Measuring Vowel Harmony within Hungarian, the Indus Valley Script Language, Spanish and Turkish Using ERGM

Front-back vowel harmony is an important characteristic of many languages. Testing whether an untranslated script has vowel harmony may aid its decipherment. This paper tests vowel harmony for three different modern languages (Hungarian, Spanish and Turkish) as well as the extinct underlying language of the undeciphered Indus Valley script. We also introduce a novel vowel harmony index based on the Exponential Random Graph Model for graphs. To achieve this, we first select words from each of the modern languages (Hungarian, Turkish, and Spanish) from their Swadesh list. Then we divide each word into syllables, isolating the vowels. We then analyze the three modern languages using Exponential Random Graph Model methods. The results indicate that this procedure and the vowel harmony index are feasible to define the degree of vowel harmony in a language. The procedure is then extended to the undeciphered Indus Valley Script. Our results indicate that the underlying language of the Indus Valley Script also had vowel harmony. We found that on average the odds of the IVS having vowel harmony were 6.61 times higher than would be found in a random graph.