Search

Het is bekend dat de variatie in het hedendaagse Afrikaans tussen (i) de complementzin met bijzinsvolgorde en het onderschikkend voegwoord dat en (ii) de zgn. "skoon bysin" zonder voegwoord en met hoofdzinsvolgorde onder meer bepaald wordt door lexicale factoren: de skoon bysin zou vooral voorkomen na matrixzinnen met sê, dink, glo, hoop of een ander (hoog frequent) werkwoord van communicatie of cognitie (zie bv. Malherbe 1966; Feinauer 1990; Bosch 1999). Empirisch onderzoek naar de lexicale voorkeuren van allerlei matrixwerkwoorden voor deze of gene constructie is echter nog nauwelijks voorhanden. In deze bijdrage wordt aan de hand van een distinctieve collexeemanalyse (Gries & Stefanowitsch 2004) nagegaan welke werkwoorden een significante voorkeur vertonen voor de dat-bijzin dan wel de skoon bysin in een corpus hedendaags journalistiek taalgebruik. We vergelijken de resultaten met de bevindingen van Van Rooy & Kruger (2016) en formuleren enkele voorzichtige conclusies over de semantische subtypes van werkwoorden die wel en niet worden aangetrokken tot de constructie met een skoon bysin. It is well-known that the alternation in current Afrikaans between (i) the complement clause with dependent word order and an overt complementiser dat ("that"), and the so-called skoon bysin "bare complement clause" without an overt complementiser and with independent word order is determined inter alia by lexical factors. The bare complement clause is said to occur particularly after main clauses with sê ("say"), dink ("think"), glo ("believe") and similar high-frequency verbs. However, not much empirical research into the preferences of various matrix verbs for the one or the other construction has been done. This article is aimed at identifying, by means of a distinctive collexeme analysis (Gries & Stefanowitsch 2004), the verbs that show a significant preference for either the dat-complement clause or the bare complement clause in a corpus of general journalistic text. We selected 180 test verbs: 76 verbs were taken from Braeckeveldt (2013) and 104 verbs were added to this corpus, mainly from Van Rooy and Kruger (2016). These verbs were all manually analysed and the observed frequencies extrapolated, using all available articles of Die Burger for the period 02/01/2003 to 01/05/2003, in total 6,3 million words. We identified 27 distinctive collexemes for the bare complement clause, i.e. verbs which show a significantly above-average preference for this construction, vs. 59 distinctive collexemes for the dat-complement clause. When we compare the results ofthe collexeme analysis with thefindings of Van Rooy and Kruger (2016) on the effect of the variable matrix verb, the similarities are greater than the differences. This indicates that the lexical preferences of verbs are quite solid: generally the same verbs are attracted to either the bare complement clause or the dat-clause in different corpora of present-day written Afrikaans. Distinctive collexeme analysis can be seen as an exploratory technique that can yield hypotheses about the semantic variables potentially playing a role in a specific alternation. The results from the quantitative analysis allow for a number of generalisations about the semantic sub-types of verbs attracted towards the bare complement clause or towards the dat-clause. For instance, causative matrix verbs consistently show a strong preference for the dat-clause: five of the ten most distinctive dat-collexemes belong to this type (sorg ("provide"), veroorsaak ("cause"), toelaat ("allow"), meebring ("cause/involve"), keer ("prevent")). The distinctive collexemesof the bare complement clause, by contrast, almost all belong to the semantic classes of verbs of communication (e.g. sê ("say"), vertel ("tell"), laat weet ("let know")) and, especially, verbs of cognition (e.g. dink ("think"), glo ("believe"), meen ("think", "mean"), besef ("realise"), weet ("know")), including the closely related sub-class ofverbs ofperception (sien ("see"), hoor ("hear"), voel ("feel")). With this article, we wanted to present empirical data on the lexical preferences of a large and heterogeneous set of matrix verbs for either the bare complement clause or the dat-clause, with a view to refining existing views on the exact kinds of verbs attracted to both constructions, and, hence, on the semantic variables that can be included in multifactorial work on the alternation in question. In future research, it would also be interesting to compare the semantic variables co-determining the alternation in Afrikaans to those that potentially play a role in other (Germanic) languages with an alternation between "ordinary" complement clauses and complement clauses displaying main clause characteristics such as different word order and/or the absence of a complementiser (e.g. English, German, Danish).