This paper addresses theoretical issues confronting cross-height harmony systems through an experimental study of Kinande, a Bantu language of the Democratic Republic of Congo. Using a combination of acoustic analysis and lingual ultrasound imaging, we evaluate previous proposals concerning the phonetic correlates of the harmonic vowel feature and the transparency of low vowels. Results indicate that (i) although a multivalued scalar acoustic feature in F1/F2 space is not adequate to distinguish all vowel categories in Kinande, the cross-height feature does correlate acoustically with F1, (ii) the cross-height feature of Kinande involves systematic tongue-root articulations and (iii) low vowels in Kinande are not neutral to harmony in the way reported in earlier work, but exhibit significant and systematic tongue-root advancement and retraction according to the dictates of harmony.

In the Yakima dialect of Sahaptin, the smallest forms of words are CCV and CVC, as established by a study of lexical patterns and the obligatory augmentation of subminimal roots. We argue that CCV words cannot be considered a type of disyllable, nor do the initial consonants in CCV words exhibit properties (apart from minimality) that would allow them to be classified as moraic. We therefore propose that Yakima Sahaptin represents a previously unrecognised type of minimality, biconsonantal minimality. This kind of minimality requires a theoretical framework in which consonants can be distinguished from vowels, as do other phonological phenomena.

In Hungarian, stems ending in a back vowel plus one or more neutral vowels show unusual behaviour: for such stems, the otherwise general process of vowel harmony is lexically idiosyncratic. Particular stems can take front suffixes, take back suffixes or vacillate. Yet at a statistical level, the patterning among these stems is lawful: in the aggregate, they obey principles that relate the propensity to take back or front harmony to the height of the rightmost vowel and to the number of neutral vowels. We argue that this patterned statistical variation in the Hungarian lexicon is internalised by native speakers. Our evidence is that they replicate the pattern when they are asked to apply harmony to novel stems in a ‘wug’ test (Berko 1958). Our test results match quantitative data about the Hungarian lexicon, gathered with an automated Web search. We model the speakers' knowledge and intuitions with a grammar based on the dual listing/generation model of Zuraw (2000), then show how the constraint rankings of this grammar can be learned by algorithm.