'La voce è mobile'. Anacrusis in Italian opera, rap and pop songs: a revisited model of Italian metrical structures

Introduction. In recent years (Baronian & Royer-Artuso, 2024; Zuraw & Roca, 2024), linguists have turned to musical practices to investigate what the artistic adaptation of a language can tell us about its phonological structure, focusing on textsetting: the system of intuitions operating when a composer assigns notes to words. The aim of this paper is to investigate anacrusis patterns in Italian songs, overcoming the traditional strict binary model of Italian prosodic structures. Particularly, our corpus of Italian opera (Rossini, Verdi, Puccini), pop music (Conte, Battiato, Battisti) and rap music (Caparezza, Frankie hi nrg, Neffa) allows us to span across different genres and musical traditions. The corpus provides a broad diachronic window for outlining different textsetting realisations and highlights a continuum in the music–text relationship (Adams, 2008): from the Monteverdian ideal of music serving the expressive meaning of the text, to contemporary rap practices where rhythmic design precedes and affects textual composition. (Corpus span: 1816–1926 for opera, 1969–1996 for pop and 1993–2004 for rap; see Di Prete, 2021 and the RVP corpus in Turconi, 2022). / Music theory. Following the Generative Theory of Tonal Music (GTTM, Lerdahl & Jackendoff, 1983) and the principle of generative textsetting (Halle & Lerdahl, 1993), we define beat as the periodic, hierarchical metric framework that predicts where stronger and weaker temporal positions occur, and grouping as the way musical events are grouped into nested, meaningful units, determining where phrases begin and end. Crucially, group boundaries need not coincide with metrical strong beats. If a group begins on a beat weaker than the strongest beat in the group (that is, if it begins on an upbeat), then the grouping and metrical structures are out of phase. Given an out-of-phase textsetting, we define anacrusis as the span from the beginning of a group to the strongest beat in the group, and downbeat as the strongest metrical position in a bar (beat 1). Moreover, weak anacrusis textsetting consists of a beat associated with one and only one unstressed syllable (either stray or degenerate foot), whereas strong anacrusis textsetting, consists of an upbeat associated with two or more syllables. / The problem. The traditional metrical representation of the Italian stress pattern is a left-headed, bounded foot (Bafile, 1999). Nevertheless, in penult-stressed words, the first syllable is usually considered stray and directly attached to the Prosodic Word (e.g., “domàni”, [do.(ˈmaː.ni)f]w, where [do] is the stray syllable, F is the foot and W is the prosodic word). Weak constituents followed by trochees are also parsed in the same way (e.g., the noun phrase “la mano”, [(ˈmaː.no)f]w). Scholars have attempted to address the empirical prevalence of initial unstressed syllables in Italian. Different solutions have been proposed across theoretical frameworks, ranging from strict binary approaches (see bimoraic trochaic foot with extrametricality right, Krämer, 2009) to weaker constraints (see Weak Local Parsing, Hayes, 1995). A further alternative is offered by internally layered ternary feet (Martínez-Paricio & Kager, 2015), which introduce an intermediate layer within the foot domain. At the opposite end of the spectrum, Faust & Ulfsbjorninn (2024), within the Strict CV framework, argue for an analysis without feet altogether. In music, when initial unstressed syllables fall at the beginning of the verse, anacrusis textsetting is often implemented to group such syllables into the musical phrase. Thanks to a heavy overlap in structural and cognitive domains (Patel & Daniele, 2003; Turconi, 2021, 2022), both language and music share similar principles for organising their metrical structures into strong and weak elements (i.e., syllables for language and beats for music). Musical meter, moreover, perfectly integrates binary and ternary constituents within its hierarchical time-span structure. Therefore, a systematic overview of how these configurations are set to music in songs can enrich the phonological literature related to this phenomenon. / Our claim. Specifically, we claim that strong anacrusis textsetting – despite including stressed syllables – is preferable within the text-tune composite (Di Prete, 2024), as upbeat strong syllables are governed by a weak node (prosodic word, foot or clitic group) at a higher level of the metrical tree. This may suggest a formal incorporation of the stray pretonic syllables into the word-level metrical algorithm rather than leaving them unparsed. Our corpus of 2392 musical phrases across the three genres (1177 in opera, 49%; 370 in pop, 15%; 836 in rap, 35%) shows that upbeat phrasing is robust and systematically supported in opera, pop and rap (71% upbeat phrasing over 29% downbeat phrasing, χ2(2)=74.8, p