Reconstructing Word Meaning in Historical Linguistics

This article concerns a methodologically and empirically neglected aspect of historical linguistic practice: the reconstruction of word meaning. Following a general introduction that contextualizes the topic with background on semantic change, I discuss lexical semantic reconstruction (i) within the diachronic trajectory of a single language; (ii) as part of the historical-comparative reconstruction of language families and/or subgroups; and (iii) in situations in which derivational morphology contributes to the meaning of attested forms. Concerning (i), I discuss evidence from collocation patterns; heterosemy; radial sense structure; and grammaticalization. I also suggest that central senses of lexical items often change first, whereas secondary, derived senses survive. Concerning (ii), I highlight the absence of universally known and accepted methodological principles to guide semantic reconstruction. This means that, in general, reconstructed meanings from e.g., etymological dictionaries cannot be taken as data for generalizations on semantic change, nor can they be reliably compared across sources. Concerning (iii), I emphasize that deviations from semantic compositionality are the norm in word formation. This has consequences for what realistic semantic reconstruction of derived terms might look like. I illustrate with examples that are relevant for cultural reconstruction and linguistic paleontology.