Final devoicing and vowel lengthening in Friulian: A representational approach

Abstract

This paper proposes an account of final devoicing in Friulian which relies on contrastive feature specification and feature geometry to explicate the connection between final devoicing and vowel lengthening. It is proposed that obstruents which are the outcome of final devoicing are phonologically distinct from true voiceless obstruents, being completely unspecified for laryngeal features. It is argued that the representational deficiency of such delaryngealized obstruents is directly connected to their inability to license a mora, which opens the way to vowel lengthening. More generally, the paper shows how feature geometry may be adapted to capture the effects of contrastive specification and express markedness relations, and proposes a novel approach to hierarchies involving the sonority of coda segments.Keywords: final devoicing, moraic theory, sonority hierarchy, feature geometry, Romance languagesThe present paper has two purposes. Empirically, it focuses on final devoicing in Friulian and on the connection between final devoicing and vowel lengthening. Its aim is to account both for the phonetic phenomena involved in final devoicing and for the fact that stressed vowels are lengthened before devoiced obstruents (but only in a word-final syllable). From a theoretical perspective, the paper takes up several strands of recent research into markedness relations. I argue that feature specifications should be assigned solely on the basis of phenomena attested within Email address: pavel.iosad@uit.no (Pavel Iosad)

Length and voicing in Friulian and Milanese

“…6 Those nominal forms that end in a vowel are stress-penultimate (code 'tail', pale 'shovel', mari 'mother', spazi 'space', Finco 2009, Iosad 2012. Antepenultimate-stress forms with a heavy penultimate syllable are not found in the native vocabulary (màrcule 'somersault' but *màcurle,ùltime 'last one-fem-sg' but *ùtilme, Roseano p.c.).…”

“…This form does not conform to a moraic trochee. However, it must be kept in mind that the unmarked stress pattern in Friulian appears to be just a general tendency in the language: light monosyllables projecting a degenerate foot are permitted as lexical words (fi 'son', di 'day', Finco 2009, Iosad 2012.…”

“…I show that the combination of constraint interaction with serial derivations makes vowel lengthening naturally follow from the need to satisfy the independently-motivated markedness constraint against voiced obstruents in coda position, NoVoicedCoda, and the constraint requiring feet to have two moras, FootBinarityµ. This serial analysis allows for a unitary treatment of devoiced and voiceless obstruents as mora-bearing units, instead of assuming that vowel lengthening is the consequence of whether or not obstruents project a mora depending on their laryngeal specification (Hualde 1990, Iosad 2012. I argue that only a serial account escapes stipulating different moraic representations among devoiced and voiceless obstruents.…”

“…In §4, Milanese is analyzed using Serial Variation (Kimper 2011). In §5, a previous parallel account of the Friulian data is discussed (Iosad 2012). This section also includes an alternative analysis using both Harmonic Serialism and Turbidity Theory.…”

“…This is due to historical compensatory lengthening triggered by consonant deletion in clusters: ["ma:ri] 'mother', ["vo:li] 'eye', ["fra:di] 'brother', ["po:re] 'fear', ["je:re] 'ivy'. All the Friulian data come from Vanelli (1979Vanelli ( , 1986, Hualde (1990), Prieto (1992Prieto ( , 1994, Repetti (1992Repetti ( , 1994, Baroni and Vanelli (2000), Youngblood (2007), Iosad (2012), Finco (2009). Other sources are cited therein.…”