Der Erwerb von es-Konstruktionen durch spanischsprachige Deutschlernende

Annette Fahrner

Résumé This dissertation is located in the field of research on Second Language Acquisition (= SLA) and follows a contrastive point of view. The basic hypothesis is that Spanishspeaking learners of German as a foreign language have serious difficulty in using the German pronoun es (i.e. English it) correctly due to differences in the language systems.
Firstly, Spanish is a pro-drop language, which can be defined by Bußmann (2008: 553): “[Im] finiten Satz [kann] eine durch das verallgemeinerte Projektionsprinzip erzwungene leere Subjektposition vorkommen […], die pronominale, d.h. referentielle Eigenschaften hat“ [English summary: a finite clause may have an empty subject position that has referential properties]. In contrast to Spanish, German is not a pro-drop language: generally speaking, sentences cannot exist without a subject. Secondly, the Spanish language system only has two genders, masculine nouns and feminine nouns, whereas German additionally has a third gender, neuter. These language properties cause selective attention
(cf. Schmidt 2010: 810) in Spanish-speaking learners, which means that they pay less attention to subject pronouns in a foreign language (cf. Ellis 2007). To sum up, the German pronoun es is a new and challenging phenomenon to which Spanishspeaking learners of German have to gradually adapt. Thus, Spanish-speaking learners
frequently make mistakes like *regnet instead of es regnet (Engl.: *is raining instead of it is raining), showing interference problems (= negative transfer from L1 into L2). However, it must be mentioned that a contrastive point of view has been viewed differently in the history of SLA research. In the 1950s, the Contrastive Analysis Hypothesis postulated that only differences in language systems account for learners’ difficulties (cf. Hufeisen/Riemer 2010; Ortega 2009). However, modern branches have come to a more differentiated view: the learners’ mother tongue is just one of many factors that influence the learners’ acquisition process (cf. Bünnagel 1993; Tekin 2012).
In addition to the contrastive point of view, the usage-based paradigm of Construction Grammar is a key element in my study. Construction Grammar in general assumes that there is no clear distinction between the lexicon on the one hand and grammatical
rules on the other; instead, the whole language system consists of form–meaning pairs (cf. Goldberg 1995, 2006). These basic assumptions are very suitable for modelling language acquisition processes (cf. Behrens 2009). The majority of studies that have already been conducted in this field focus on First Language Acquisition (e.g. Tomasello 2003), whereas there is only very little research that applies the theoretical assumptions of Construction Grammar to Second Language Acquisition. However, some studies indicate that this approach is very fruitful and there is a need for more empirical studies (cf. Ellis 2013).
The usage-based paradigm of Construction Grammar holds that language is dynamic and flexible, and that language learning and usage are characterized by general cognitive processes (cf. Bybee 2010). Every usage event shapes the speaker’s mental representation and his or her knowledge of a language. Thus, language system and language usage cannot be separated from each other (cf. Beckner et al. 2009). According to this view, the frequency of a linguistic pattern is important to its usage and acquisition, and
many studies have already shown that frequency plays a central role in language acquisition, usage and change (cf. Bybee 2007; Diessel 2007; Hilpert 2014; McDonough/Kim 2009; Stefanowitsch 2009). However, there is still a lack of empirical studies that analyse
frequency effects in SLA in close detail (cf. Ellis 2013). My study aims to help to reduce this desideratum, and seeks to prove a frequency effect in the acquisition of the German pronoun es by Spanish-speaking learners. The hypotheses to be tested are the following:
Hypothesis 1: Spanish-speaking learners of German as a foreign language have less difficulties in using high-frequency German es constructions than lowfrequency ones.
Hypothesis 2: Spanish-speaking learners of German as a foreign language have less difficulties in using German es constructions which are to some extent equivalent to Spanish constructions.
In order to test these hypotheses, two main steps were necessary: firstly, I had to establish how many and which types of German es categories exist. As the traditional classification of the German pronoun es is not sufficient – as I will explain below –, I followed a usage-based constructionist approach and identified German es constructions. As a second step, I conducted an empirical investigation of the acquisition of these constructions by Spanish-speaking learners.
In the first empirical study – about German es constructions –, two spoken language corpora of contemporary German were used, provided by the University of Freiburg: the Big Brother corpus and the Call Home corpus, which consist of carefully transcribed
spoken language according to the GAT2 conventions (cf. Selting et al. 2009) and contain about 200,000 words each. I researched these two corpora and conducted a constructionist analysis of German es occurrences, following Goldberg’s (1995, 2006) definition
and assuming that a construction is a form–meaning pair that is non-compositional and non-predictable. According to Goldberg (2006), a further aspect can be the frequency with which a linguistic pattern occurs. Goldberg argues that even completely predictable patterns should be considered as constructions if they occur with sufficient
frequency. This frequency argument is difficult to apply (cf. Traugott/Trousdale 2013). In the constructional analysis, I drew up an argumentation of how to use frequency as an aspect and how to deal with the problem of frequency as the sole aspect. My study shows that the traditional classification model of German es, containing only four classes (cf. Eisenberg 2006), is not sufficiently fine-grained as a categorization, and does not mirror the mental representation that should be assumed in the speaker’s mind. In this corpus study I identified 21 es constructions. This is not the whole system of German es occurrences. However, it is sufficient to establish a profound idea of what such a system could look like. By additionally consulting the vast deWaC corpus (about 1.7 billion words) and using a randomised sample of 4,000 es occurrences, the frequencies of each were identified and a frequency hierarchy of the 21 identified es constructions was established.
The second empirical study investigated the acquisition of these constructions by advanced Spanish-speaking learners of German. It aimed to prove that high-frequency es constructions are better acquired than low-frequency ones. However, it would be
impossible to test 21 es constructions in an experimental setting. Therefore, some constructions had to be selected. For this purpose, the constructions were divided into frequency classes, with a differentiation between low frequency, middle frequency and
high frequency (hypothesis 1). Secondly, I built two groups concerning the equivalence to Spanish constructions: es constructions with no equivalent in Spanish, and es constructions where a somewhat similar structure in Spanish is at least possible (hypothesis
As a result, I had two criteria with which to differentiate between the es constructions: the degree of frequency in German and the degree of difference in comparison to Spanish constructions. I then had to select a few constructions from each class to systematically
compare them. However, I considered it important to have a closer look at empirical data before deciding which constructions to test. To this end, a small empirical study with authentic learner speech data was conducted. Here, I used a self-compiled corpus of non-guided interviews with six Spanish-speaking learners of German. These
learners were at an intermediate level of German, between B1 and B2 according to the Common European Framework of Reference for Languages (cf. Europarat, Rat für Kulturelle Zusammenarbeit 2007). Each of them took part in an interview of one hour, consisting of informal conversation. These interviews were transcribed and the transcripts analysed regarding the use of the pronoun es by the learners. This qualitative analysis of the learner data delivered interesting results. Generally speaking, highfrequency
constructions are mostly used correctly, whereas low-frequency constructions are more problematic for the learners. For example, the very high-frequency construction [[es][Copula][Predicative] + subsequent clause], as used in a structure like es ist schön, dass … (Engl.: it is lovely that …), does not seem to cause difficulties for the
learners. In general, the usage-based point of view helps to explain the empirical findings from the qualitative analysis. However, some findings of the corpus study cannot be explained with frequency values. For example, there are high-frequency constructions
that are not used by any of the six learners. In summary, it can be said that the results of the small qualitative analysis confirm the usage-based idea that high-frequency constructions are more quickly acquired than low-frequency ones. However, there are findings that do not easily fit the frequency-related explanation. Therefore, this first empirical step has provided a valuable indication of which constructions could be interesting to test in a systematic experiment. What is more, I gained important ideas about some methodological issues: some learners told me in the interview (without being asked) that they had already identified the pronoun es as a problem, and so they always made a conscious effort to use the pronoun when necessary. This leads me to the conclusion that in the experiment the test persons should not notice that the study focuses on the pronoun es, as they probably will not make any es mistakes if they are concentrating on this. Based on the considerations about frequency classes of the es constructions as well as their similarity to Spanish equivalents, and with the help of the findings from the qualitative analysis, eight es constructions were selected for the test.
Subsequently the test design had to be determined. Initially, it was planned to elicit these eight constructions in a guided interview task. However, this method of data collection would be impossible, as some constructions cannot be elicited discreetly. Therefore, the two test methods of grammaticality judgment task and cloze test were
chosen. For each of the eight selected es constructions, a small dialogue with a specific subject was designed, where the test persons rate the grammaticality of each of the sentences in that dialogue. The idea here is that the participants will instinctively concentrate on the topic of the setting, which distracts them from the real purpose.
Additionally, different distractors were added such as missing articles and incorrectly declined articles and nouns. All these steps aim to prevent the test persons from noticing that their assessment of sentences with a correctly placed or a missing pronoun
es is of importance. For the cloze test, seven short texts were designed, thematically linked to the seven dialogues, where every fourth word is missing.
The whole test was administered on the LimeSurvey platform, with the advantage that it can be carried out online, and was – after a pilot study – completed by 53 Spanishspeaking learners of German as a foreign language. The test persons were mainly selected according to their level of German (B2) and the amount of German input they had. As the experiment aimed to examine the influence of frequency distributions in the learners’ input, only test persons who had been living in Germany for at least half a year and who had a high level of German input were permitted.
After excluding some participants (e.g. with less than seven hours of German input per week), the data of 44 Spanish-speaking learners remained. The data were analysed regarding the two hypotheses to be tested. For the quantitative analysis, a linear mixed model in the software environment R was used. Although the participants performed as expected with regard to the frequency-based hypothesis, a deductive statistical analysis did not reveal a significant effect. The same is true for hypothesis 2. Therefore, neither
hypothesis could be verified. However, the test design was evaluated and found to be suitable. It was concluded that speakers’ L2 performance is highly variable and that the experiment should be replicated.
Finally, some didactic implications of the presented study were considered. It was proposed to use corpus analyses for instruction on German as a foreign language; for instance, high-frequency es constructions should be learned first, and low-frequency
constructions should follow later. Construction-based learning (or teaching, respectively) was suggested.
The present study contributes to Construction Grammar research on the German language as well as to usage-based research on SLA. Although no frequency effect was proved, it threw light on the acquisition of the German pronoun es in SLA and revealed
that a usage-based and constructionist approach may enrich the existing research on, and didactic of, the German es.
Citation Fahrner, A. (2016). Der Erwerb von es-Konstruktionen durch spanischsprachige Deutschlernende, Doctorat, Université de Neuchâtel et Albert-Ludwigs-Universität Freiburg i. Br., Neuchâtel et Freiburg i. Br..
Type Thèse (Allemand)
Année 2016
Departement academique Faculté des lettres et sciences humaines, Institut de langues et littératures anglaises et Albert-Ludwigs-Universität Freiburg i. Br.
Université Université de Neuchâtel et Albert-Ludwigs-Universität Freiburg i. Br. (Neuchâtel et Freiburg i. Br.)
Degré Doctorat
URL https://doc.rero.ch/record/322780?ln=fr