Literary and Linguistic Computing Advance Access originally published online on August 27, 2006
Literary and Linguistic Computing 2006 21(4):543-557; doi:10.1093/llc/fql036
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
Mutual Comprehensibility of Written Afrikaans and Dutch: Symmetrical or Asymmetrical?
Department of Scandinavian Studies, University of Groningen, Groningen, The Netherlands
Department of Linguistics, Radboud University Nijmegen, Nijmegen, The Netherlands
Correspondence: Charlotte Gooskens, Department of Scandinavian Studies, University of Groningen, P.O. Box 716, 9700 AS Groningen, The Netherlands. E-mail: c.s.gooskens{at}rug.nl
The two West-Germanic languages Dutch and Afrikaans are so closely related that they can be expected to be mutually intelligible to a large extent. The present investigation focuses on written language. Comprehension was established by means of cloze tests on the basis of two newspaper articles. Results suggest that it is easier for Dutch subjects to understand written Afrikaans than it is for South African subjects to understand written Dutch. In order to explain the results, attitudes as well as several types of linguistic distances were assessed. The relations between attitude scales and intelligibility scores were few and weak. Asymmetries in the linguistic relationships between the two languages are probably more important, especially the asymmetries in the number of noncognates and the opacity of the relatedness of cognates. These asymmetries are caused by historical developments in Dutch and Afrikaans, with respect to the lexicon, grammar, and spelling.
1 The quantitative information about Afrikaans is from Webb (2002).
2 We aimed at texts with an average level of difficulty. That is why we selected human interest items from a regional newspaper (De Leeuwarder Courant), which addresses readers of all educational levels. To make sure that the texts indeed had an average level of difficulty, we applied the so-called LIX-index (Björnsson, 1968). This is the mean number of words per sentence plus the percentage of words which are longer than seven letters. Texts with a LIX-value of 35 to 44 have a mean level of difficulty. The mean LIX-value for the two Dutch texts was 42. So, this confirmed that the texts indeed had the intended level of difficulty.
3 The cloze test was developed by William Taylor in 1953 in America. Since then it has been a widely used tool for measuring the intelligibility of texts. Sometimes the words are placed above the text, like in our investigation, but sometimes the test subjects have to think of words to fill in themselves.
4 In principle, we had two options. The first option was to take original texts in both languages. However, in that case the texts might not have been comparable. To assess a possible asymmetry in intelligibility, it is imperative that there be no differences between the texts in the two languages in vocabulary and style. We therefore decided to make use of translation. We took great care to obtain high-quality translations. The Dutch texts were translated into Afrikaans by a South African student of English and corrected by several South African linguists.
5 In addition, the subjects were also tested for their listening comprehension (van Bezooijen and Gooskens, 2005b). The listening tests preceded the cloze tests.
6 In fact, we found no relationship between the subjects’ performance for the two languages. The Pearson correlation between the scores for the Dutch and Afrikaans texts was –0.09 for the Dutch subjects and 0.06 for the South African subjects. Neither coefficient is significant at the 5% level.
7 We normalized for the difference in level of difficulty of the two texts in the following manner. We determined the mean intelligibility score for each of the Afrikaans texts and next calculated the difference between the two means. The relatively difficult feminist text was on average 2.0% easier than the overall mean and the relatively difficult dating text was on average 2.0% more difficult than the overall average. In order to normalize, all subject who read the Afrikaans feminist text had 2.0% added to their score and subjects who read the Afrikaans dating text had 2.0% subtracted from their score. The same procedure was followed for the Dutch texts. In this case 0.15% was added or subtracted.
8 Sometimes the frequency of word pairs related via a synonym may differ somewhat in the two languages. However, we have taken the view that as long as the word can be assumed to be known by readers of the age group at hand, the frequency does not have to be identical.
9 We thank Bart Alewijnse for calculating the Levenshtein distances.
10 We measured the distances on the basis of binary differences. Different symbols increase the distance, identical symbols do not increase the difference. For spoken texts, a more advance method can be used, where the phonetic distances between the individual sounds are taken into account.
11 We thank Vincent van Heuven for his help.
12 We thank Sulette Bruwer for her help.
13 It was not possible to carry out a t-test for dependent samples since the Dutch text were aligned differently with the Afrikaans texts than the Afrikaans texts with the Dutch texts.