Excerpt
Table of Content
List of Figures
List of Tables
List of Abbreviations
Abstract
1. Introduction
2. Theoretical Background
2.1. The Common European Framework of Reference for Languages
2.2. Theories about the Origin of Cross-Linguistic Differences
2.2.1. The Generative Approach: Universal Grammar and the Principles-and-Parameters Theory
2.2.2. The Cognitive Approach: The Competition Model
2.3. Characteristics of Verb Placement and Subject-Verb-Agreement in German and English
2.3.1. Subject-Verb-Agreement in English and German
2.3.2. Verb Placement in English Main and Subordinate Clauses
2.3.3. Verb Placement in German Main and Subordinate Clauses
2.3.4. Underlying Syntactic Structures
3. Hypotheses and Variables
3.1. Variables
3.2. Hypotheses
3.2.1. In the Setting of the LiSe-DaZ test
3.2.2. In the Setting of Spontaneous Speech Production
4. Design of the Study
4.1. Participants
4.2. Materials and Methodology
5. Results: Disclosing the L2 Competence by Analysing German and English Morphosyntactic Features
5.1. Realisation of Subject-Verb-Agreement
5.2. Realisation of Verb Placement in German Main and Subordinate Clauses
6. Application to Second Language Assessment Testing
7. Conclusion and Future Research
8. Bibliography
List of Figures
Figure 1. Possible parameter with regards to headedness in German and English
Figure 2. Head-final VP structure in German
Figure 3. Head-initial VP structure in English
Figure 4. Language by Word Order Interaction in English, German and Italian
Figure 5. Syntax tree of an English main clause 'I also see the dog.'
Figure 6. Syntax tree of a German main clause 'Ich sehe auch den Hund'
Figure 7. Syntax tree of a German subordinate clause 'dass die Kinder halten.'
Figure 8.Example of a picture in the picture story 'Abenteuer im Park'
Figure 9. Ratio of produced utterances showing SVA in the LiSe-DaZ test setting sorted by German language proficiency levels according to the CEFR
Figure 10. Ratio of produced utterances showing SVA in spontaneous speech sorted by German language proficiency levels according to the CEFR
Figure 11. Mean ratio of produced utterances showing SVA in the LiSe-DaZ test setting sorted by German language proficiency levels according to the CEFR
Figure 12. Mean ratio of produced utterances showing SVA in spontaneous speech sorted by German language proficiency levels according to the CEFR
List of Tables
Table 1. Common Reference Levels: global scale
Table 2. The topological fields of the German main and subordinate clause
Table 3. Comparison of the ratio of targetlike structures in the LiSe-DaZ test and in spontaneous speech
Table 4. Error analysis in the area of SVA in the LiSe-DaZ test setting sorted by German language proficiency levels according to the CEFR
Table 5. Error analysis in the area of SVA in spontaneous speech sorted by German language proficiency levels according to the CEFR
Table 6. Comparison of the acquired stages of syntactic development
List of Abbreviations
illustration not visible in this excerpt
Abstract
In this thesis, the accuracy of the Common European Framework of Reference for Languages (CEFR) is explored with the help of a study investigating the German L2 morphosyntactic competence of 15 English L1 speakers in two settings: a language assessment test setting as well as in online processing. Within this thesis L2 morphosyntactic competence was limited to the two investigated areas of subject-verb- agreement (SVA) as well as the distinctive verb placement in German main and subordinate clauses. Previous research has identified these two linguistic areas as being especially prone to error in adult L2 acquisition of German. Since all test takers varied in their German language proficiency levels (encompassing a range from A2 to B2 according to the CEFR) it was hypothesised that L2 learners of a B2 language proficiency level generally outperform L2 learners of a lower language proficiency level (B1 and A2) in all investigated areas. However, this hypothesis could not be affirmed since the study did not find significant differences in the morphosyntactic competence of English German L2 learners between test takers of distinctive CEF language proficiency levels.
KEYWORDS: adult second language acquisition; syntax-morphology interface; verb placement; subject-verb-agreement; acquisition of German; areas of difficulty in German L2A; English L1 and German L2; language assessment testing
1. Introduction
With more bi- and multilingual than monolingual speakers in the world, multilingualism is no longer the exception but rather a common phenomenon permeating all age groups as well as social classes. The puzzling questions arise how speakers of different first languages (L1) successfully manage to acquire a second or third language (L2) being of such seemingly distinctively different underlying structure?
Within the context of this thesis a "second language refers to a language that is acquired after the first language has been established in early childhood" (Benati & Angelovska, 2016, p.3). Thereby, as Tracy (2014) points out it is to note that a successful L2 acquisition does not imply a perfect ultimate attainment of the L2 but entails rather a broad array of degrees of competence (p.17). Those differences in degrees of L2 competence are most commonly classified according to the Common European Framework of Reference for Languages (CEFR).
This thesis will primarily investigate the interaction of English as a L1 and German as a L2 in two morphosyntactic contexts. Their shared origin from the Indo-European languages (König & Gast, 2012, p.188) qualifies English and German as an interesting enough language pair to establish comparability. Not only do they share a sufficient number of features but they are distinctive enough to reveal significant contrasts. A first comparative glance at English and German structures may convey the impression of formal parallels, however, a thorough second look discloses very different underlying internal structures. This highly distinctive skeleton promotes numerous potential areas of interference and accounts for challenges, that especially adult English L1 speakers are confronted with when learning German.
By initially exploring the theory behind the origin of cross-linguistic differences primarily referring to the Generative Grammar approach as well as the cognitive approach of the Competition Model, the different underlying syntactic structures of English and German are thoroughly described and contrasted. The outlined contrasts serve as the basis for generating subsequent hypotheses about specific linguistic areas in German which are known to be problematic and challenging for foreign speakers acquiring German as a L2. Subsequently, the methodology as well as results of a study with 15 English L1 speakers who are German L2 learners of varying degrees of proficiency are empirically analysed and discussed. The study constituted of an individual test called LiSe-DaZ test with 15 English L1 and German L2 learners. It investigated the test takers' German L2 competencies primarily in the following two specific linguistic areas of German which are known to be especially prone to difficulties or even errors for English L1 speakers: verb placement in German main and subordinate clauses as well as subject-verb-agreement. Ultimately, the performances from the LiSe- DaZ test are contrasted to those of an equivalent online processing setting.
Afterwards this thesis draws conclusions about the accuracy and reliability of language assessment tests exemplified at the Common European Framework of Reference for Languages (CEFR) and concludes that the classification of language proficiency in the CEFR is to be treated with caution.
2. Theoretical Background
2.1. The Common European Framework of Reference for Languages
Universities, countries, schools and international institutions all make use of distinctive methodologies to describe and assess language proficiency. Comparability is thereby challenging to maintain. What may be an advanced language level in one country, may be only an intermediate one in another. Besides, the language proficiency description intermediate or advanced remains quite vague and is hard to grasp: what exactly are speakers with an intermediate language proficiency level able to produce? In what specific competencies do they differ from advanced speakers? This shall be explored in the thesis at hand.
The demand for a mutual basis of internationally recognised language qualifications and a uniform framework for language assessment was satisfied by the development and implementation of the Common European Framework of Reference for Languages (CEFR) in 2001. Its original aim was to transform Europe's rich linguistic and cultural diversity from a "barrier to communication into a source of mutual enrichment and understanding" (Council of Europe, 2001, p.2) by facilitating mobility, cooperation and unity within the European Union. By now, it is primarily known for being the dominant reference point in language assessment testing (Tracy, 2017, p.1).
The CEFR outlines an ascending series of six distinctive language proficiency levels (A1, A2, B1, B2, C1, C2) differentiating between basic users, intermediate users and advanced users.
- Basic user: A1 and A2 language proficiency level
- Independent (intermediate) user: B1 and B2 language proficiency level
- Proficient (advanced) user: C1 and C2 language proficiency level
According to each level, the CEFR describes in depth what competencies an L2 learner should be equipped with referring to effective communication skills and knowledge competencies as well as spontaneous situated speech production competencies. Although not explicitly stated, it is implied that a good command of grammatical rules is inevitable in order to communicate effectively.
The CEFR intends to assign each L2 learner to a language proficiency level that mirrors his L2 competence most accurately. The different levels build up upon another and are successively and ascendingly organised with each level indicating a different degree of L2 competence. Although the language levels cannot be considered clearly cut-off from another but rather transitional, each level is expected to display a reasonably consistent distinction (Council of Europe, 2001, p.21). Thus, with a growing L2 competence a L2 learner is expected to gradually move along from the lowest category being A1 to the highest being C2 while having passed A2, B1, B2 and C1. The language assessment criteria of each language level are expressed in "can-do"- statements and pinpoint what a L2 leaner is expected to express in the L2 (Council of Europe, 2001, p.16). While the A2 level learner is expected to "handle very short social exchanges (...) but cannot usually understand enough to keep the conversation going" and the B1 level learner "can deal with most situations likely to arise whilst travelling (...) and can enter unprepared into conversations", the B2 level learner "can interact with a degree of fluency and spontaneity that makes regular interaction with native speakers quite possible" (Council of Europe, 2001, pp.26-27). A complete description of all six language levels is taken from Council of Europe (2001, p.24) and shown in Table 1. Of course, the climb from a B2 to a C1 level is harder than transitioning from a A2 level to a B1 level. Underlying this is the premise that the more advanced a learner is, the bigger the range of language knowledge and competencies that he needs to acquire in order to be allocated to the next higher language level.
For the purpose of this study, participants with a German A2, B1 and B2 proficiency level were recruited. Those levels were considered to be advanced enough to enable the test takers to not only follow the test successfully but also to be capable of providing enough relevant data. Additionally, the range of anticipated distinctive L2 competencies was considered broad enough to yield in the expected significant differences.
Table 1. Common Reference Levels: global scale
illustration not visible in this excerpt
2.2. Theories about the Origin of Cross-Linguistic Differences
The following chapter attempts to present a multi-faceted introduction into the theories of origins for cross-linguistic differences. There are certainly a number of different relevant theories, however, due to a word restriction this thesis is going to limit the theoretic discussion to two approaches. Starting off with the Universal Grammar (UG) approach initially proposed by Noam Chomsky, this chapter also deals with a languageprocessing based approach, the Competition Model (CM). The presented approaches primarily differ in their assumption regarding how much of language acquisition can be accredited to an innate faculty, how language is processed and that the CM is based on language performance whereas the UG approach is based on language competence.
2.2.1. The Generative Approach: Universal Grammar and the Principles- and-Parameters Theory
Within the Universal Grammar (UG) framework shaped by Noam Chomsky language is described as a pure system of constraints surfacing in the form of principles and parameters (Chomsky, 1993). Claiming that all languages are subject to universal principles and hence are not supposed to differ, is in conflict with the existence of hundreds and thousands of language systems worldwide. Consequently, cross-linguistic variation must be explained by something else inherent in the UG, which stimulates speakers to specific language properties.
In the 1980s Chomsky postulated a theory, the so-called Principles-and- Parameters theory (P&P theory), which did not only clear the mystery around cross- linguistic variation but also provided an explanation for the emergence of cross-linguistic differences. From the perspective of the P&P theory, invariant principles as well as binary parameters constitute but also restrict all natural languages.
Presuming that human beings are equipped with an innate UG and hence universal principles by birth, the existence of cross-linguistic differences suggests that the binary parameters are language-specifically developed. Before the L1 acquisition has been completed, it is assumed that all parameters are set on neutral. Parameters interact and are set in response to the experience with the language and the input (Mitchell, Myles, & Marsden, 2013, p.81) waiting for the input to firmly trigger and chose a parameter. Having eventually filled all empty parameters certifies the learner a successful acquisition of the language. Thus, from the perspective of the Principles-and- Parameters theory, language acquisition is described as "a process of the successive setting of parameters" (Bates & MacWhinney, 1987, p.158).
While a principle, for instance, posits that all languages have the same basic sentence structure, the parameters may address headedness and distinctive feature strengths which differ across languages. A possible parameter is illustrated in Figure 1.
illustration not visible in this excerpt
Figure 1. Possible parameter with regards to headedness in German and English.
The different VP structures are depicted in Figure 2 and Figure 3 and later become relevant again when the cross-linguistic differences in German and English sentence structures are discussed; especially in the context of subordinate clauses, in which English follows a strict subject-verb-object (SVO) word order but German follows a subject-object-verb (SOV) word order.
It can thus be argued that parameters are responsible for cross-linguistic variation and "that the ability to acquire a given language may thus be construed as the ability to set the parameters of UG" (Roberts, 2007, p.22). From the perspective of structurally similar languages, it may be interesting to note that two languages with the same parameters should exhibit no structural differences in this particular area of interest and are expected to be acquired more easily due to positive interference (Saville-Troike & Barto, 2017, p.35).
Usually every parameter carries covarying features which disclose further structural characteristics that come along with a set parameter and "(are) acquired 'for free' once another is acquired" (Roberts, 2007, pp.23-24). For instance, as demonstrated in Figure 3 in languages like English with a VO structure, verbs and their direct objects merge to the right to form a verbal phrase (VP). The covarying property suggests that auxiliaries and verbs also merge in the same direction forming an auxiliary-verb phrase (AuxV), whereas in languages which display OV structures, the direction to merge auxiliaries and verbs is also in reverse with the verb preceding the auxiliary forming a verb-auxiliary phrase (VAux). The differences are exemplified in (1).
(1) a. I will soon have to go home. (AuxV)
b. Ich werde bald nach Hause gehen müssen. (VAux)
Applying the P&P theory to the theme of this thesis comes to the conclusion that areas of difficulties for non-native speaker can be expected. The respects in which parameters of languages differ are at the same time assumed to be challenging for nonnative speakers to acquire and constitute areas of potential errors. As illustrated before, in the respect of German and English, the property of VPs to merge either head-initial or head-final is such a case for instance.
Employing the P&P theory to L2 acquisition raises more essential questions: Do the invariant principles as well as parameters remain present and accessible throughout the life? Can parameters be reset or do they wither after the L1 acquisition is completed? Do L2 learners still have full access to their UG in L2 acquisition and what happens to those modules of the UG that are not required in L1 acquisition? Linguists today have not agreed on consent since empirical evidence remains ambiguous. They suggest a variety of models which differ in the extent to which the initial state of L2 equals the final state of L1 (extent of transfer) as well as in the extent to which the UG wields its constraints on L2 acquisition (extent of availability). A number of linguists take the stance that after the completed L1 acquisition the full inventory of options remains accessible including those which have not even manifested in the L1. Others argue that in L2 acquisition, L2 learners have only those settings on hand which were established in L1 while all the other options vanish after the L1 acquisition is completed. Since the theme of the thesis at hand does not deal with finding empiric evidence for one of the particular models but scrutinizes the contrastive morphosyntactic structures of German and English a further discussion would go beyond the scope.
2.2.2. The Cognitive Approach: The Competition Model
Whereas the UG approach characterises language acquisition based on innate but language-specific mechanisms, the Competition Model (CM), first developed by Bates and MacWhinney, attempts to view language competence as a result of the permanent interaction of input with general learning mechanisms. It argues for language acquisition as a process consisting of a sequence of interacting cognitive processes (Bates & MacWhinney, 1987).
According to the CM, permanently analysing the competing linguistic cues within a sentence, which appear for instance in the form of morphological inflections, syntactic structures and grammatical relations, enables human beings to eventually draw situated conclusions about the meaning of a sentence. Each word literally contributes to the overall understanding of the sentence. When contrasted to the previously outlined UG approach, it is important to note that in the Competition Model speakers are not at all subject to the innate systems of the UG but rely solely on their contextual and general knowledge of the world alongside their cognitive abilities. While each word of the sentence is closely examined and the competing linguistic cues of the sentence are scrutinized in detail, a set of possible interpretations is compiled each carrying an individual probabilistic value. Eventually the interpretation generated with the highest cue reliability suppresses all other possible interpretations and becomes predominant (Bates & MacWhinney, 1987, p.164). According to Bates and MacWhinney (1987) languages exert distinctively strong cues which are the sources of cross-linguistic variation (p.164). Implicitly formed by the language input, cue strengths profoundly impact language processing (Bates & MacWhinney, 1987, p.163). The essence of the CM can best be exemplified by contrasting how the subject is identified in English and German (Figure 4).
Figure 4 taken from Bates and MacWhinney (1987) clearly illustrates the differences: 95% of the time English speakers choose based on word order the first noun to be the subject of the sentence, whereas with 60% the German speakers are less certain about the identification of the subject solely based on word order. In their case, the identification of the subject seems to be less predictable than in English.
The question that emerges is how this difference comes about. The reason lies in the distinctive cue strengths that English and German accommodate in respect to the identification of the subject. Figure 4 provides clear evidence that for English speakers, word order exhibits the strongest linguistic cue to identify the subject of a sentence. This also explains why English speakers will most likely always choose the first noun to be the subject of a sentence. In contrast, the German equivalent looks less clear and more complex: first, for this purpose, it is crucial to know that German word order is grammatically more flexible than its English counterpart. This makes the identification of the subject solely based on word order in German less certain and harder to predict. Figure 4 indicates that Germans assign the same probability to the sentence-initial noun being the subject as to the sentence-final noun being the subject. Hence, German speakers cannot solely trust word order to infer correctly about the subject of a sentence but also use their elaborate morphological system with case-marking as well as subject-verb-agreement to affirm their assumption (Bates & MacWhinney, 1987, p.165). Hence in German, morphological inflections are the cues that ultimately provide clear indications about the semantic relations within a sentence. The following sample sentences in (2) shall further illustrate this: since there is no sufficiently clear casemarking in English for the subject (2a + 2b), case marking is by no means a sufficient cue for English speakers and totally irrelevant in the identification of the sentence, whereas examples (2c) and (2d) showcase that in German, word order is not a sufficiently reliable cue for the identification of the subject.
(2) a. The womans eats an apple.
b. The apples eats a woman.
c. Die Fraus aß einen Apfel.
d. Einen Apfel aß die FrauS.
For learners of any L2, the challenge constitutes exploring which cues are valid and most reliable in the L2 language in order to successfully interpret sentences in the target language. For English L1 speakers to learn that word order is not a trustworthy indicator for the identification of the subject in German is a gradual process. "Two competing tendencies may coexist for prolonged periods of time, cycling in and out" (Bates & MacWhinney, 1987, p.158). Data from the study produced by test takers confirms this gradual transition empirically.
2.3. Characteristics of Verb Placement and Subject-Verb-Agreement in German and English
Previous studies on adult L2 acquisition have hypothesised that some linguistic levels may converge easily to an ultimate targetlike representation while others may be more prone to difficulties and errors. Similar structures in the L1 as well as in the L2 which are alike in form, meaning and distribution are expected to be acquired first due to a positive transfer (Saville-Troike & Barto, 2017, p.35). Any other structure in the L2 which does not have an equivalent counterpart in L1 needs to be learned such as the gender system in German for English L1 speakers. Thereby, the greater the difference between the patterns of the two structures in terms of form, meaning and distribution, the greater is also the degree of perceived difficulty of acquisition which again eventually logically leads to a higher probability of errors (Ellis, 1994, p.308). However, it is to note that an area of difficulty cannot automatically be equalled to the occurrence of an error. Learners may perceive an area as extremely difficult but may still be able to produce the correct structure since they focus a lot of extra attention on this area. Structures perceived to be most difficult are those which partially overlap but differ in meaning, form and distribution (Saville-Troike & Barto, 2017, p.35) such as the two linguistic areas discussed in the thesis at hand.
This section provides the basis for the subsequent generation of hypotheses about linguistic areas being prone to difficulties for English German L2 learners by disclosing the underlying structural cross-linguistic differences of German and English syntax.
From a cross-linguistic perspective, specifically challenging to acquire for German L2 learners is the strongly overlapping interface of morphology and syntax in German. The phenomenon that many syntactic and semantic relations are predominantly marked on a morphological level, for instance with the help of an elaborate case system and a rich inflectional morphology for nouns, verbs, adjectives and pronouns, is unknown to many other language - also to English. Example 3 is taken from the study and demonstrates that even advanced English German L2 learners often consider this interface as challenging with many possibilities to produce errors.
(3) Test taker: *Die Baum hastsiNG;2.PERs gefallen mit die Sägen.
*The tree HûSsing;2.pers fallen with thesNGNOM chainsaws.
The tree was cut with a chainsaw.
Der Baum wurde mit einer Säge geschnitten.
(B2i_USA_L_8.1)
Earlier research (Hopp, 2010, p.902) has concluded that during the acquisition process adult German L2 learners frequently encounter two basic challenges: on a syntactic level, German L2 learners have to discover the German-specific verb placement in sentences, while on a morphological level, they are confronted with questions such as when and how to correctly mark the verb for finiteness. A typical difficulty for German L2 learners comprises the targetlike production of syntax structures such as the construction of main clauses with the finite verb surfacing in the V2 position and the construction of grammatically correct subordinate clauses by placing the finite verb in the VE position (Repp, 2014, p.110). in German, the positions of the finite and the non-finite verbs take over the "basic anchoring points of constituent order" (König & Gast, 2012, p.196). investigating the position of finite verbs in German sentences hence perfectly qualifies to indicate the learner's state of German.
Therefore, the following cross-sectional study systematically scrutinises the syntactic position of the finite verb in German main and subordinate clauses as well as the appropriate realisation of subject-verb-agreement in German. At first, the fundamental characteristics of the two investigated areas are outlined and subsequently thoroughly contrasted to their English counterparts.
[...]
Comments