How closely does the grammar of the Dutch word 'er' correspond to the use of 'er' found in Dutch corpora?
Isa M. Apallius de Vos · 2020
In the context of natural language processing, it is critical to use grammar rules that are accurate and supported by credible sources. In this research, the grammar of the Dutch word 'er' is tested to examine whether the literature on 'er' corresponds to its use by Dutch speakers. Four uses of 'er' are considered: existential 'er' (erX), pronominal 'er' (erP), locative 'er' (erL), and quantitative 'er' (erQ). Three structures with 'er' uses not supported by the literature are examined: adjacent 'ers' in the midfield, non-adjacent 'ers' in the midfield, and 'erX' in the prefield with 'erL' or 'erP' in the midfield. These structures are explored by using two corpora to find sample sentences and to calculate the frequency of these phenomena. From the results it has been concluded that adjacent and non-adjacent 'ers' in the midfield do not occur frequently enough to deem them acceptable (the frequency is 2.3% for both corpora compared to the acceptable 'erX' with 'erQ' structure) to conclude that it needs more research to test its use by native speakers.