Digital archives reveal how English evolved—and why it matters for AI
Researchers analyzing vast historical databases of English texts have identified blind spots in how we document language change, with major implications for training artificial intelligence systems. The findings highlight a critical gap: millions of words from everyday speech, informal writing, and marginalized communities remain missing from the digital record that shapes modern language technology.
Originaltitel: Historical Corpora of English
<p>In this chapter, we address some ways in which the use of corpora has revolutionised the study of the history of English. We first account for the development of historical corpora of English and discuss advantages and drawbacks associated with different corpus sizes. We also address types of language use that are not well represented in existing corpora, potential clashes between comparability and representativity, and features such as tagging and spelling normalisation. We then consider contributions that historical corpora have made to specific linguistic fields, notably in variationist studies, historical sociolinguistics and historical pragmatics, and illustrate historical corpus methodology by presenting a case study on sentence-initial and in Late Modern English based on the Corpus of Historical American English (COHA). We conclude the chapter with a list of desiderata for future corpus-based research on the history of the English language.</p>