Even though the corpus of naturally occurring data is sizeable, it did not and could not answer all the questions that the project sought to answer. Therefore, further data were elicited in surveys specifically for the purposes of the project in two broad categories, non-linguistic and linguistic. The surveys were administered in various ways, such as Google forms, Moodle’s quiz function (2016) and to a lesser extent on paper. The following subsections describe the elicited data in more detail.
3.2.1 Non-linguistic data
The non-linguistic data collected consist of educational background information about the students and their motivation and attitude to various issues concerning their studies. Non-linguistic data were only elicited from the Danish informants.
Non-linguistic data were not elicited from the Slovene and Serbian informants for two reasons (except for basic demographic information on age and gender). First, I did not want to outstay my welcome at my hosts’ since I was very much dependent on the Slovene and Serbian students’ willingness to do something that was indeed foreign to them. Therefore, I focused on the most essential and posed them as few questions as possible.
Second, even if willingness had been in great abundance, it would have been out of the scope of the project to include a socio-psychological study of the Slovene and Serbian informants. Therefore, it sufficed for me to merely ascertain that all the informants were comparable on the most essential account: being freshmen of Eng-lish language studies.
3.2.1.1 Data on educational background
The elicitation of data on the Danish students’ educational background went through some refinement after the first version of the survey in 2012-2013. Unfortunately, it was not possible to do a survey retroactively with the informants from the years prior to 2012 or re-do the newest version of the survey with the informants from previous years. Thus, only the informants’ age and gender are known from before 2012, and data collected from 2012 onwards are not completely comparable across the academic years.
Appendix G lists the items surveyed beginning in 2012-2013. The surveys on educational background were administered some time into the academic year in order to distribute the extra workload caused by the surveys evenly because some linguistic surveys (cf. Section 3.2.2.1) had to be done at the beginning of the aca-demic year, and because the information collected in the surveys on educational background is static. It would not change during a semester; therefore, it did not matter when this information was elicited.
The data on the educational background were used to some extent in Madsen 2014 (Section 6.1). The findings therein suggest that these data deserve further re-search.
3.2.1.2 Data on attitude and motivation
As the project progressed, I realised that not every aspect of the informants’ linguis-tic performance could be explained on purely linguislinguis-tic grounds. Therefore, I decid-ed to perform an analysis of the students’ motivation to study, attitude to various linguistic matters and their linguistic self-awareness.
I had carried out some rudimentary survey in the Department of Business and Social Sciences, Aarhus University, where I had been a part-time teacher in previous years. I then made a somewhat more concerted effort in my home Department of English Business Communication in the academic year 2013-2014. The results of this were presented at a national conference in 2014 and are described in Section 6.2 and summarised in Table 6-2 and Table 6-3.
The final attempt was loosely based on Gardner’s Attitude and Motivation Test Battery (1985) and of a pre-test-post-test design. The pre-test survey was adminis-tered on the first day the students had class with me in the autumn semester of 2014, and the post-test survey was administered on the last day of that semester. Its find-ings were published in Madsen 2016 (Section 6.5).
3.2.2 Linguistic data
Various questionnaires were employed to probe linguistic phenomena which were underrepresented in the corpus of naturally occurring texts, or whose use was to be tested under specific circumstances. As far as possible, these extracurricular surveys were also used pedagogically, providing instantaneous feedback to the students about their performance.
3.2.2.1 Linguistic knowledge upon entry
One of the major questions for the project was how much knowledge of theoretical grammar the Danish students brought with themselves from high school upon enter-ing the university. There was no entrance exam to the Department of Business Communication, and the students gained admission depending on their high-school grades. Since the high-school grades were not particularly revealing, no one knew what level of practical command of written English and of theoretical grammar the students had reached by the time they entered the university.
A survey, called survey 0 in the database, was designed to measure the stu-dents’ knowledge of English upon their entry. It was introduced in the autumn se-mester of 2014 and administered on the very first day the students had English Grammar. Originally, this survey had been envisaged to serve as an object of com-parison for the grammar exam so that the students’ development could be gauged.
However, I had to realise that only within two topics of the grammar exam, parts of speech and clause constituents, could the students’ knowledge be tested because only these two topics could reasonably be expected that the students had been taught in school. The survey was composed of 15 questions concerning clause constituents and 9 questions about parts of speech, all taken from the grammar exam of 2005.
Ten more questions tested whether the students could distinguish grammatical-ly incorrect sentences from correct ones. Some of the incorrect sentences were taken from previous students’ texts in Production of Written Texts, containing typical Danishisms. Typically, the students performed better on the grammaticality-judgment part than on the linguistic terminology part of the survey.
3.2.2.2 Relativization
Another major topic of the linguistic surveys was the use of relative clauses and relative pronouns. Based on informal pre-project observations, a contrastive analysis of English and Danish and preliminary statistics of the project, I reckoned that the use of English relative pronouns might be one of the major grammatical challenges facing Danish students. The articles Madsen (2015b in Section 6.6, and forthcoming in Section 6.9) were dedicated to this issue.
As mentioned in the Introduction, the Slovene and Serbian informants were to serve as an object of comparison for the testing of the theory of cross-linguistic influence. These informants were given the same questionnaires on relativization which the Danish informants received (see Appendix H).
3.2.2.3 Grammar exam for the Slovene informants
In 2014, Slovene informants were given the same exam in theoretical grammar that the Danish students had just taken. Since it was an extracurricular activity for the Slovene students, it counts as elicited data in the project’s terminology. The Slovene students performed markedly poorer than the Danish students had; only 1 out of 17 would have passed, and only barely.
Admittedly, the students were not prepared for the test. Neither were they in-formed of the content of the upcoming test in advance, nor did their curriculum focus on theoretical grammar. Especially the latter made me realise how much weight theoretical grammar is given in my department. Nevertheless, despite what-ever weaknesses the Slovene freshmen may have with theoretical grammar, their practical command of English does not seem any less developed than that of Danish freshmen, judging from informal conversations and chats with them.
4 Theory
As described in Chapter 2, this study is highly data-driven in its approach. Neverthe-less, three theories were tested explicitly in the project: the contrastive hypothesis, also called the theory of cross-linguistic influence (Lado 1957, Odlin 1989, Ring-bom 2007, Jarvis and Pavlenko 2008, Jarvis 2011), Krashen’s monitor theory (1978, 1982), and the accessibility hierarchy of relativization by Keenan and Comrie (1977).
The sections below elaborate the theories mentioned above and why they were selected for testing in the project. It is not my aim to give a detailed list of pros and cons of the theories in this thesis; it has been done by numerous authors (for exam-ple, McLaughlin 1978, and Krashen himself concerning contrastive analysis (1981:
64ff). Instead, I want to focus on testing the theories against my data. As Mackey expressed it (2014), it is important to test and re-test theories continuously if social sciences are to approach the rigour of natural sciences. The theories that provided the conceptual background of the project are described in Chapter 5.