The shortness of roots explains limits on lexification richness

Talk by Martin Haspelmath, Max Planck Institute for Evolutionary Anthropology.

In this talk, I propose a new explanation of universal limits on lexification richness. In a range of different domains, it has been observed that more frequently needed meanings tend to be expressed in a more differentiated way. For example, close kin are more lexically differentiated than distant kin (e.g. brother/sister vs. cousin), modes of walking are more richly lexified than modes of flying (e.g. run/walk/creep vs. fly), and snow and ice are expressed my more different words in cold regions.

This has been explained with respect to a trade-off between informativeness (for the addressee) and complexity (for the speaker) (e.g. Kemp et al. 2018). However, “system complexity” does not seem to be a relevant processing bottleneck, and I will argue that the cost incurred by speakers primarily concerns articulation: While hearers want messages to be clear (or informative), speakers want them to be short (or easy to articulate). And since roots cannot be too long (cf. the “Root Length Constraint” first observed in Haspelmath 2023), this means that only highly frequent meanings can be expressed by roots (given the Zipfian efficiency principle). 

That frequency leads to greater differentiation was already observed by Mańczak (1970), but I argue that this is best explained by root shortness: If a meaning is rare, it must be expressed by a long form, and if the form needs three or four syllables, it must be a composite form (e.g. “fly fast”, contrasting with “run”). I will speculate that this has not been noted before because researchers thought of the lexicon as consisting of “words”, whereas it actually consists of a large number of diverse composite expressions in addition to roots (Haspelmath 2024), where “words” have no particular status. And finally, I will suggest that Brøndal’s “Principle of Compensation” may also find its explanation in a similar way.

Bio

Martin Haspelmath is a comparative linguist who studies the diversity of the world's grammatical and lexical systems and tries to understand what is universal about them. He is best known for coediting the "World atlas of language structures" (2005/2013), and as a co-founder of "Language Science Press". He is also an honorary (adjunct) professor at Leipzig University.

References

• Brøndal, Viggo. 1943. Structure et variabilité des systèmes morphologiques. In: Brøndal, Viggo, Essais de linguistique générale. Copenhague: Munksgaard.
• Haspelmath, Martin. 2023. Coexpression and synexpression patterns across languages: Comparative concepts and possible explanations. Frontiers in Psychology 14. (doi:10.3389/fpsyg.2023.1236853) 
• Haspelmath, Martin. 2024. Four kinds of lexical items: Words, lexemes, inventorial items, and mental items. Lexique: Revue en Sciences du Langage (34). 71–95. (doi:10.54563/lexique.1737)
• Kemp, Charles & Xu, Yang & Regier, Terry. 2018. Semantic typology and efficient communication. Annual Review of Linguistics 4(1). 109–128. (doi:10.1146/annurev-linguistics-011817-045406)
• Mańczak, Witold. 1970. Sur la théorie die catégories “marquées” et “non-marquées” de Greenberg. Linguistics 8(59). 29–36. (doi:10.1515/ling.1970.8.59.29).