ELSEVIER Lingua Lingua 102 ( 1997) 265-290 Problems in constraining High tone spread in Ekegusii’ Lee S. Bickmore Department of Anthropokqp. Social Scienw 263. iJni\~ersity at Albany. Albany, NY 12222, USA Received 15 November 1996; revised version 4 April 1997 Abtract Ekegusii, a Bantu language spoken in southwestern Kenya, exhibits a wide range of stem tonal patters in the finite verbal system. Ekegusii exhibits the bounded spreading of prefixal and lexical High tones and the unbounded spreading of grammatical High tones. Yet each of these apply in certain tonal configurations and not others. Predicting just when a High exhibits bounded spread, unbounded spread, or neither will be the focus of this paper. I sug- gest that accounting for the lack of spread is quite difficult to do under standard generative assumptions on the application of phonological rules. Specifically, I conclude that to account for the attested verbal patterns it becomes necessary to either 1) relax the principle of adja- cency that holds between the trigger and target or 2) allow the Obligatory Contour Principle to block the spread of certain (morphologically defined) High tones, but not others. 1. Introduction This paper examines a wide range of stem tonal patterns attested in the finite ver- bal system of Ekegusii, a Bantu language spoken in southwestern Kenya. Ekegusii exhibits the bounded spreading of prefixal and lexical High tones and the unbounded spreading of grammatical High tones. Yet each of these apply in certain tonal con- figurations and not others. Predicting just when a High exhibits bounded spread, unbounded spread, or neither will be the focus of this paper. I will explore an account of these facts within the autosegmental framework (cf. Goldsmith, 1976) and suggest that accounting for the lack of spread under a derivational account is quite difficult to do under standard generative assumptions on the application of phonological rules. Specifically, I conclude that to account for the attested verbal patterns it becomes necessary to either (1) permit the use of the ‘ellipsis notation’ which violates the principle of adjacency between the trigger and target (cf. Odden, 1994) or (2) allow the Obligatory Contour Principle (cf. Goldsmith, 1976; Odden, I would like to thank my language consultant, Robert Ndemo Achenchi from whom all the Ekegusii data was elicited. Additionally, in preparing this paper I benefited from discussions with David Odden and Chuck Kisseberth. Any errors or inconsistencies are entirely my own. 037%2166/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PII SOO24-384 1(97)0009-O