An acoustic and statistical analysis of spanish mid-vowel allophones

In 1918 Navarro Tomás claimed that Spanish mid vowels have open and close allophones. Whilst some acoustic studies have supported these claims, others have not. The conclusions in these studies were subjective since they failed to apply statistical analysis to the acoustic data. Statistical analysis is required to determine whether data with a range of values can be grouped into distinct categories. The present study investigates the production of Spanish /e/ and /o/ in several contexts in which Navarro Tomás claimed open and close allophones would be found. Recordings were made of a male and a female speaker of educated Madrid Spanish, the dialect which Navarro Tomás originally described. First and second formants were measured and statistically analysed. Formant frequencies did not cluster into two groups associated with the contexts in which Navarro Tomás claimed each allophone would occur. Other potential allophones were identified: fronted and retracted allophones for /o/; and close-fronted, central, and openretracted allophones for /e/.