This paper looks at two recent approaches to accentuation in Ancient Greek, Steriade 1988 and Sauzet 1989. Both Steriade and Sauzet include treatments of enclitic accentuation in Ancient Greek which I will argue need to be revised. Steriade offers a metrical analysis that is consistent with most of the data but theoretically suspect. Sauzet 1989 offers a mixed metrical/autosegmental account that is theoretically more appealing but-fails to account for established generalizations about enclitic accentuation. I will adopt the general framework of Sauzet, which seems to be more in line with normal (non -enclitic) accentuation in Ancient Greek, but revise his analysis of enclitic accent. The result, I hope, will be a more insightful approach to enclitic accent than either Steriade's or Sauzet's. An added bonus of the present analysis is that it uses the same footing procedures that Allen (1973 ) has motivated independently for Ancient Greek primary and secondary stress- -this is true of neither Sauzet's nor Steriade's analyses.

In this paper, I provide an analysis of vowel harmony in Izi, an Igbo language spoken in the East - Central State of Nigeria. Using data from Meier, Meier, and Samuel (1975; hereafter MMB), I argue that harmony in complex verbal structures in Izi is inadequately accounted for within a level ordered model of lexical phonology (Kiparsky 1982, Mohanan 1982, etc...), claiming instead that harmony facts are best accommodated within a non-level ordered approach (cf. Halle and Vergnaud 1987, Halle and Kenstowicz 1991; Halle, Harris, and Vergnaud 1991). In sections 1 and 2, I provide a description of the general pattern of the [ATR]-based vowel harmony system in Izi and motivate [+ATR] as the only value of the feature [ATR] present at the level of underlying representation. In section 3, data are presented demonstrating the inadequacy of a level -ordered treatment of vowel harmony in verbal structures. Finally, in section 4, I propose an alternative, non-level ordered analysis that derives the attested harmony facts via cyclic rule application at a single level. Crucially, particular morphemes in verbal structures are claimed to undergo a pass of the cyclic rules prior to concatenation, a phenomenon which I call selective cyclicity.

It is widely acknowledged that certain feature combinations are more likely to occur than others. For example, the feature of nasality is much more likely to appear on segments that are voiced than on segments that are voiceless (see discussion below). Several properties of such combinatorial restrictions are important, including the following: (i) the motivation or source of such restrictions, (ii) their cross-linguistic variability, (iii) their language -internal strength, (iv) the manners in which they manifest themselves. This paper examines certain aspects of the phonology of nasal segments that bear on these issues. The paper focusses on the phenomenon of nasal opacity, where opacity is used to refer to the arresting of a process of feature propagation. When some feature (in this paper, nasality) is transmitted throughout some domain, the presence of certain opaque segments interrupts such a transmission. It is shown that in a wide range of cases involving nasality, the class of opaque segments is systematically defined. Blocking is not due to the lexical idiosyncracy of particular segments; the class of blockers is defined in terms of particular phonological features. This property raises two important issues. On the one hand, how can the possible classes of blockers be characterised in terms of their feature composition? On the other hand, by what mechanism do the opaque elements actually accomplish blocking. In the following sections, I first discuss certain cross-linguistic generalisations concerning cooccurrence restrictions involving nasality; I go on to demonstrate that the types of cooccurrence restrictions governing segmental inventories also define typical classes of opaque segments; finally, it is demonstrated that the actual mechanism for accomplishing the blocking of feature transmission involves feature cooccurrence restrictions in a central way.

It is argued that echo -words result from the tension between a requirement that penalizes a sequence of two identical stems, *REPEAT(Stem), and one that requires two identical stems, REPEAT(Stem). Based primarily on data from Javanese, I make three main points. First, at least some inputs to the Optimality Grammar must be abstract morphological specifications like PLURAL. They are phonologically incomplete outputs of the morpho-syntax. Second, morpheme realization results from an attempt to meet output targets in the form of constraints: REPEAT, σ₂ =a; PL=s, and so on. Such morphemes do not have underlying forms in the familiar sense (cf Hammond 1995, Russell 1995). Third, the target constraints may be out -ranked by phonological constraints of various kinds, particularly constraints against the repetition of elements, here called *REPEAT. The elements may be phonological (feature, segment) or morphological (affix, stem). These findings support the view of Pierrehumbert (1993a) that identity has broad cognitive roots. The primary data comes from Javanese, but the paper also touches on English and Turkish. Section 1 gives some background on the handling of morphological data in OT. Section 2 discusses identity avoidance in morphology, sets out the basic proposal, and gives sketches of English and Turkish. Section 3 is an extended discussion of Javanese. Section 4 looks at secret languages, and section 5 sums up.

It is well known that there are four classes of Japanese vocabulary with respect to its origin; Yamato vocabulary consists of native morphemes, Sino- Japanese consists of borrowed morphemes from Chinese, Foreign is a loanword from a language other than Chinese, and Mimetic describes sounds or manners. Each of these classes has different phonological properties.1 There are three phenomena with respect to the distribution of voice in Japanese. One of them is that post-nasal obstruents in Yamato vocabulary and Mimetic are mostly voiced while those in Sino-Japanese and Foreign are not. I will mainly focus on this property in this paper. However, I will also discuss the other phenomena, namely the compound voicing alternation (Rendaku) and the restriction of voiced sounds in a morpheme (Lyman's Law). These phenomena typically occur with Yamato vocabulary only. Although the domain of each phenomenon largely overlaps with a certain class of lexical origin, they do not match completely with each other. The purpose of this paper is to account for the distribution of voice in Japanese by establishing a constraint ranking that covers Japanese vocabulary of any origin. The organization of the paper is as follows. In section 2, I will present data and four problems to be solved. General tendency of Yamato vocabulary are summarized in 2.1, and many exceptions to the generalization are presented in 2.2. In section 3, I will give an analysis using a unified ranking rather than different rankings depending on origins of the vocabulary. In section 4, I will present two pieces of evidence --- historical and acquisional---to support my claim that Japanese has only one ranking.