Over the last century, we’ve gained a great deal of evidence that language structures evolve in ways that optimize communication efficiency. In the lexicon for example, Zipf (1939) famously showed that words which are more predictable tend to be shorter, and vice versa. This relationship reduces overall speaker effort while preserving communication accuracy. In the first part of this talk, I will review some of the most interesting recent findings that illustrate the apparent influence of communication efficiency on lexicons and grammars.

In the second part, I will present two strands of our research in this area that are based on the fact that listeners process the speech stream incrementally, continually updating their lexical search as the phonetic signal unfolds. As a consequence of this incremental processing, segments earlier in words contribute on average more disambiguating information to lexical access than later segments. If languages evolve to optimize communication efficiency, we expect therefore that informative segments should be concentrated early in words where they can do the most work in lexical disambiguation – and further that this tendency should be strongest for the least predictable words where comprehension accuracy depends more on an informative acoustic signal. Here I’ll use data from a wide range of languages to show that this is in fact the case: words that are on average less predictable have relatively more informative early segments, while preserving a longer tail of redundant, confirmatory segments.

Second, I’ll review our recent work suggesting that the relatively low information of late segments in a word may influence the development of phonological rules which reduce lexical distinctiveness (such as word-final devoicing in German). In a typologically-balanced sample of 50 languages, we find that phonological rules which neutralize lexical distinctions are common at word-ends, but very rare at word-beginnings where neutralization would have a greater negative impact on lexical access.  Interestingly, we find this asymmetry in languages from every family and from every region of the world, suggesting that a bias toward word-final neutralization is a strong language universal. This is what we would expect if this asymmetric distribution in grammatical rules stems from a basic property of human linguistic cognition.