Investigating the Phonetic Organisation of the English Language via Phonological Networks, Percolation and Markov Models
Applying tools from network science and statistical mechanics, this paper represents an interdisciplinary analysis of the phonetic organisation of the English language. By using open datasets, we build phonological networks, where nodes are the phonetic pronunciations of words and edges connect words differing by the addition, deletion, or substitution of exactly one phoneme. We present an investigation of whether the topological features of this phonological network reflect only lower or also higher order correlations in phoneme organisation. We address this question by exploring artificially constructed repertoires of words, constructing phonological networks for these repertoires, and comparing them to the network constructed from the real data. Artificial repertoires of words are built to reflect increasingly higher order statistics of the English corpus. Hence, we start with percolation-type experiments in which phonemes are sampled uniformly at random to construct words, then sample from the real phoneme frequency distribution, and finally we consider repertoires resulting from Markov processes of first, second, and third order. As expected, we find that percolation-type experiments constitute a poor null model for the real data. However, some network features, such as the relatively high assortative mixing by degree and the clustering coefficient of the English PN, can be retrieved by Markov models for word construction. Nevertheless, even Markov processes up to third order cannot fully reproduce other patterns of the empirical network, such as link densities and component sizes. We conjecture that this difference is related to the combinatorial space the real and the artificial phonological networks are embedded into and that the connectivity properties of phonological networks reflect additional patterns in word organisation in the English language which cannot be captured by lower order phoneme correlations.
KeywordsMarkov Process Regular Graph Phonological Similarity Giant Component Mental Lexicon
The authors acknowledge the Doctoral Training Centre in Complex Systems Simulation at the University of Southampton, in the completion of this work. MS was supported by an EPSRC grant (EP/G03690X/1).
- 1.Solé, R.V., Seoane, L.F.: Ambiguity in Language Networks. arXiv:1402.4802 (2014)
- 2.Kello, C.T., Beltz, B.C.: Scale-free networks in phonological and orthographic wordform lexicons. In: Approaches to Phonological Complexity (2009)Google Scholar
- 4.Aitchison, J.: Words in the Mind: An Introduction to the Mental Lexicon. Wiley (2012)Google Scholar
- 5.Ferrer-i-Cancho, R., Solé, R.V.: The small world of human language. Proc. R. Soc. Lond. Series B: Biol. Sci. 268(1482), 2261–2265 (2001)Google Scholar
- 15.Ke, J.: Complex Networks and Human Language. arXiv preprint cs/0701135 (2007)Google Scholar
- 16.Siew, C.S.: Community structure in the phonological network. Front. Psychol. 4 (2013)Google Scholar
- 17.Luce, P.A., Pisoni, D.B.: Recognizing spoken words: the neighborhood activation model. Ear Hear. 19(1) (1998)Google Scholar
- 19.Vitevitch, M.S.: The neighborhood characteristics of malapropisms. Lang. Speech 40(3), 211–228 (1997)Google Scholar
- 20.Stella, M., Brede, M.: Patterns in the English language: phonological networks, percolation and assembly models. J. Stat. Mech. P05006 (2015)Google Scholar
- 21.Newman, M.: Networks: An Introduction. Oxford University Press (2010)Google Scholar
- 23.Grimmett, G., Stirzaker, D.: Probability and Random Processes. Oxford University Press (2001)Google Scholar