Problems using the traditional acoustic cues for the phonological interpretation of vowels

The need to add acoustic characteristics to description of the Latvian sounds traditionally described and phonologically classified only on the basis of their articulation has raised the question on relations between acoustic properties of sounds and features significant for their perception and classification. Using formant data of Cardinal and IPA vowels and plotting them on two-dimensional acoustic or psycho-physical vowel planes, e. g., F2 vs. F1, F2′ vs. F1, F2-F1 vs. F1, F2′-F1 vs. F1 etc., it has been illustrated that it is problematic to make distinction between front, central and back unrounded vowels if the classification is attempted on the basis of values used to define axis of these plots. The same is true if the phonological classification is attempted by assigning values to parameters corresponding to acoustic features derived from the dimensions “flat” and “spread” suggested by Gunnar Fant. Although the use of 3-D vowel space instead of 2-D vowel plane resolves overlapping of different vowels one has to bear in mind that changes in one articulatory dimension evoke changes in the values of several formants thus not allowing direct relations between articulatory and acoustic or psycho-physical parameters.