Every human language builds words from syllables, but the rules for building them differ so dramatically that some languages allow vowel-free words like the Berber *rgl* 'close!' while others permit only simple consonant-vowel pairs. This matters because linguists have long catalogued these differences—English allows *gl* but not *lm* at the start of a word, for example—without a systematic understanding of *how* syllable structures evolve over time. The researcher will develop open-source computational tools to infer syllable structure from word lists across 897 languages, then model how those structures change through time. This fills a gap: existing typologies are based on small language samples, and the evolutionary pathways that turn a simple-syllable language into a complex one remain poorly understood. This is fundamental science. It will not directly change how people speak or build consumer products. But understanding how sound systems diversify could eventually inform speech recognition software, language teaching tools, or efforts to document endangered languages before they vanish. Past work on sound change has already shaped everything from forensic linguistics to how digital assistants parse accents—a deeper evolutionary model adds a layer of predictive power to that foundation.
View original technical description
All languages have syllables, but all languages do syllables differently. Syllables are universal building blocks, binding together consonants and vowels into larger units, which can then be formed into words and phrases, and further into speech and song. Yet despite being ubiquitous, syllables possess a structure and complexity varies enormously from language to language. In this Catalyst project I will uncover the processes by which this diversity evolves by developing open-source computational tools for syllable analysis and applying these to a selected dataset of words from 897 languages. Syllables are defined by both their shape and their content. Some languages allow only syllables with a simple shape, consisting of a consonant and a vowel (CV, as in the syllables lo and go in the word logo). Others, such as English and many other European languages, permit more complex shapes, e.g. CVC (as in both syllables of san.dal) or CCV (as in the glo of glory). At the extreme, languages can permit highly complex syllable shapes, forming words without any vowels at all, e.g. Tashlhiyt Berber rgl 'close!'. Languages may also impose different restrictions on the content which can fill these shapes. While English allows a cluster of two consonants at the beginning of the word, not every combination of consonants can occur in this position: a cluster like gl, as in glory, is permissible, but one such as lm, as in Pashto lmundz ‘prayer’, is not. Languages evolve with respect to the types of syllables they allow. Old Church Slavonic migla ‘fog’ and Polish mgla ‘fog’ have the same Common Slavonic origin, but Polish has lost the first vowel, creating an initial cluster /mgw/, which is impossible in Old Church Slavonic. English words like knee and knight used to be pronounced with both /k/ and /n/, but the /k/ sound has been lost and Modern English words cannot begin with the sequence /kn/, unlike in German, where in the corresponding words Knie and Knecht both /k/ and /n/ are pronounced. While syllables can change in a multitude of ways, much remains to be understood about how and why these changes occur. In this project, I will systematically investigate how languages differ in terms of the shape and content of the syllables they allow and model how this changes through time. To do so, I will develop a suite of open-source computational tools for syllable analysis and apply these tools to a specially developed dataset of word forms in a diverse set of languages. First, I will write tools to infer syllable structure from lists of word forms. This will allow me to evaluate syllable structure for many more languages than has hitherto been possible, allowing me to refine existing typologies and to state more clearly the distribution of different syllable types across the languages of the world. Second, I will develop tools to model the evolution of syllable structure through time, so arriving at a more detailed and nuanced understanding of how sound systems diversify and change in human languages.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know