In this paper, we argue that comparable collections of historical written resources can help overcoming typical challenges posed by heritage texts enhancing spelling normalization, POS-tagging and subsequent diachronic linguistic analyses. Thus, we present a comparable corpus of historical German recipes and show how such a comparable text collection together with the application of innovative MT inspired strategies allow us (ⅰ) to address the word form normalization problem and (ⅱ) to automatically generate a diachronic dictionary of spelling variants. Such a diachronic dictionary can be used both for spelling normalization and for extracting new "translation" (word formation/change) rules for diachronic spelling variants. Moreover, our approach can be applied virtually to any diachronic collection of texts regardless of the time span they represent. A first evaluation shows that our approach compares well with state-of-art approaches.
展开▼
