The use of zo’n versus zulke ‘such’ in Belgian and Netherlandic Dutch. Testing hypotheses relating to lexical biases, function, register and noun type.

Topic & objectives – This study investigates the use of the determiners zulke vs. zo’n ‘such’ in front of plural and uncountable singular nouns in European Dutch (Ghesquière and Van de Velde 2011; Van Olmen 2019). Countable singular nouns are not taken up because previous research indicates that zulke is exceedingly infrequent in that context (Van Olmen and van der Auwera 2014: 217). In addition, Van Olmen and van der Auwera (2014: 217) show that zo’n is generally more popular in Belgian Dutch than in Netherlandic Dutch. Three hypotheses are put to the test. The first predicts that both Belgians and Dutchmen will be more inclined to use the ‘Belgian’ variant zo’n in phrases that are more often used in Belgian Dutch. This would be an effect of lectal contamination (Pijpops and Van de Velde 2018). The second hypothesis states that in Netherlandic Dutch, there will be an important distinction between the functions that both determiners fulfill, among the instances in front of uncountable, singular nouns. In that context, zulke will be preferred for identifying the following noun, whereas zo’n will be preferred for intensifying the following noun (Van Olmen 2019: 218). The third hypothesis claims that in Belgian Dutch, there will be an important distinction in register, among the instances in front of plural nouns. In that context, zulke will be preferred in a formal register, whereas zo’n will be preferred in an informal register (Taaladvies.net).
Data – To test these hypotheses, data were drawn from the Sonar corpus of written Dutch (Oostdijk et al. 2013) and the Corpus of Spoken Dutch (Oostdijk et al. 2002). Speaker information is available for the entire Corpus of Spoken Dutch, so all of its material was used, whereas for the Sonar corpus, only the material was used for which writer information is available. First, a list of all potentially uncountable nouns was drawn from the Reference File Dutch (van der Vliet 2007; Referentiebestand Nederlands - RBN 2014). Next, all instances of zo’n and zulke, including spelling variants, were extracted from the corpora, when they were followed within the five words by a plural noun or a potentially uncountable singular noun. This yielded 9206 instances, which were then manually checked.
Method and preliminary results – To test the first hypothesis, a measure was calculated for each unique phrase, i.e. each unique lexical realization of the noun phrase containing zo’n or zulke, that quantified how often the phrase is used in Belgian Dutch compared to Netherlandic Dutch, regardless of whether it is used with zo’n or zulke (Pijpops & Van de Velde 2018). To test the second hypothesis, all data were manually annotated for their function, based on the criteria mentioned in Altenberg (1994: 234) en Van Olmen (2019: 227, 232, 235, 255). Finally, all hypotheses were put to the test used mixed model regression analysis (Gries 2015; Speelman, Heylen and Geeraerts 2018). The results show that only one hypothesis is straightforwardly confirmed, while another is confirmed pending important qualifications, and another is not confirmed.