Open Access

In a recent study, Bromham, Yaxley and Cardillo (BYC) assembled an impressive database of 1,197 languages spoken across 13,100 islands to examine how predictions from island biogeography theory (IBT) apply to linguistic diversity and whether islands act as drivers of language change. I commend the authors for the significant effort in compiling such a truly global dataset of island languages, and for quantitatively demonstrating that islands hold a disproportionately greater share of the world’s languages than expected based on their land area. However, I here contend that the core model used to test the predictions of IBT and the hypothesis that islands shape language evolution is empirically implausible and plagued by several critical issues.

From a cross-linguistic perspective, language models are interesting because they can be used as idealised language learners that learn to produce and process language by being trained on a corpus of linguistic input. In this paper, we train different language models, from simple statistical models to advanced neural networks, on a database of 41 multilingual text collections comprising a wide variety of text types, which together include nearly 3 billion words across more than 6,500 documents in over 2,000 languages. We use the trained models to estimate entropy rates, a complexity measure derived from information theory. To compare entropy rates across both models and languages, we develop a quantitative approach that combines machine learning with semiparametric spatial filtering methods to account for both language- and document-specific characteristics, as well as phylogenetic and geographical language relationships. We first establish that entropy rate distributions are highly consistent across different language models, suggesting that the choice of model may have minimal impact on cross-linguistic investigations. On the basis of a much broader range of language models than in previous studies, we confirm results showing systematic differences in entropy rates, i.e. text complexity, across languages. These results challenge the long-held notion that all languages are equally complex. We then show that higher entropy rate tends to co-occur with shorter text length, and argue that this inverse relationship between complexity and length implies a compensatory mechanism whereby increased complexity is offset by increased efficiency. Finally, we introduce a multi-model multilevel inference approach to show that this complexity-efficiency trade-off is partly influenced by the social environment in which languages are used: languages spoken by larger communities tend to have higher entropy rates while using fewer symbols to encode messages.

For language scientists, a prima facie advantage of AI-generated data over human-created content is that AI outputs are generally regarded as free from copyright. This submission addresses this issue in some detail.

In the present contribution, we describe the features of the CLARIN SIS (Standards Information System) that have been designed to assist data-deposition centres in CLARIN. We also show what is needed to go beyond the originally designated target, in order to provide service to sibling and descendant research infrastructures, of which DARIAH and Text+ are taken as examples.