Language-specific coarticulatory patterns: a comparative study of Arabic and Spanish

Background. We investigate a hypothesis for languages to be phonologically categorised based on direction of coarticulation: i.e. whether coarticulation mainly occurs from consonant to vowel (C-to-V) or from vowel to consonant (V-to-C), and whether this is connected to language-specific phonological contrasts. We report preliminary findings from an Ultrasound Tongue Imaging (UTI) study on Arabic and Spanish focusing on the velar stop /k/. Previous studies highlight language-specific preferences for the direction of coarticulation: e.g. consonant to vowel (Caucasian) versus vowel to consonant (English; Greek). Furthermore, in an acoustic study of V1CV2 utterances, it was noticed that English and Swedish allow the articulators to start moving toward the configuration of V2 at the acoustic end of V1 before the consonant occlusion, while Russian did not. Our initial hypothesis was that coarticulatory preferences mainly result from the ratio between vowels and consonants in the phoneme inventory: i.e. languages with a high number of consonantal contrasts should show different coarticulatory patterns from languages with fewer phonemic consonants. With this in mind, Arabic and Spanish form a good comparative pair: Modern Standard Arabic has 39 consonants and 3 vowel qualities that also admit a quantity contrast. Spanish has a smaller system, comprising 17-19 consonants (dialect dependently) and 5 vowel qualities.