PHONOTACTIC CONSTRAINTS AND WORD DEMARCATION IN ROMANCE 1 João Veloso & Pedro Tiago Martins (Center of Linguistics of the University of Porto, Portugal) Some Basic Assumptions 1 – “Words” are not universal linguistic units; in some languages, they can be conveniently and completely replaced, for the sake of linguistic description, by units such as the “morphemes”. 2 – However, in languages that admit WFRs (inflectional languages, mainly), words do seem to behave as linguistic units/structures. Romance Codas - tendency towards empty codas; - *[ComplexCoda]; - “unmarked” Coda: [+son], [+cor]. (see, e.g., Glessgen 2007: 142) Word-final codas: diachronical evidence of their special status. (see, e.g., Lausberg 1963) But: - some evidence of the existence of word-final codas different from word-medial codas: Peninsular Spanish: admits /d/-filled codas word-finally: «sed » ‘thirst’, «huésped » ‘host’, «césped » ‘grass’, «juventud » ‘youth’. Catalan: admits /d/-filled codas word-finally (even though they are often phonetically deleted). CUNY CONFERENCE ON THE WORD IN PHONOLOGY (January 14 th -15 th , 2010), City University of New York Final Remarks 1 – Some phonotactic constraints can be accepted as word-boundary cues at least in some languages. The Case of European Portuguese European Portuguese phonology: highly restrictive as far as coda-filling is concerned: • filled codas are less frequent than empty codas • no complex codas • only /l/ ([]), // and /S/ are admitted as coda-fillers Declarative Phonology-Based Formalizations of Possible "PTWRB Phenomena" and Other Word-Boundary- Sensitive Phonotactic Constraints in Romance P.1: Prohibition of word-initial //, // and // in European Portuguese [Seg = ( ∨  ∨ )] → ~[# Init ℘Seg] P.2: PTWRB in European Portuguese {Seq = [(n.) ∨ ((k ∨ p)s.) ∨ (VGN) ∨ (VGNS)]} → → [(Seq℘# Fin ) ∧ ( . = # Fin )] S.1: Word-final /d/-codas in Pensinsular Spanish [Seq=(d.)] → [(Seq ℘# Fin ) ∧ ( .=# Fin )] C.1: Word-final /d/-codas in Catalan [Seq=(d.)] → [(Seq ℘# Fin ) ∧ ( .=# Fin )] Key in the handout. Abstract In many languages, some phonological segments or structures are allowed at word boundaries only. The importance of this fact is manifold: it shows that phonology is not “blind” to the word as a linguistic unit; it can explain some aspects of speech processing (how speech continua are split into words); it may help to build software aimed at identifying word boundaries within larger continua. For European Portuguese and for other Romance languages possibly, it is proposed that a “Prosodic Tolerance of the Word Right Boundary” should play an important role as a phonotactic cue for word-endings: these languages are known to be very restrictive as far as the segmental coda- filling is concerned. However, these restrictions are somehow cancelled at word-endings, which admit phonotactic combinations that are not allowed elsewhere. A formalization of this phenomenon, inspired by the logical formalisms of Declarative Phonology, is also proposed here. 1 Part of this research is funded by the Centro de Linguística da Universidade do Porto (Fundação para a Ciência e a Tecnologia, Portugal, FEDER/POCTI, U0022/2003). 3 – Words as linguistic units are ruled by morphological, syntactical, phonological and semantic properties which interact language-specifically. 4– Phonotactic constraints act in many languages as word-boundary cues (Jones 1931; Anderson 1965 ........); this may be properly described by means of logic-based formalisms, as it is proposed by frameworks like Declarative Phonology (Scobbie et al. 1996; Angoujard 2006). 2 – Apart from the descriptive interest of this regularity, its relevance is twofold: (a) it can explain how hearers may identify word- boundaries in speech processing tasks; (b) it may be useful for the development of automatic tools for word demarcation within speech continua. 3 – It must be added, though, that in languages where PTRWB is found, it is a sufficient but unnecessary condition for word boundary identification. 4 – This kind of linguistic data offer us an extra amount of evidence favouring the word as a linguistic domain/unit and the necessity of including phonological aspects among the “wordhood criteria”. 5 – Declarative Phonology-based formalisms offer adequate descriptions of all relevant, surface- observable phonotactic regularities of the lexicon of a given language. References Anderson, J. M. 1965. The demarcative function. Lingua. 13: 185-188. Angoujard, J.-P. 2006. Phonologie Déclarative. Paris: CNRS. Glessgen, M.-D. 2007. Linguistique romane. Domaines et méthodes en linguistique française et romane. Paris: Armand Colin. Jones, D. 1931. The ‘word’ as a phonetic entity. Le Maître Phonétique. 36: 60-65. Lausberg, H. 1963. Romanisches Sprachwissenschaft. Berlin: Walter de Gruyter. Mateus, M. H. M.; Andrade, E. 2000. The Phonology of Portuguese. Oxford: Oxford University Press. Scobbie, J. M.; Coleman, J. S.; Bird, S. 1996. Key Aspects of Declarative Phonology. In: J. Durand & B. Laks (Eds.). Current Trends in Phonology: Models and Methods. Manchester: ESRI/University of Salford, II, 685-709. Veloso, J. (forthcoming). Rimes /VGNS/ en position finale de mot en portugais: une contrainte «sensible au mot». Proceedings of the XXV International Congress of the Société de Linguistique Romane [Innsbruck, 2007]. Veloso, J. 2008. Coda-avoiding: Some Evidence from Portuguese. Romanitas. 3. http://humanidades.uprrp.edu/romanitas. Veloso, J. 2009. Découpage de continuums phonétiques en mots: Critères formels vs. indices substatiels. In : O. Crouzet, A. Tifrit & J.-P. Angoujard (Eds). Proceedings of JEL’2009/JEG’2009. 6.èmes Journées d’Etudes Linguistiques. Nantes: Université de Nantes, 85-90. Goals of This Study 1 – To analyze how phonotactic constraints cue word- boundaries in Romance languages. 2 – To propose some formalizations of the role of phonotactic constraints as cues for word-demarcation in Romance languages. 3 – To underline the theoretical implications of such findings for a correct evaluation of the linguistic units admitted by linguistic descriptions and eventual applications in domains such as automatic processing. Word-Sensitive Phonological Phenomena (some examples) • Vowel Harmony • Stress Assignment • Prohibition of Segment(s)/Clusters in Word-Initial Position • Restriction of Segment(s)/Clusters to Word-Initial Position Only • Prohibition of Segment(s)/Clusters in Word-Final Position • Restriction of Segment(s)/Clusters to word-Final Position Only • Restriction of Occurrence of Certain Segments/Clusters at Word-Endings Examples and sources in the handout. (see, e.g., Mateus & Andrade 2000) But: Word-finally, (1) ‘extra-heavy rhymes’ (2) codas filled by consonants different from /l/, // or /S/. (see, e.g., Veloso 2008, 2009, forthcoming) (1) Extra-heavy rhymes admitted word-finally only in European Portuguese: /VGN/# « pão» ‘bread’ [ ] « ontem » ‘yesterday’ [ ] « homem » ‘man’ [ ] « ruim » ‘bad’ [ ] /VGNS Pl /# « mãos » ‘hands’ [] « irmãos » ‘brothers’ [] «alemães » ‘German [plural]’ [] /VGNS Lex /# « Guimarães » (place-name) [] « Simões » (person-name) [] Unattested in non-final position: *[ .], *[.] (2) Word-final codas filled by consonants different from /l/, // and /S/ Segmental /n/:« gérmen » ‘germ’, []; « plâncton » ‘plankton’, [] Unattested in non-final position: *[k.tu], *[e.] /ks/ and /ps/: « tórax » ‘thorax’ [] « fórceps » ‘forceps’ [] « córtex » ‘cortex’ [] « bíceps » ‘biceps’ [] Unattested in non-final position: *[.], *[.] “PROSODIC TOLERANCE OF THE WORD RIGHT BOUNDARY” (PTWRB) (Veloso forthcoming)