Abstract
The results of the crosslinguistic (Dutch, Italian and German) database analysis and experiments presented in my study show that there are interesting similarities between the pattern of article omission by adults in newspaper headlines and the pattern of article omission observed in language acquiring children. Further, I show that in
... read more
both languages and in both groups of speakers article omission is optional. Moreover, article omission appears to be influenced by the presence of a finite verb or a preposition and by the sentence position of the article-requiring noun. Although this effect works in the same direction in all three languages (less omissions if the verb is finite, less omissions after a preposition, less omissions in sentence-internal position), the concrete consequences are different. In order to come to a unified model that can account for these findings I introduce a base model of article selection that describes the process of article selection under normal and specific circumstances, with normal speakers and speakers with limited processing resources. According to this model, for an article to be produced, the targeted form needs to receive the highest activation among the members of the article set. I argue that the final output of the article selection process is the result of a competition between all members of the set. In a sense, the article selection process therefore can be characterized as resolving uncertainty with regard to which element of the set will win the competition. The higher the uncertainty, the more processing resources are necessary to select an article from the set. In my study I present an information-theoretical method to calculate the uncertainty level in the article set, the relative entropy value. The relative entropy of the article system reflects the differences in probability distribution between the elements: the higher the differences in probability distribution (and, thus the higher the value of relative entropy), the more processing resources are required to select an article from the article set. I show that the relative entropy value is higher in Dutch than in German, and higher in German than in Italian. I argue that both children and headline writers are sensitive to differences in the processing resources necessary for article selection, children because of their limited processing resources, headline writers (and readers) because of the restricted time readers have available. I show that with the measure of relative entropy we can account not only for the crosslinguistic differences in omission pattern, but also for the relation between article omission and finiteness, for the influence of the use of a preposition before the article-requiring noun on the omission of articles, for the relation between article omission and sentence position and for differences in omission pattern of different articles (particularly the differences in omissions of ‘de’ and ‘het’ in Dutch and the differences between omission of definite and indefinite articles in Italian).
show less