1
The Acquisition of Modal Uses of the Dutch Ook by English and German Second Language Learners of Dutch
Research Master Thesis Universiteit van Amsterdam M.J.M. Post, student number: 11145080 Supervisors: Lotte Hogeweg and Petra Sleeman
Second reader: Ineke Vedder
ABSTRACT
This study tries to find an answer to the question whether L1-transfer is a factor in the acquisition of the Dutch particle ook. To answer this question, I compared 9 English and 20 German learners of Dutch to 17 Dutch native speakers with a grammaticality judgment task. I also performed a sociolinguistic questionnaire to see which other factors affected the acquisition. The results of my study indeed point to L1-transfer. The English speakers performed worse (p<0.05) on uses of ook that did not occur in their mother tongue but were present in German and Dutch. Aside the L1, also the frequency of a reading of ook and the proficiency affected the results. The results of my study could be used for educational purposes. For example, the inclusion of discourse markers in educational methods seems to be helpful to L2-speakers, especially if these markers are not omnipresent in their native language.
2
1. Introduction
Particles play an important role in Dutch. They are used often and are necessary to create a natural-sounding discourse. According to Van der Wouden & Caspers (2010), there are three types of particles in Dutch: focus particles, modal particles and discourse particles. They describe these three types of particles in the following way: Focus particles emphasize only one part of a sentence or they suggest a contrast between one part of a sentence and another. Examples are vooral and met name. Modal particles usually occur in the middle of a sentence and modify the entire sentence. For example, the maar in Doe je trui maar uit! indicates that the sentence is an advice rather than an order. Modal particles are usually phonetically unstressed. Ultimately, discourse particles organize the conversation and can occur in any place in the sentence. An example is hoor in Frankrijk is een groot land hoor. In this sentence, hoor indicates that the speaker does not tolerate objections. The present study is about the acquisition of the second type of particles in Dutch: modal particles.
Van der Wouden & Caspers (2010) argue that Dutch distinguishes itself by the modal particles. Together with German, Mandarin and Classical Greek, Dutch is one of the languages where these particles are most frequent. Other languages, like English and French are less productive in the use of modal markers. The frequency of modal particles in Dutch makes it essential for a second language learner to acquire how to apply these markers. For learners who have fewer particles in their mother tongue, this can be challenging.
Previous research has explored the acquisition of various Dutch modal markers by differing groups of speakers: Caspers & der Wouden (2010) showed that Spanish L2-speakers of Dutch have difficulties with the perception of modal particles, and that prosody plays a role in their difficulties. A later study by van Balen (2010) indicates that those difficulties also occur in the production of these particles. Her results indicate that Spanish L2-speakers of Dutch use fewer particles than native L2-speakers and that the number of particles increases when proficiency improves.
In the present study, I focus on the L2-acquisition of the Dutch discourse marker: ook. I choose this particle, because it has not been studied before from the viewpoint of L2-acquisition, even though it induces some interesting problems for the learners, such as the prosody and the placement of the particle within a sentence. The unstressed variant of ook is mostly modal, whereas the stressed variant is additive. In this study, I compare English and German learners of Dutch in their perception of the particle ook. In German, the translational equivalent of ook (auch) is used in a comparable way whereas the English equivalent (also) is used differently. This reveals the role of the mother tongue in the acquisition of this particle.
3 The rest of this text is organized as follows: First, I will present some background information on the L2-acquisition of particles and on the features of additive markers (to which ook belongs) in English, German and Dutch. Based on this literature, I will formulate my research questions and hypotheses. Then, I will explain my method, which not only includes a grammaticality judgment task, but also a corpus investigation. The corpus investigation is a helpful tool to choose the material for the grammaticality judgment task, given that it provides us with the frequencies of the different uses in Dutch. After that, I will discuss the results of these investigations. Based on the results, I will formulate a discussion and a conclusion.
2. The L2 Acquisition of Pragmatics and Discourse Markers in Particular
In the handbook of Verschueren (1999: 11) pragmatics is defined by the following features:
- As the linguistics of language use
- Giving neither its own unit(s) of analysis nor its own correlational object(s)
- Constituting a general functional (i.e. cognitive, social and cultural) perspective on language
- With as its topic of investigation the meaningful functioning of language in actual use, as a complex form of behavior that generates meaning
- And serving within the realm of the language-related sciences, as a point of convergence for the interdisciplinary fields of investigation and as a latch between those and the components of the linguistics of language resources.
Pragmatics, thus, studies language use. It is concerned with the way language functions in social contexts and how people construct meaning and interpret meanings based on linguistic input. Topics that are studied in pragmatics are speech acts, linguistic politeness, conversational implicatures, verbal tenses and discourse markers (Kasper & Blum-Kulka, 1993:4–8) Compared to other linguistic disciplines like phonology and syntax, pragmatics is more interdisciplinary. It is studied by linguists, as well as social scientists, psychologists and other disciplines.
Pragmatics is also a topic in second language acquisition. It is difficult to acquire, as demonstrated by researchers who study the so-called field of interlanguage pragmatics. One of the arguments why pragmatics causes so many problems is provided by the Interface Hypothesis (Sorace & Filiaci, 2006), which states that narrow syntactic abilities are completely acquirable in an L2, whereas interface properties involving syntax and another cognitive domain may not be fully acquirable. In a later study Sorace & Serratrice (2009), also distinguish between the semantics-syntax interface and the pragmatics-syntax interface. In the first interface, two formal systems are involved. A mistake at this interface causes ungrammatical
4 utterances. In the latter, it is about the pragmatic conditions that determine the appropriateness of a sentence. That means that a form might be syntactically correct at this interface, but pragmatically infelicitous. Since it does not cause absolute grammatical mistakes, the pragmatics syntax interface is found to lead to more acquisitional problems than the semantics syntax interface.
Summarizing the previous paragraph, most L2 speakers do not reach a native-like level in their pragmatics and suffer from incomplete acquisition. According to Bardovi-Harlig (2012), the pragmatic competence of an individual L2-speaker is affected by many different factors, both language internal and language external factors. Language internal factors include the L1 of the speaker, the grammatical competence in the L2 and the frequency of a construction in the input and language external factors the environment where the speaker learned the language, the length of exposure to the L2 and the age of onset of the L2. In the following paragraphs, I will first discuss the language internal factors and consecutively the language external factors.
Multiple studies have focused on L1-influence in pragmatics. L1-influence surfaces in different forms. These different types of cross-linguistic influence are grouped under the term pragmatic transfer. According to Bou Franch (1998) the L1 can lead to avoidance of certain constructions, borrowing and incomplete acquisition of constructions that are not present in the L1. Pragmatic transfer takes place when an L2-speaker uses his L1-pragmatic knowledge to understand or bring across the pragmatics in his L2 (Kasper, 1997).
Bou Franch (1998) argues that the transfer is positive when the L1 helps the speaker to learn the pragmatics in the L2. This type of transfer takes place when the two languages are congruent with each other in the use of a certain construction. In her article she mentions the study of Weizman (1993) as an example. Her study found that Danish and German L2 speakers of Hebrew and English had few problems in acquiring indirect requesting strategies, because they also occurred in their L1. The transfer is negative when the L1 leads to the incorrect application or avoidance of certain pragmatic features in the L2. This tends to occur when the two languages are different in the use of a construction. An example of this mechanism is found in the study of Richards & Sukwiwat (1983), who studied the English of Japanese L1-speakers and found that they use inappropriate strategies to express gratitude in English due to the constructions existing in their mother tongue.
Other studies have focused on the relationship between pragmatic competence and grammatical competence. According to Kasper (2001), there are researchers who claim that grammatical knowledge precedes pragmatic knowledge and those who claim the opposite. The
5 grammar precedes pragmatics scenario comes in different forms. Some theories claim that the L2-speakers do know the pragmatic functions of a form, but do not apply them actively, whereas others argue that they do not even know the functions. An important supporter of the grammar precedes pragmatics is the study of Takahashi and Beebe's study (1987) that shows that more proficient speakers are more vulnerable to negative transfer, as they have the grammatical means to realize the transfer in the L2. In the opposite scenario, L2-speakers use the apt pragmatic strategy with an incorrect grammatical form. This scenario, for example, has been argued for by Schmidt (1983), who found that a Polish speaker of English, Wes, could use pragmatically appropriate strategies with incorrect grammar after having lived in an English speaking country for a prolonged period of time.
Finally, frequency and saliency also affect the acquisition. Ellis (2002) argues that a speaker gets more chance to create a form function connection, if it is frequent in the input. This association is necessary for the acquisition. As a counter argument to this theory, Gass and Mackey (2002) come forward with the study of Pienemann & Johnston (1986). This study states that, unless the frequency, it is hard for most L2-speakers to learn the third person’s -s in English. In a later article, Ellis (2006) argues that saliency also affects the acquisition. The third person’s singular -s is vulnerable to lenition, and not very prominent in the input. Due to this low prominence, L2-speakers have difficulty acquiring such forms.
Switching to the language external factors, second language acquisition is influenced by the environment where the speakers learned the language and the type of instruction they received. Bardovi‐Harlig & Dörnyei (1998) found that learners living in an English-speaking environment rated English pragmatic errors more severely than English grammatical errors, whereas English learners living in Hungary did the opposite. L2-speakers living in a native environment, thus, seem to have an advantage for the pragmatic part. The type of instruction also affects the pragmatic competence. Jeon & Kaya (2006) did a meta-analysis of different studies and found that a direct instruction on pragmatics is more beneficial compared to an indirect instruction or no instruction at all.
The age of onset is likely the most important language external factor. The study of Johnson and Newport (1989) found prove for a critical period in second language acquisition. Their results supported what they call the maturational state hypothesis, which states that “Early in life, humans have a superior capacity for acquiring languages. This capacity disappears or declines with maturation” (1989: 94). More specifically they found that from age seven on till adulthood, the capacity to acquire languages at a native-like level declined gradually, so there was not a specific point in which the critical period expired. Furthermore, their results disproved
6 previous statements that the decline only started after puberty. Also, they found that after puberty the decline fossilized. In the learners who acquired a language after puberty, there was considerable individual variation.
The factors that affect the acquisition are not the only point of discussion in Interlanguage Pragmatics. There is also a debate on how it should be studied. In this paper, we do this by means of studying discourse markers, but there are also other possibilities like speech acts, linguistic politeness and the use of verbal tenses. The term discourse marker has received many different interpretations in the literature. Some literature makes a strict distinction between discourse markers and discourse particles. An example of this approach is Fraser (1999: 950) who argues that the meaning of a discourse marker is procedural: It is used to signal a semantic relationship between two parts in a discourse, whereas a discourse particle gives a comment on an individual proposition in the discourse (1999: 942- 943). Others like Schiffrin (1988) do not make such a distinction. I will use the notion discourse marker for both categories.
Discourse markers have not been studied intensively in the field of interlanguage pragmatics. It has been demonstrated by different authors that L2-speakers start using discourse markers more strategically when they become more proficient, because their speech becomes more coherent in the course of time (House, 2013; Polat, 2011). It is also beneficial if the mother tongue resembles the second language in its particles (Hogeweg, De Hoop, Ramachers, Van Der Slik, & Wottrich, 2016). Besides that, discourse markers that resemble the L1 emerge earlier than those that do not (Andersen, 2014). Kasper & Rose (2002) suggest that some of the uses of a polyfunctional discourse marker are acquired earlier than others: A speaker can know a form without knowing all its functions.
Summarizing this section, we have seen a more general picture of how pragmatics is acquired by second language learners and how discourse markers are studied in interlanguage pragmatics. In the following sections, I will enter in detail about the functions of one of the Dutch discourse markers, ook. These functions play an essential role in the acquisition of this marker, which is the eventual focus of this study.
3. Functions of Ook
In this section, I will discuss the different functions of ook and its English and German translational equivalents. I start this section by defining the term additive marker, the category to which ook belongs. Next, I define the basic, non-modal functions of these markers in English, Dutch and German. In the following part, I go into detail about the modal functions of ook that
7 are shared between German and Dutch, but do not exist in English. In the last part of this section, I give a short overview of the functions of ook.
3.1. Additive Function
Additive markers are part of the broad category of discourse markers (Forker, 2016; Gast & Van der Auwera, 2011). In their basic sense, they introduce an alternative for an element in the proposition, as can be witnessed in the following example:
(1) John also plays in a soccer team.
The sentence in (1) has two different interpretations. First, John could be playing in another team besides the soccer team. Second, there can be another person, apart from John, who plays in a soccer team. The constituent to which the additive marker gives an alternative, is often referred to as associate. In more formal terms, the particle refers to an alternative to the associate of the additive. Therefore, additive particles are sometimes described as presupposition triggers (König, 2002).
The interpretation of (1) depends on the placement of the stress in the sentence. In both interpretations, ook gets the primary stress in the sentence. If the secondary stress (and focus) is on “soccer team”, the sentence gets the first interpretation and if the secondary stress (and focus) is on “John”, it receives the second interpretation. This is consistent with König (2002), who argues that the scope of an additive marker is determined by the focus and intonation structure of a sentence and that the additive marker takes scope over the stressed element in the proposition.
The basic function of the Dutch and German equivalent auch/ook resembles the English also, as shown by (2) and (3):
(2)
(3)
These Dutch and German utterances have the same two alternative interpretations as the English translational equivalent: Jan could be playing in another team besides the soccer team or there
Jan speel-t ook in een voetbalteam
Jan play-3SG.PRS also in a soccer team
‘John also plays in a soccer team’
Jan spiel-t auch in ein-er Fußballmannschaft
Jan play-3SG.PRS also in a-DAT soccer team
8 could be another person apart from Jan playing in the soccer team. Just like in English, the interpretation of the sentence depends on the intonation structure of the sentence.
If the additive ook is negated, a difference occurs between English on the one hand and Dutch and German on the other hand. English uses another form in negations, either or neither (Rullmann, 2003). Either and neither are Negative Polarity Items. They are licensed by a negation and by some other elements, like the adverbs few and hardly. German and Dutch, in contrast, simply add niet or nicht to the sentence in those cases:
(4) John does not play in a soccer team either. (5)
Cross-linguistically, the additive function has semantically widened. According to Forker (2016), an additive particle can express scalar additivity (comparable to the meaning of the English even), introduce a contrastive topic or a topic switch, express indefiniteness (like the English whenever) and introduce a concessive conditional (like the Dutch ook al). These functions are shown by the following examples:
(6) Dolakha Newar
(7) Udihe
(8) Amharic
(9) Dutch
Jan speel-t ook niet in een voetbalteam
Jan play-3SG.PRS also not in a soccer team
‘John does not play in a soccer team either’
bā̃d ̣ā=ku thi-gur=uŋ ma-da-i don-ju
pot=LOC {one-CL}=ADD NEG-exist-INF finish-3SG.PST
‘In the pot there was not even one (grain of rice) left.’ (Genetti, 2007: 244)
min-du sata bie s’ei=de anči
me-DAT sugar be.PRS.HAB salt=ADD no
‘I have sugar, but no salt.’ (Nikolaeva & Tolskaya, 2001: 822)
mən yə-ddɛrrɛg, mɛčče=mm yəhɛw wusɛǧi-w!
what 3SG.M-be.done.JUS when=ADD voilà take.IMP.2SG.F-3SG.M
‘What should be done, voilà, take it whenever you want!’ (Demeke & Meyer, 2008: 624)
Hij deed mee met schaken ook al hield hij niet van spelletjes
He participate-3SG.PST in chess also though like-3SG.PST he not of games
9 The examples give an indication of the kind of variation we encounter cross linguistically. Some of the examples show close resemblance with English and other Germanic languages in their use of additive markers (6 and 9), whilst others seem rather unnatural from an Indo-European point of view (7 and 8).
Besides the additive function and other closely related functions, additive markers also have modal functions in some languages like Dutch and German (Dittmann, 1980; Meijer, 2016; Thurmair, 1991; Vismans, 1994). In the following section, I discuss these modal functions.
3.2. Modal Functions
The Dutch ook and the German auch are highly polysemous. Apart from the additive uses, they also have so-called modal uses. A modal use is usually phonetically unstressed, whereas additives are stressed (Thurmair, 1989). In this section, I discuss some of these cases where ook and auch are used in contexts where English does not allow too, as well and also.
One of the modal uses of ook is in polar questions:
(10)
(11)
According to Dittmann (1980) and Vismans (1994) auch and ook express politeness in these examples , therefore we call this use a politeness use. If this utterance is used without ook, the message is more direct and therefore less polite. To soften the impact of the message, ook and auch are added to the proposition. The use of these markers is not allowed in all contexts. Just like in the case of the additive use, there must be an alternative to an element in the proposition. That means that using the questions in (10) and (11) is not correct when you aim to get a specific cheese from somebody (without referring to possible other cheeses he or she could have). To get the politeness reading, it is necessary that ook is unstressed. In this case zoutloze gets the stress in the sentence.
Ook and auch are also used in wh-questions (Thurmair, 1989). According to Thurmair, those questions are always rhetorical, like we can see in the following example:
Heef-t u ook zoutloze kaas?
have-3SG.PRS you also saltless cheese
‘Do you have saltless cheese? (Vismans, 1994: 66)
Haben Sie auch salzlose Käse?
have-3PL.PRS you also saltless cheese
10
(12)
In the context preceding to this sentence, a woman got an oil stain on her blouse. The speaker continues the discussion by the sentence in (12). With the use of this utterance, he/she indicates that it was expected that you would wear an apron in this situation. Also, he/she connects to the sequence where the woman got a stain on her blouse. Due to this connection with a previous element in the discourse, it is possible to add dan creating the Dutch translation Waarom trek je (dan) ook geen schort aan bij het koken! In the present paper, we call this use the rhetorical use.
Dutch and German also use their additive particles in WH-exclamations (Thurmair, 1989, 1991), as can be seen in the following examples:
(13)
(14)
In her article, Thurmair (1989) provides an analysis of this specific use of auch and ook. She claims that this use is very similar to the wh-questions, given that it also refers to previous elements in the discourse; in this case, a circumstance where the speakers showed that they were stupid. Other articles, like d’Avis & Lohnstein (2016) claim that the use of ook and other modal particles is quite common, but not obligatory in this type of sentences in German. As to the meaning, the effect seems comparable to the effect in the former two examples: It mitigates the content of the remark. The utterances without ook and auch appear to be more categorical than the version that does include these modal particles. Just like (10) and (11), these uses are politeness uses.
Auch and ook also occur in declarative sentences. In this type of sentence, the additive markers can fulfill different functions:
Warum sieh-st du auch zum Kochen keine Schürze an?
Why pull-2SG.PRS you also by cooking no apron on
‘Why don’t you pull on an apron while cooking?’ (Thurmair, 1989: 159)
Was sind wir auch für blöde Kerle!
What are-1PL.PRS we also for stupid fellows ‘What stupid fellows are we!’ (Thurmair, 1989: 159)
Wat zijn we ook voor stomme kerels!
What are-1PL.PRS we also for stupid fellows What stupid fellows are we!
11
(15)
(16)
By using ook in (15) and (16), the speaker implicitly agrees with what has been said before (Karagjosova, 2003; Meijer, 2016). In (15), speaker B explicitly says that it would be a contradiction to say that two and two is five and in (16) he agrees by giving a reason why Peter is looking so bad. According to Thurmair (1989), ook is used to establish a connection between the proposition and the discourse and to indicate that the information given by speaker A meets the expectations of speaker B. These uses are highly similar in German and Dutch. In (16), it is also possible, just as in (14), to add dan in the Dutch translations: (16) would translate as Hij is (dan) ook lang ziek geweest in Dutch. In both cases, the sentence with auch/ook explains a previous element in the discourse. The use in (15) is a so-called counterfactual use, whereas the use in (16) is an explanatory use.
The literature also refers to two more uses in Dutch: an contrastive use and an anti-mirative use:
(17)
A: Twee plus twee is niet vijf
A: Two plus two are-3SG.PRS not five
B: Dat zou ook een tegenstelling geweest zijn
B: That will.3SG.PST also a contradiction been are
‘A: Two plus two is not five. B: That would have been a contradiction.’ (Meijer, 2016: 22)
A: Peter sieh-t sehr slecht aus
A: Peter look-3SG.PRS very bad out
B: Er ist auch lange krank gewesen
B: He are-3SG.PRS also long ill been
‘A: Peter looks very bad B: He has been ill for a long time.’ (Meijer, 2016: 29)
Vader: Moe-t je geen huiswerk maken?
Father: Have.to-3SG.PRS you no homework make
Zoon: Dat doe ik ook
Son: That do-1SG.PRS I also
“Father: Weren’t you supposed to do your homework?
12
(18)
According to Schmitz, Hogeweg, & De Hoop (to appear), the father in (17) implies that the son is not doing his homework. The son does not agree with this statement, so he answers with “Dat doe ik ook”. According to Schmitz et al. (to appear), who followed Saebo (2004) ook is used to avoid a relation of contrast between the situation the father desires and the actual situation. This is referred to as an anti-contrastive use. In (18), it does not come as a surprise to the speakers that there are few gas stations in Iceland. Ook is used to indicate this lack of surprise. In linguistic terms, this is referred to as an anti-mirative marker.
3.3. Summary
In the previous part of this section, I described the different additive and modal uses of ook. In Table 1, the different uses are summarized. The table shows that ook and auch are used in all sentence types; declarative, interrogative and exclamative, and that the type of meaning expressed by the particle is partly depended upon the type of sentence. Behind every category, the corresponding examples in the previous sections are indicated. The English translational equivalent of ook is only used for the additive affirmative cases. For modal uses, English uses another particle or no particle at all.
In the next section, I will elaborate on how additive markers are acquired in different European languages. This allows us to formulate the research questions of this investigation. The following sections will be devoted to answering these research questions.
Table 1
Summary of the different uses of ook
Main categories Sentence type Type of modal meaning
Additive Negation (1-3)
Affirmation (4-5)
Modal Interrogatives (including subordinate questions) Rhetorical question (10-11) Politeness marker (12)
We zaten op een gegeven moment nog van ojee als we maar genoeg
We sit-1PL.PST at a given moment still of oh.dear if we but enough
benzine hebben want ja benzinepompen heb je ook niet
fuel have-1PL.PRS because yes gas station have-2SG.PRS you also not
overal op IJsland
everywhere in Iceland
‘At a given moment we were like: ‘oh dear, hopefully there is enough fuel left, because there are few gas stations in Iceland.’
13
Other
Exclamative Politeness marker (13-14)
Other Declarative Counterfactual (15) Explanatory (16) Anti-contrastive (17) Anti-mirative (18) Other
4. Previous Research on The L2-Acquisition of Additive Markers
Previous research on the L2-acquistion of additive markers has mainly focused on the additive function. Some of these studies found a clear relationship between the acquisition of additive markers (and resembling particles) on the one hand and syntactic development on the other. Benazzo (2002) describes the following stages of linguistic development and the related uses of additive particles in untutored second language learning:
- Pre-basic variety: mainly nominal constituents, related based on pragmatic and discourse organization structures. For example, “Mädchen auch Auto”
- Basic variety: sentences organized around non-finite verb and its arguments. There is no correct morphology yet. For example, “Die Mädchen willen auch Auto”
- Post-basic variety: correct morphology begins to appear, first in functional verbs, like modal verbs preceded by lexical verbs. For example, “Das Mädchen will auch ein Auto.”
In the rest of her study, Benazzo argues that the restrictive (for example the German nur) and additive particles are acquired in the Pre-Basic variety, iterative particles like noch in the basic variety and other temporal particles like schon in the post-basic variety. She relates these developments to the emergence of the finite verb. In other words, to express iterativity, it is necessary to have a verb in a sentence, and correct morphology is needed to express other temporal meanings like the meaning of schon.
Other authors, like Andorno and Turco (2015) also relate the development of the different meanings of additive particles to these developmental stages. They say that additive particles usually precede the element in its scope in the pre-basic variety. In the basic variety, it can also be put in the post-finite position. It is only in the post-basic stage that it can appear in other peripheral places. In these peripheral places, the additive particle can have a modal reading, for example an explanatory reading. However, as described by Becker and Dietrich
14 (1996), speakers still have difficulties in placing these modal particles in the correct peripherical position. So, the application of the additive particle is not yet completely functional.
Furthermore, there is also a difference between how additive particles are applied by child L1-learners and adult L2-learners. According to Benazzo et al. (2012) the L2-learners correctly apply the pragmatic functions of particles, like ook, in sometimes incorrect grammatical sentences, whereas children are more accurate with regard to their grammar, but tend to omit particles in places where they are pragmatically necessary or add them in places where they are not pragmatically appropriate.
5. Research Questions and Hypotheses
Previous research has shown that pragmatics, and particularly discourse particles, are difficult to acquire for second language learners. There are different explanations why discourse markers are so difficult: First, they are often polyfunctional and second, there are subtle differences between languages. The acquisition is affected by various speaker internal and external factors. One of these factors is L1-interference. This mechanism applies when two languages are very similar. In that case the acquisition could be facilitated, as some structures can be transferred from one language to another.
In the preceding sections, I described the different uses of the additive markers in English, Dutch and German. My examples showed that the three languages all apply these markers in an additive use. However, Dutch and German have some additional uses: modal uses. Those uses are not found in English. Accordingly, Dutch and German are very similar, but English deviates from these two languages. Thus, German learners could be helped by their mother tongue in the acquisition of modal uses of ook in Dutch.
In this paper, I would like to find out whether it is indeed easier for German learners to acquire the different uses of ook, especially the modal uses, than for English speakers, because they can profit from their native language. In other words, I would like to examine whether L1-interference is a determining factor in the acquisition of ook. This means that I would like to answer the following research questions:
- How is the particle ook used in Dutch? Is the type of input a predictor of the acquisition by L2-learners and if so what does this prediction look like?
- To which extent does the L1 affect the acquisition of the particle ook? Is it facilitating to have a language that resembles Dutch as a mother tongue?
15 For the first research question, I predict that both modal and additive uses are present in normal every day Dutch, even though the additive uses are likely more frequent. The frequency of the uses is potentially a predicting factor in how the L2-learners acquire them. If a use is very frequent, it is easier to acquire, because the learners get a lot of input and get a reasonable chance to form a connection between the form and function, as was argued by Ellis (2006). Furthermore, there are various modal uses: The more frequent ones could be easier to acquire than the less frequent ones.
My hypothesis for the second research question is that the L1 indeed affects the acquisition. If the results of my study meet my predictions, I expect that English, Dutch and German native speakers perform similarly in contexts where the three languages overlap (so in the additive uses). Second, I predict that English speakers behave differently from Dutch speakers in their judgment of the modal uses of ook, because the two languages are not similar on that point. And finally, I expect that the German speakers react like Dutch speakers in all uses, because the uses are all shared between the two languages.
In the following section, I introduce a method to answer the research questions. The research consists of two parts: a corpus study and an experiment. The corpus study is used to find answer to answer the first research question. It tries to find out how frequently native speakers of Dutch apply the different uses of ook and the kind of input the second language learners have. The experiment consists of a grammaticality judgment task and a sociolinguistic questionnaire. The grammaticality judgment task investigates the perception of the additive and modal uses of ook by English, German and Dutch speakers of Dutch and the sociolinguistic questionnaire their linguistic background.
6. Method 6.1. Method of the Corpus Study
With this corpus study, I try to find an answer to my first research question how the particle ook is used in Dutch. The frequencies I encounter in this corpus study are used for the election of the material for the grammaticality judgment task.
For the corpus study, I made use of the Spoken Dutch Corpus (CGN). This corpus contains oral language data from both Dutch and Flemish speakers in different language contexts, like monologues and dialogues, spontaneous and prepared discourse and public and private discourse. For the present investigation, I only made use of dialogues produced by speakers from the Netherlands to make the data more homogenous, I randomly chose 50 dialogues from the corpus: 25 face-to-face dialogues and 25 phone conversations. These
16 dialogue types were chosen, because they form a good reflection of the informal spoken language that is encountered by L2-speakers. Furthermore, the modal uses of ook mainly occur in spoken language.
There were around 1200 occurrences of ook in these dialogues. I analyzed every fourth example yielding a total number of 273 examples. The main goal of the analysis of these examples is to see how frequent the modal uses of ook are compared to the additive uses. The examples were therefore annotated as additive or modal. In the latter group, I categorized the different uses based on their meaning and sentence type. If I doubted about the interpretation of an example, I listened to the audio transcripts to listen where the stress was placed in the sentence. The categorization is further explained in the appendix with the annotation protocol.
6.2. Method of the Grammaticality Judgment Task 6.2.1. Participants
17 Dutch native speakers (Age: 31,17 (13,98), 14 females, 3 males), 20 German learners of Dutch (Age: 25,8 (4,76), 14 females and 6 males) and 9 English learners of Dutch (Age: 38,55 (19,3), 6 females, 3 males) were compared in this experiment. The participants were mostly students at higher education institutions or recent graduates, but especially the English group also contained people that already had more working experience. Apart from Dutch, English and German, some of the speakers had a good command of other language too. In the English group, there were participants who spoke Swedish, French, Malay, German or Bengali. Some of the German speakers spoke Italian, Norwegian, French, Swedish and Spanish. The Dutch speaker group had participants who had knowledge of Spanish, French, German, Arabic and Papiamentu. One of the Dutch speakers was raised as a bilingual in a Frisian speaking home.
The participants were recruited through different channels. I wrote messages on my personal Facebook page as well as on group pages that are visited by many international students. Besides, I wrote e-mails to Dutch language schools in the Netherlands and Dutch departments in Germany, Austria, the United Kingdom and the United States. That means that part of the learners learned Dutch in a native speaking environment, whereas others learned it as a foreign language. Before starting the test, the participants had to give their informed consent. The text in the informed consent was approved by the Ethical Committee of the University of Amsterdam. The participants were not payed for their participation.
After completing the grammaticality judgment task, the participants were also asked to fill in a short sociolinguistic questionnaire. These questions were taken and adapted from the LEAP-Q (Marian, Blumfield, & Kaushanskaya, 2007). The sociolinguistic part contained
17 general questions about age, gender and linguistic background for all speakers. There was also one set of questions about the amount in which the speakers used Dutch (calculating the exposure of the speakers to Dutch). These questions had a 5-point Likert scale ranging between Always and Never and one set of questions about the proficiency in speaking, writing, listening and speaking. In these questions there was also a Likert scale that ranged between Very good and Very poor. The exact questions are added to this paper in an appendix.
The results of the sociolinguistics questionnaire are shown below. Most of these data were only collected for the German and English speaker group. Therefore, there are some empty cells in Table 2:
Table 2
Means and standard deviations for the demographic and linguistic data per group
Item type Mean score on the Likert scale per group
Dutch English German F/t
Age (in years) 31.18(13.98) 38.50(19.31) 25.80(4.76) 3.390¹
Proficiency (on a scale from 1-5) 2.50(1.12) 2.14(0.59) 1.153
Exposure (on a scale from 1-5) 3.53(1.03) 3.54(0.55) -.0.27
Age of onset (in years) 23.56(13.9) 17.50(7.22) 1.556
Stay in a Dutch speaking country (in months) 137.33(159.21) 20.70(49.22) 2.152
¹The result was significant at p <0.05
The L2-learners had a relative good command of Dutch, because the test contained some items in relatively complicated Dutch. Overall, the proficiency level of the two L2-groups was comparable. According to the self-reported proficiency, the English speakers had a moderate to good knowledge of Dutch (corresponding to 2.50 on the Likert scale) and the German speakers a good knowledge (corresponding to 2.14 on the Likert scale). However, this difference was non-significant. Both speaker groups used Dutch half of the time (3.53 and 3.54 on the Likert scale).
On average, the English speakers started learning Dutch at the age of 23,5 and the German speakers at the age of 17,5. This difference was non-significant, because there was more variation in the English group than in the German group. Consequently, the standard deviation in the English group was considerably higher. 8 of the 9 English participants lived in the Netherlands for on average 13 years and 14 of the 20 German participants lived in the Netherlands for 3 years on average. The high difference in length of stay was also non-significant. It was caused by two outliers in the English group, whereas the rest of the English speakers had a similar length of stay in a Dutch speaking country as the German speakers. Apart
18 from the speakers who resided in a Dutch speaking country, there was also one German participant who lived close to the Dutch border.
There is only one significant difference between the groups, in the category of age, F (2, 43) =3.390, p < 0.05. The German speakers were significantly younger then the Dutch and English speakers. To make sure this age difference is not affecting the results, I will treat it as a possible covariate from now on in the analysis. Despite being similar in most measurements, an appealing difference from this table is that the standard deviation is usually higher for the English group than for the German group. This suggests more heterogeneity in the English group.
In Table 3, it is shown how the speakers performed on the filler items with present and past tenses:
Table 3
Means and standard deviations for the control items per group Item type
Mean score on the Likert scale per group
Dutch English German F ηp²
Present tense 1.39(0.49) 1.59(0.69) 1.35(0.45) 0.713 0.03
Past tense 4.63(0.49) 4.09(0.96) 4.66(0.44) 3.2631 0.13
1This difference was significant at p<0.05
The results of the control items show that the English and German and Dutch native speakers are equal on the items with a present tense, F (2,43) = 0.713. The effect size for this difference was also small, ηp² = 0.03. However, there is a difference between the groups on the judgment of items in which a past is correct, F (2,43) =3.263. The effect size of this difference was medium to large, ηp² = 0.13. This means that the German and English group are not similar in this respect. This might be related to the mother tongue of the speakers. The German use of the past tense is like the Dutch, whereas the English deviates. English has a continuous past tense that partially overlaps with the Dutch and German simple past in its functions. That might cause that English speakers do not recognize all the past tenses in Dutch.
6.2.2. Materials
The test for this investigation was a grammaticality judgment task. According to Blume & Lust (2017) a major advantage of this method is that it assesses a speaker’s intuition about the use of distinct linguistic items in very direct way: it gives the possibility to test the pragmatic knowledge of L2-speakers in an easy way.
However, there are two important downsides to this method: First, it can be difficult to determine where the judgment of the speaker is based on, given that many factors (syntax,
19 semantics, context) affect the way people judge a sentence. To reduce the effect of this kind of factors, the participants were presented with a detailed context in which the item was uttered so that the dialogues are more likely to be interpreted and understood in the same way by all the participants. Apart from that, the same sentence with ook was used in two different testing conditions (one correct, one incorrect) to make sure that the difference in judgment was attributable to ook and not to other factors. A second disadvantage of this method is that a simple “yes” or “no” might not be a good representation of what the speaker thinks. Therefore, I used a five-point Likert scale in the answers.
In the test, I focused on explanatory uses and politeness uses. The explanatory use proved to be relatively frequent in a small sample from the Corpus Gesproken Nederlands and occurs in both German and Dutch but not in English. In contrast, the politeness use, was infrequent. This enables us to see the difference between a frequent and infrequent modal use of ook. The results of this corpus study will be discussed in more detail in the result section.
The questionnaire included 72 items. There were 14 items about the additive use of ook (7 correct uses and 7 matched incorrect uses), 14 items of ook as politeness marker (7 correct and 7 incorrect) and 14 explanatory uses (7 correct and 7 incorrect) and 30 filler items about the use of the present/ past tense. This type of fillers was used because the use of the present/past tense are a good way to compare the speakers in their proficiency. In the matching of the items, I paid attention to the length and grammatical and lexical complexity of the entire dialogue. The sentences containing ook occur twice in the exact same form in the two matched items. By randomization, I made sure that two identical sentences with ook never followed each other in the experiment. This made it harder for participants to identify the matched items design.
The test items were presented as contextualized dialogues. The dialogues were introduced by a short fragment of around three sentences. This introduction gave information on the protagonists, their actions and their conversations. The introduction was followed by a short dialogue of at most 6 sentences. The dialogue contained a gap. At the end of the dialogue, the participant was asked to choose the best answer for this gap (one option with ook (A) and one option without ook (B)). The stress in the target sentences was indicated by a word (or several words), that was/were underlined and bold. The scale for the choice has five points (high preference for A, preference for A, no preference, preference for B, high preference for B). In half of the cases, the answers were reversed, so that answer A contained the option without ook and B the option with ook. Below are two examples of test items:
20
Test item 1
Joost en Sara zitten allebei bij de voetbalclub FC Doelgericht. Ze hebben het over het clubfeestje dat aanstaande donderdag gehouden gaat worden en wie daarbij aanwezig zullen zijn.
Joost: Komt Jan donderdag naar het feest? Sara: Ja, die komt!
Joost: En Merel?
Sara: TS (Target Sentence)
Welke van deze twee reacties past het beste in deze context? A: Die komt ook.
B: Die komt.
Grote voorkeur voor A Voorkeur voor A Geen voorkeur Voorkeur voor B Grote voorkeur voor B
Test item 2
Jan en Els hebben toetsweek. Ze willen samen met een paar klasgenoten gaan studeren in de bibliotheek. Jan vraag zich af wie er vandaag mee komen.
Jan: Komt Johanna mee leren voor de wiskundetoets? Els: Nee, die komt niet.
Jan: En Anton? Els: TS
Jan: Gezellig dan zijn we met zijn drietjes.
Welke van deze twee reacties past het beste in deze context? A: Die komt ook.
B: Die komt.
Grote voorkeur voor A Voorkeur voor A Geen voorkeur Voorkeur voor B Grote voorkeur voor B
In the two test items, I presented above, there are two matched additive items, the first correct and the second incorrect. As can be seen, the two items are comparable in various aspects. First, the two dialogues are about people that are (not) joining others in an event. Apart from that the two target sentences (TS) are the same. In the first case, the expected answer is A, whereas in the second case, most people will probably answer B.
6.2.3. Procedure and design
The test was presented to the participants in Qualtrics. The participants were able to fill in the test individually from their own laptop or computer at any given moment. In total, the test took about 40 to 50 minutes. There were 4 different randomized orders of the test that were randomly assigned to the participants by Qualtrics.
21 Before the actual test started, the participants had to give their informed consent. Consecutively, the participants read some instructions on what they had to do during the test. The instructions stated that the test was about Dutch language proficiency and the way speakers judge different sentences in an ongoing discourse. I explicitly said that the participants had to rely on their intuition and that they had to consider the context and the phonetic stress (as indicated by a bold and underlined word) in the rating of the sentences. As the test was about pragmatics, the participants were asked which answer they found most acceptable in a given context (and not to choose the “correct” answer in a given context). After completing this part of the survey, the participants did the sociolinguistic part of the questionnaire.
6.2.4. Statistics
As I stated before, I compared English and German learners of Dutch to Dutch native speakers on their rating of the different uses of ook. The ratings of the three groups were measured on three independent variables: additive uses of ook, explanatory uses of ook and politeness uses of ook (all have correct and incorrect versions). The rating the speakers gave on the 5-point Likert scale is the dependent variable in this research. In the scale 1 stands for the sentence without ook and 5 for the sentence with ook yielding the prediction that the sentences where ook is correct receive a higher rating than those contexts where it is incorrect. If the score was around 3, the speakers had no preference for one of the uses. There were also filler items with past and present tenses. In these items, there was also a Likert scale, with on one side a sentence with a present tense (1 on the Likert scale) and on the other hand a past tense (5 on the Likert scale)
The ratings on all the items were compared between groups. I performed individual ANOVA’s on all the item types in SPSS. Given that the age difference between the different groups was considerable, I used age as covariate in the ANOVA’s. As the group of participants was relatively small (especially the English group), I checked the sphericity and normality of the data. This proved to be unproblematic. Since I did multiple comparisons for the same group of participants, I applied a Bonferroni correction to the p-values. The effect size of the ANOVA’s was calculated by means of partial-eta squared. Aside the ANOVA’s, I also calculated the inter-item reliability for all categories.
A second important step in my statistical analysis, were two multiple regression analyses: one with all correct items and one with all incorrect items. For this, I first calculated the mean over all (in)correct items for a participant. I used these two scores as dependent variables in the multiple regressions. The independent variables I used were L1, age, exposure, proficiency, length of stay and age of onset. I checked whether the independent variables were
22 related to each other by means of the multicollinearity values. This did not prove to be the case. So, I did not exclude any of these independent variables from the analysis.
7. Results 7.1. Results of the Corpus Study
This section discusses the result of the corpus study that I presented before. In Table 4, it is shown that the modal uses form about 19% of the total:
Table 4
Uses of ook in the Corpus of Spoken Dutch
Additive/modal Characteristics of the modal uses
Sentence type Type of modal meaning
Additive 222 Modal 51 Exclamation 1 Politeness 1 Question 2 Indefinite 2 Declarative 48 Anti-contrastive 27 Explanatory 17 Indefinite 1
Concessive (“ook al”) 1
Other 2
Total 273 51 51
In the corpus, we mainly encounter modal uses of the anti-contrastive type. These uses resemble additives in some respects. In the first example, ook is used as an additive marker, whereas in the second it is anti-contrastive:
(19) Uh dat kasteel is echt de moeite waard om uh om te bezichtigen.
Uh that castle are-3SG.PRS really the effort worth to uh to to visit
Het is ook niet duur en ‘t is midden in de stad.
It are-3SG.PRS also not expensive and it are-3SG.PRS midst in the city
23
(20)
In (19), the speaker makes an enumeration why they should visit a castle. It is worth visiting, it does not cost a lot of money and it is in the city centre. In this sentence, ook adds another feature to the other positive features of the castle. In (20) the discussion involves the photos that should be printed for a holiday album. As in the case of (19), the speaker adds an element to the sentence that is in accordance with the rest of the sentences. The ook in the third sentence makes clear that it indeed is not necessary to print more photos, as the Atrium photos are not that attractive. In this case the speaker establishes that there is no contrast between the fact that no photos should be added and that the Atrium photos are not that spectacular.
Another frequent modal use is the explanatory use. In this case, ook occurs in a sentence, which gives an explanation to another sentence, as in the following example:
(21)
In the last sentence, speaker A explains why his grandmother can bring so much wool in her suitcase on a plane trip. Ook is used to indicate that this is an explanation alongside with natuurlijk. It is phonetically unstressed unlike the additive ook. This prosody makes an additive interpretation impossible.
A: We gaan niet meer [foto’s] scannen A: We go-PL not more photos scan
B: Jij je zin
B: You your volition
A: Ja ‘t Atrium stelt ook helemaal geen drol voor als je dat zo ziet he?
A: Yes the Atrium put-3SG-PRS also at.all no shit for if you that so see-3SG.PRS uh A: We are not going to scan anymore photos. B: What you want. A: Yes, the Atrium is not that special, if you see it there.
A: Da’s zo handig met wol zeg-t ze dan want je
A: That’s so handy with wool say-3SG.PRS she then since you
kun-t ‘t overal tussen stoppen
can-3SG.PRS it everywhere between put
B: Ja, alsof ‘t dan geen plek inneem-t B: Yes like it then no place take-3SG.PRS
A: En uh nou dat weegt natuurlijk ook niet veel dus dat kan
A: And uh then that weigh-3SG.PRS of.course also not much so that can-3SG.PRS
A: That’s so practical about wool, she says, you can put it everywhere B: Yes, like it does not take place then A: And well, it does not weigh much, so it is possible
24 The corpus indicates that about one fifth of the uses are modal. These uses are non-occurring in English, and therefore form a challenge for English speakers who are learning Dutch. However, they are prone to positive L1-transfer from German speakers, as these uses do occur in their mother tongue. In the grammaticality judgment, which I discuss next, this L1 transfer is studied.
7.2. Results of the Grammaticality Judgment Task
In this section, I discuss the results of the grammaticality judgment task. One of the central questions of this paper is whether having German as a mother tongue facilitates the acquisition of the Dutch ook compared to English speakers who learn Dutch. I start by assessing the reliability of the items. In the second part of this section, I will focus on the differences between the German, English and Dutch speakers in the different item types and on the effect of the age, age of onset and other sociolinguistic factors on these results.
7.2.1. Reliability of the items
In this section, I will go into detail about the reliability of the test items in the different testing categories. Table 5 shows that the reliability scores for the correct items are mostly higher than those for the incorrect items. I found the following scores for Cronbach’s alpha:
Table 5
Reliability scores per item type
Item type Number of items Cronbach’s alpha
Additive correct 7 0.665 Additive incorrect 7 0.736 Politeness correct 7 0.697 Politeness incorrect 7 0.385 Explanatory correct 7 0.666 Explanatory incorrect 7 0.552
Present correct (filler) 15 0.860
Past correct (filler) 15 0.907
If the Cronbach’s alpha is above 0.6, it is considered acceptable. This was the case for the correct additive items (a = 0.66), the incorrect additive items (a= 0.736), the correct politeness items (a= 0.697), the correct explanatory items (a= 0.666) and the filler items (a= 0.860 for the items with a present tense and a= 0.907 for the items with a past tense). Unfortunately, the Cronbach’s alpha in the matched incorrect items for the politeness uses (a= 0.385) and the explanatory uses (a= 0.552) was lower. It seems that some of the items are more likely to be judged as correct than others in these testing categories. Since the matched correct items score
25 high enough on the Cronbach’s alpha and the matched item design is important for the results of my study, I decided to use all the items in the statistical analysis of my results. However, as the Cronbach’s alpha was low in some cases, the results should be interpreted in that light.
7.2.2. Comparison between the groups on the items with ook
In this section, I formulate the results that are necessary to answer my second research question. The results of my questionnaire support my hypothesis that the L1 is a deciding factor in the acquisition of ook. The results show that the English speakers differ from the native speakers of Dutch in the modal uses of ook and that they resemble the German and Dutch speakers in the additive uses, which are shared between the three speaker groups. In Table 6, my results are summarized:
Table 6
Means and standard deviations per item types per group
Item type Mean score on the Likert scale per group Effect of the L1 Age Effect
Dutch English German F ηp² F ηp²
Additive correct 4.46(0.65) 4.19(0.84) 4.20(0.53) 0.921 0.042 0.394 0.009 Additive incorrect 1.40(0.63) 1.57(0.66) 1.49(0.56) 0.211 0.010 0.020 0.000 Politeness correct 3.56(0.64)² 2.62(0.78)² 3.05(0.82) 5.305¹ 0.202 0.801 0.019 Politeness incorrect 2.43(0.56) 2.11(0.69) 2.11(0.49) 1.774 0.078 0.246 0.006 Explanatory correct 3.80(0.59)² 2.94(0.83)² 3.60(0.72) 3.454¹ 0.141 3.126 0.069 Explanatory incorrect 1.99(0.57) 2.57(0.74) 2.27(0.66) 3.186 0.132 2.590 0.058
¹The difference was significant at p < 0.05
²There is a significant difference between the English and Dutch speakers in the post-hoc Bonferroni (p<0.05)
The table shows that there are some tendencies in the means of the item types. First, we can see that the speakers tend to accept a sentence with ook in a context where one could correctly use it, whereas they opt for a sentence without ook if the use is not allowed. In all item categories, this difference is significant over all speaker groups. However, the differences per category are different. In the additive items, the difference has the highest effect size, t (90) = 21.997, p < 0.05, d=4.60. In the explanatory and politeness category the difference is smaller, but significant. Nonetheless, the effect sizes are smaller, t (90) =6.344, p<0.05, d=1.33 for the politeness category and, t (90) =8.908, p<0.05, d=1.84 for the explanatory category. The effect of age on the results is non-significant for the additive and politeness items. In the explanatory items, the effect comes close to significant, but does not yet reach a p< 0.05 level. In summary, the correct-incorrect division we made in the task turned out to be identifiable for the participants and age did not have a significant effect in neither of the categories.
26 The difference between the groups on the additive condition was non-significant, both for the correct as well as the incorrect items. The difference between correct and incorrect uses is the most pronounced for Dutch speakers. On a five-point Likert scale they scored 4.46 for correct items and 1.40 for incorrect items. The difference for the English and German speakers was a bit smaller. They scored the correct items as 4.19 and 4.20 respectively and the incorrect items as 1.57 and 1.49. The high difference between correct and incorrect items shows that the speakers are not confused about the contexts. They clearly recognize when ook is used (in-)correctly.
In the politeness condition, the speakers were significantly different in the correct uses, F (2,43) =5.305, p<0.05, but not on the incorrect uses, F (2, 43) =1.774, p >0.05. A post-hoc Bonferroni test showed that there was a significant difference between the English and Dutch speakers in the correct condition. It is apparent from the table that the differences between correct and incorrect items are a considerably smaller for this category. For the English speakers, there is almost no difference at all (2.62 and 2.11 at the Likert scale) and the difference for the German and Dutch speakers is rather small (around 1 point on the Likert-scale). In other words, the English speakers are confused about the correctness of ook and prefer a sentence without ook even in contexts where it could potentially be used (the 2.62 on the Likert scale corresponds to “small preference for a sentence without ook”). The German speakers don’t prefer a sentence with(out) ook and the Dutch speakers have a small preference for a sentence with ook. The small differences between correct and incorrect sentences show that this type of ook is harder to understand for the speakers. This use is the most complex for the English speakers, given that they are least able to distinguish correct and incorrect uses.
In the explanatory uses, there is also a significant difference between the groups in the correct items, but not in the incorrect items. Just like in the case of the politeness items, the post hoc Bonferroni showed that the English speakers were significantly different from the Dutch speakers. Compared to the correct politeness condition, the effect size of the correct explanatory condition was smaller. This shows that the groups are closer to each other than in the politeness condition. In comparison to the politeness condition, the difference between correct and incorrect items is also more pronounced for the German and Dutch speakers (around 1.5 point on the Likert-scale). The English speakers barely show a difference between correct and incorrect items (2.94 and 2.57 at the Likert scale). This shows that the understanding of the explanatory use of ook is problematic for the English speakers.
After running the ANOVA’s, I also performed two multiple regressions (one with all the correct and one with all the incorrect items) to see which sociolinguistic values contributed
27 most to the results on the grammaticality judgment task. Table 7 with the results on this regression analyses is shown below.
The table shows that all factors together do predict the outcomes on the correct items, F (6) =4.214, whereas they do not predict the results on the incorrect items. In the regression analysis on the correct items, the only significant predictor for the outcome, is the L1. The factor proficiency came close to significance (p=0.052). If we put this in words, the more proficient one is, the more likely one chooses ook in a context where that is correct. And second, if one is German, one is more likely to choose ook in a correct context then if one is English. Surprisingly, neither of the other factors contributed significantly to the results on the correct items of the grammaticality judgment task. For the incorrect items, neither of the sociolinguistic factors formed a significant predictor for the outcome.
Table 7
Results of Multiple Regression Analysis by Sociolinguistic Factors
t β F df Adj. R2
Correct items
Overall model 4.2141 6 0.408
L1 2.3961 0.452
Stay in a Dutch speaking country 1.212 0.283
Age of onset -0.164 -0.050 Exposure to Dutch -1.354 -0.215 Proficiency -2.0582 -0.423 Age 0.438 0.115 Incorrect items Overall model 1.430 6 0.084 L1 -0.622 -0.146
Stay in a Dutch speaking country 0.900 0.261
Age of onset -0.132 -0.05
Exposure to Dutch 0.694 0.137
Proficiency 1.506 0.385
Age -1.201 -0.394
1This result was significant at p<0.05 2This result was marginally significant p=0.052
8. Discussion
In this paper, I tried to find an answer to two research questions. My first question was: “How is the particle ook used in Dutch? Is the type of input a predictor of the acquisition by L2-learners and if so what does this prediction look like?”. To answer this research question, I
28 described the use of the Dutch ook in the corpus of spoken Dutch. In the corpus study, I found that about a fifth of the uses of ook in Dutch is modal, while the rest is additive. The modal uses mostly occur in declarative sentences and are often used to avoid a contrast and/or to explain a previous element in the discourse. Besides those uses, there are also other modal uses like a politeness use, but these uses are less frequent.
Based on the outcome of the corpus study, I included three different uses of ook in the grammaticality judgment task: the additive use (very frequent), explanatory (relatively frequent) and the politeness use (infrequent). My prediction was that a more frequent use would be acquired easier by all L2-speakers than a less frequent use. I verified my prediction by comparing the judgment on the three uses I mentioned. My results showed that there was a difference between the groups on the correct versions of the explanatory uses and politeness uses and not on the additive uses. It was apparent that the effect size of the less frequent use (politeness) was bigger. This suggests that the differences are more pronounced in this condition. In the incorrect versions, the differences were especially high in the explanatory incorrect uses. In these uses, the English speakers leaned towards the middle of the Likert scale and consequently did not prefer a sentence with(out) ook, whereas the German and Dutch speakers clearly preferred a sentence without ook.
The results of my study are in line with previous findings, like the findings of Ellis (2002, 2006). According to the work of Ellis, the more frequent a form is, the easier it is to establish a connection between the form and function. My study points to the following order of acquisition: additive → explanatory → politeness. Based on frequency, this is the expected order, given that the additive uses are most recurring, and the politeness uses are least recurring. Aside the frequency, the additive ook is also more salient in a sentence: It is stressed, and it usually occurs in a non-peripherical position. This saliency is also an argument why the additive use is acquired before the modal uses.
The second research question was: “To which extent does the L1 affect the acquisition of the particle ook? Is it facilitating to have a language that resembles Dutch as a mother tongue?” In my theoretical framework I made clear that the modal uses of ook were shared between German and Dutch, but not between Dutch and English. This led to the prediction that English speakers would have more difficulties in the understanding of the modal uses of ook than the German speakers of Dutch. To confirm that prediction, I compared 9 English speakers and 20 German speakers to 17 Dutch native speakers in a grammaticality judgment task.
The results show that the English speakers do not react as Dutch speakers on the modal uses, whereas the German speakers mostly do. As I mentioned in the previous paragraphs, my
