Data collected in the course English Grammar

This course stretched over the first and second semesters, consisting of 10 and 9 sessions of 90 minutes, respectively. In the first semester, the students had to master basic concepts of morphological and syntactical analysis such as parts of speech, roots, affixes, clause constituents, phrase constituents, use of comma, etc. The sec-ond semester featured a detailed analysis of the verb phrase and the verbal categories of tense, mood, aspect, voice and person-number. In both semesters, focus was on grammatical theory, however, with frequent references to the course Production of Written Texts, which was dedicated to the practical mastery of written English.

The first semester in each academic year concluded with a written exam of theoretical grammar. The students were asked 100 questions (with the exception of the exam in 2015, see below in Table 3-2), which they had to answer within 2 hours without the use of any aid.⁸ The students had to answer 60 questions correctly in order to pass the exam. In the years 2009-2011 and again from 2014 onwards the exam was fine-graded, whereas in 2012 and 2013, it was simply pass or not pass.

The results of the grammar exam played a crucial role in the analyses in the project.

Approximately 95% of the questions required no production of “ordinary”

English, but had to be answered by naming the linguistic element that the given question concerned, by using linguistic terminology. Hence, the exam required only receptive knowledge of English besides metalinguistic knowledge. Section 3.1.2 below gives a detailed overview of the topics of the exams.

Questions were of the type What is the underlined clause constituent? or What part of speech is “part” in this sentence? At most, 5% of the questions required actual, albeit very limited use of English, involving the insertion of missing commas into sentences lacking commas – or the statement that no comma was needed in a given sentence, or the insertion of a missing relative pronoun.

The students had to hand in three home assignments each in both the first and second semesters. The home assignments had not been standardised before 2013, nor were they optimised for a statistical analysis. For this reason, home assignments prior to 2013 were not error-analysed and tagged with metadata in the database, and therefore, the following describes only the home assignments from the autumn of 2013 onwards. Home assignment 1 in the spring of 2014 is also missing because it was an unsuccessful attempt at introducing a different type of grammar home

8 Madsen (2014) contains a complete set of exam questions.

signment. In any case, the home assignments played only a minor role in the anal-yses thus far. Nevertheless, also the home assignments which have not been ana-lysed yet are available for study in the future should the interest for it arise.

In the first semester, the assignments consisted of 100 questions each – just as the exam – although the composition of the question sets in the first and second assignments differed substantially from that of the exam. The reason for this was simply that the students could not reasonably be expected to know all the exam topics for the first two assignments. The third assignment, on the other hand, resem-bled the exam very closely. In fact, all the questions in all three home assignments in the autumn semesters were always composed of questions from previous exams. In both semesters, the students were given one week to complete the assignments, and they were allowed to use any aid except for human help.

In the second semester, the home assignments focused – besides a little repeti-tion of first semester topics – on the new topics of verbal categories. The second semester did not conclude with a dedicated exam. In fact, until the academic year 2014-2015, there was no exam at all in grammar at the end of the second semester.

This explains the meagre number of informants in the spring of 2014. However, in the spring terms of 2015 and 2016, the home assignments of the second semester formed part of the combined portfolio exam of Grammar and Production of Written Texts. This is the reason why these home assignments consist of fewer questions than the home assignments in the other semesters.

Table 3-2 summarises the home assignments and exams in English Grammar that have been error-analysed. Some of the home assignments were the same; this was most notably the case in autumn 2015, in which all three home assignments were the ones that had been employed in the autumn of 2014.

Table 3-2: Overview of the home assignments and exams in English Grammar

Semester Type of text Submissions Questions

Autumn 2009 Exam 56 100

Autumn 2010 Exam 62 100

Autumn 2011 Exam 54 100

Autumn 2012 Exam 58 100

Autumn 2013 Home assignment 1 92 100

Home assignment 2 90 100

Semester Type of text Submissions Questions

Home assignment 3 90 100

Exam 100 100

Spring 2014 Home assignment 2 32 100

Home assignment 3 30 100

Autumn 2014 Home assignment 1 77 100

Home assignment 2 74 100

Home assignment 3 73 100

Exam 68 100

Spring 2015 Home assignment 1 63 45

Home assignment 2 58 50

Home assignment 3 60 50

Autumn 2015 Home assignment 1 86 100

Home assignment 2 86 100

Home assignment 3 84 100

Exam 84 95

Spring 2016 Home assignment 1 76 45

Home assignment 2 76 45

Home assignment 3 73 45

Total 1702 148040

The home assignments and exams were evaluated, and the correctness of the stu-dents’ responses was noted as metadata in a binary fashion as either correct or incor-rect, as mentioned in Section 2.3.5. Missing responses counted as incorrect respons-es. Only the results of the first exam attempt at the end of the autumn semesters were stored in the database. The results of the reexams of students who failed the first exam were not kept and later discarded irretrievably because it was not stipulat-ed to pay special attention to these students.