Phonological theories invoke a number of different articulatory parametrizations. Traditional phonetic theories and some phonological theories parametrize vowel articulation in terms of tongue height and backness. Particle phonology [S. A. Schane, Phonology Yearbook 1, 129-155 (1984)] parametrizes vowel articulation as blends of particular (extreme) articulations (i,u,a). Harshman, Ladefoged, and Goldstein Acoust. Soc. Am. 62, 693-707 (1977)] parametrize English vowels using front-raising and back-raising factors, each of which generates a family of tongue positions. A multispeaker, cross-linguistic database of x-ray tracings [M.T.T. Jackson, UCLA Working Papers in Phonetics 71, 1–232 (1988)] allows quantitative comparison of these proposals. Each proposal defines a vowel space: The coordinates of vowels in the vowel space can be fixed, and the articulatory correlates of the parameters using multivariate regression can be estimated. Each model can, therefore, be evaluated by standard methods, e.g., χ2 a tests. A number of such articulatory models have been investigated. The results suggest that front-raising and back-raising factors generalize well across languages and fit articulatory data better than height/backness- or particle-based models.