Abstract: A plurality of source text files are read, representing similar information but in different natural languages. The files have correlated layouts, in that the same layout commands are employed at similar points in the files. Similar text, from the respective files, is aligned by identifying its position between equivalent word processing commands. Preferably, intermediate files are produced in which the word processing (WP) commands are converted into an identifiable form. Aligned text, which differs between the intermediate files whereas WP commands will not differ, is identified by a differential comparison operation, such as a call to DIFF within a UNIX environment.

Abstract: Aligned corpora are generated or received from an external source. Each corpus comprises a set of portions aligned with corresponding portions of the other corpus, for example, so that aligned portions are nominally translations of one another in two languages. A statistical database is compiled. An evaluation module calculates correlation scores for pairs of words chosen one from each corpus. Given a pair of text portions (one in each language) the evaluation module combines word pair correlation scores to obtain an alignment score for the text portions. These alignment scores can be used to verify a translation and/or to modify the aligned corpora to remove improbable alignments.