I kill you' b. /ček/ → čéka 'stagger' mayákte 'you kill me' /khuš/ → khúša 'lazy' wičháyakte 'you kill them' /čap/ → čápa 'trot' owíčhayakte 'you kill them there' cf. /kte/ → kté 's/he, it kills'This pattern of stress-epenthesis interaction presents a clear counterexample for the Bottom-Up theory: a-epenthesis is syllabically conditioned; and since the organization of syllables into stress feet proceeds from the bottom-up, the epenthetic syllable should be stressed according to the regular pattern of peninitial stress. But this is not correct for Dakota, which calls into question the explanation that the theory offers for other languages. In order to account for the Dakota pattern, stress assignment must be ordered after a-epenthesis, and once rule ordering is admitted in the theory, the account of the Swahili pattern is no different from the Rule Ordering approach.We are left, it would seem, with some version of the Rule Ordering theory, and there is a reason for rejecting this theory as well. Epenthetic syllables do not always behave in a uniform way in relation to stress. They can be ignored in some environments, and yet incorporated into the stress pattern in others. Stress-epenthesis interaction in the Papuan language Yimas is like this, and as we will see in detail below, such patterns point to a real flaw in the Rule Ordering approach.In Yimas, the main stress regularly falls on the initial syllable of a word (3a). Epenthesis into this position, however, creates exceptions to regular initial stress, pushing stress forward a syllable (3b).