Investigating word recognition in intercomprehension: Methods and findings

Abstract

This article presents methods and findings from research on factors that determine the recognizability of cognate words in Germanic intercomprehension (more precisely: German speakers' reading comprehension in Germanic languages they have not learnt). Different types of written tests were carried out (free response, multiple choice, judgments on the probability of two words being cognates) in order to assess the importance of different aspects of linguistic similarity for the transparency of written cognates. Apart from the overall amount of differing segments of the cognate words, phonetic similarity between the differing segments – in particular an identical place of articulation – turned out to be most important, which can reflect either a spontaneous sense of similarity or familiarity with phenomena of variation between phonetically similar sounds. By adjusting similarity measuring to these results (weighted Levenshtein distance), it is possible, to a certain degree, to assess the transparency of cognates. Nevertheless, certain results could not be explained by phonetic similarity. Thus, recordings were made of subjects commenting on their way of proceeding in such tasks as well as in text decoding, which made clear that the subjects often attribute the same importance to other associations as they do to phonetic ones – even in the recognition of isolated words semantic connections in the mental lexicon are involved. In text context, the most frequent strategy seems to rely on an interplay of phonetic similarity and inference; however, in the subjects' reflections, the aspect of semantic probability manifestly overrides intuitions about phonetic similarity.