linguistic Approaches to Bilingualism, 2(3), 219–253. https://doi.org/10.1075/lab.2.3.01dab

18.

Dąbrowska

(2020). Language as a phenomenon of the third kind. Cognitive Linguistics, 31(2), 213–229. https://doi.org/10.1515/cog-2019-0029

19.

De Smet

. (2016). Entrenchment effects in language change. In Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge (pp. 75–99). https://doi.org/10.1515/9783110341423-005

20.

De Sutter

Van de Velde

. (2007). Do the mechanisms that govern syntactic choices differ between original and translated language? A corpus-based translation study of PP placement in Dutch and German. In R. Xiao, L. He, & M. Yue (Eds.), Proceedings of the international symposium on using corpora in contrastive and translation studies (UCCTS 2008) (pp. 1–38). Hangzhou.

21.

DudenOnline. (n.d.). Verwendung von “um zu” vs. “zu.” [use of “um zu” and “zu”]. https://www.duden.de/sprachwissen/sprachratgeber/um-zu-zu#:~:text=Vor%20allem%20in%20der%20geschriebenen,um%20zu%20und%20zu%20austauschbar

22.

e-ANS. (n.d.). Het gebruik van om in beknopte bijzinnen met te + infinitief [The use of “om” in reduced dependent clauses with “te” + infinitive]. https://e-ans.ivdnt.org/topics/pid/ans190303lingtopic

23.

Fitch

G. W.

(2011). Changes in frequency as a measure of language change. Studies on German-Language Islands, 123, 371–384. https://doi.org/10.1075/slcs.123.14wes

24.

Flores

(2010). The effect of age on language attrition: Evidence from bilingual returnees. Bilingualism: Language and Cognition, 13(4), 533–546. https://doi.org/10.1017/S136672890999054X

25.

Goldberg

A. E.

(1995). Constructions: A construction grammar approach to argument structure. University of Chicago Press.

26.

Heine

Kuteva

(2003). On contact-induced grammaticalization. Studies in Language, 27(3), 529–572. https://doi.org/10.1075/sl.27.3.04hei

27.

Heine

Kuteva

(2005). Language contact and grammatical change. Cambridge University Press. https://doi.org/10.1017/CBO9780511614132

28.

Hubers

Trompenaars

Collin

Schepper

K. D. E.

De Hoop

(2020). Hypercorrection as a by-product of education. Applied Linguistics, 41(4), 552–574. https://doi.org/10.1093/applin/amz001

29.

Ibbotson

(2013). The scope of usage-based theory. Frontiers in Psychology, 4, Article 255. https://doi.org/10.3389/fpsyg.2013.00255

30.

Kühl

Braunmüller

(2014). Linguistic stability and divergence: An extended perspective on language contact. In Braunmüller

Höder

Kühl

(Eds.), Stability and divergence in language contact: Factors and mechanisms (pp. 1–38). John Benjamins.

31.

Kupisch

(2014). Adjective placement in simultaneous bilinguals (German—Italian) and the concept of cross-linguistic overcorrection. Bilingualism: Language and Cognition, 17(1), 222–233. https://doi.org/10.1017/S1366728913000382

32.

Langacker

R. W.

(1987). Foundations of cognitive grammar (1st ed.). Stanford University Press.

33.

Langacker

R. W.

(2016). Entrenchment in cognitive grammar? In Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge (pp. 39–56). De Gruyter. https://doi.org/10.1515/9783110341423-003

34.

Mougeon

Nadasdi

Rehner

(2005). Contact-induced linguistic innovations on the continuum of language use: The case of French in Ontario. Bilingualism: Language and Cognition, 8(2), 99–115. https://doi.org/10.1017/S1366728905002142

35.

Quick

A. E.

Backus

Lieven

(2021). Entrenchment effects in code-mixing: Individual differences in German-English bilingual children. Cognitive Linguistics, 32(2), 319–348. https://doi.org/10.1515/cog-2020-0036

36.

Ribbert

Kuiken

(2010). L2-induced changes in the L1 of Germans living in the Netherlands. Bilingualism: Language and Cognition, 13(1), 41–48. https://doi.org/10.1017/S1366728909990320

37.

Schmid

H.-J.

(2007). Entrenchment, salience, and basic levels. In The Oxford handbook of cognitive linguistics (pp. 117–138). https://doi.org/10.2307/j.ctvgc6251

38.

Schmid

H.-J.

(2016). A framework for understanding linguistic entrenchment and its psychological foundations. In Schmid

H. J.

(Ed.), Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge (1st ed., pp. 9–36). De Gruyter.

39.

Schmid

M. S.

Köpke

(2017). The relevance of first language attrition to theories of bilingual development. Linguistic Approaches to Bilingualism, 7(6), 637–667. https://doi.org/10.1075/lab.17058.sch

40.

Swerts

van Wijk

(2005). Prosodic, lexico-syntactic and regional influences on word order in Dutch verbal endgroups. Journal of Phonetics, 33(2), 243–262. https://doi.org/10.1016/j.wocn.2004.09.002

41.

Tagliamonte

S. A.

Baayen

R. H.

(2012). Models, forests, and trees of York English: Was/were variation as a case study for statistical practice. The Mind Research Repository, 24(2), 135–178. https://doi.org/10.1017/S0954394512000129

42.

Ten Cate

A. P.

Lodder

H. G.

Kootte

. (1998). Deutsche Grammatik. Eine kontrastiv deutsch-niederländische Beschreibung für den Zweitsprachenerwerb [German grammar. A contrastive German-Dutch description for second language acquisition]. Coutinho.

43.

Tomasello

(2000a). First steps toward a usage-based theory of language acquisition. Cognitive Linguistics, 11(1–2), 61–82. https://doi.org/10.1515/cogl.2001.012

44.

Tomasello

(2000b). The item-based nature of children’s early syntactic development. Trends in Cognitive Sciences, 4(4), 156–163. https://doi.org/10.1016/S1364-6613(00)01462-5

45.

Treffers-Daller

(2012). Grammatical collocations and verb-particle constructions in Brussels French: A corpus-linguistic approach to transfer. International Journal of Bilingualism, 16(1), 53–82. https://doi.org/10.1177/1367006911403213

46.

Treffers-Daller

Mougeon

(2005). The role of transfer in language variation and change: Evidence from contact varieties of French. Bilingualism, 8(2), 93–98. https://doi.org/10.1017/S1366728905002191

47.

Treffers-Daller

Sakel

(2012). Why transfer is a key aspect of language use and processing in bilinguals and L2-users. International Journal of Bilingualism, 16(1), 3–10. https://doi.org/10.1177/1367006911403206

48.

Van Oost

Willems

De Sutter

. (2016). Asymmetric syntactic patterns in German-Dutch translation: A corpus-based study of the interaction between normalisation and shining through. International Journal of Translation, 28(1), 7–25.

49.

Verhagen

Mos

Backus

Schilperoord

(2018). Predictive language processing revealing usage-based variation. Language and Cognition, 10(2), 329–373. https://doi.org/10.1017/langcog.2018.4

50.

Verhagen

Mos

Schilperoord

Backus

(2020). Variation is information: Analyses of variation across items, participants, time, and methods in metalinguistic judgment data. Linguistics, 58(1), 37–81. https://doi.org/10.1515/ling-2018-0036

51.

Wittenberg

Jackendoff

Kuperberg

Paczynski

Snedeker

Wiese

(2014). The processing and representation of light verb constructions. In Bachrach

Roy

Stockall

(Eds.), Structuring the argument: Multidisciplinary research on verb argument structure (pp. 61–80). John Benjamins.

52.

Wittenberg

Paczynski

Wiese

Jackendoff

Kuperberg

(2014). The difference between “giving a rose” and “giving a kiss”: Sustained neural activity to the light verb construction. Journal of Memory and Language, 73, 31–42. https://doi.org/10.1016/j.jml.2014.02.002

53.

Zenner

Backus

Winter-Froemel

(2019). Introduction. In Zenner

Backus

Winter-Froemel

(Eds.), Cognitive contact linguistics: Placing usage, meaning and mind at the core of contact-induced variation and change (pp. 1–23). Mouton de Gruyter.

Investigating language transfer from a usage-based perspective

Abstract

Aims and objectives:

Design:

Data and analysis:

Findings/conclusions:

Originality:

Significance/implications:

Keywords

Introduction

Schematicity continuum

Covert and overt transfer

Method

Design

Participants

Stimuli

Lexically specific lvcs

Partially schematic um-constructions

Fully schematic word order patterns

Procedure

Data analysis

Data exclusion

Analysis extent of transfer

Analysis speaker and item variation

Results

Analysis extent of transfer

Analysis speaker and item variation

Discussion

Conclusion

Footnotes

Appendix 1

Authors’ note

Declaration of conflicting interests

Funding

ORCID iD

Notes

Author biographies

References