Rhythm Types and the Speech of Working-Class Youth in a Banlieue of Paris 125
19 Beur, -ette (n.m./f.) descendent of immigrants from North Africa, born in France,
20 Certaines voyelles ont tendance à tomber. Mais les consonnes—c’est manifeste
dans les milieux arabes des banlieues—deviennent beaucoup plus explosives. Un
type de prononciation que les rappeurs [... ] ont reprise. Par exemple, au lieu
de “partir”, on dira “ p’r t’r”: les voyelles disparaissent presque totalement. Et les
consonnes explosent, comme dans “Rrrspect!” (respect). C’est l’accent beur. Le
français intègre parfaitement cette inÀ uence nouvelle, comme il a intégré celles
de l’italien, de l’anglais, pour sa plus grande vitalité!
21 The idea of raw spectral measures of sonority was explored recently by Galves
and his colleagues (2002) who found that such measures lead to the same cluster-
ing of rhythmic classes conjectured by Pike, ¿ rst shown by Ramus and his col-
leagues’ 1999 study.
22 White and Mattys (2007) have proposed the VarcoV and VarcoC indexes, accom-
plishing essentially the same task as Grabe et al.’s various PVI indexes. These two
measures were tested on an extended corpus of speech samples recorded in La
Courneuve in Fagyal (in press).
23 AF speakers in this corpus referred to their heritage language as “Arabic,” and
reported to have at least passive knowledge of the language. Although none of
them reported speaking Berber, some speakers might have used “Arabic” as a
unifying label for a language from North Africa (see Fagyal in press).
24 Some of the students might have repeated classes, and did not communicate this
information during or after the interview. Therefore, their exact age might not
always correlate with their grade.
25 The terms active or productive vs. passive or receptive bilingual refer to an active
vs. passive knowledge of languages, following established terminology in studies
of bilingualism (see Romaine 1989).
26 These numbers are based on the 1999 census data, analyzed and publicly available
in the Centre de Documentation of La Courneuve (ORGECO 2001).
27 Headline in The New York Times published on October 16, 2001.
28 Ramus et al. (1999) and Ghazali et al. (2002) measured vocalic and consonantal
intervals in utterances, while Low et al. (2000) did so in intonation phrases. Grabe
and Low (2002) later broke this tradition, and computed rhythm type measure-
ments on the entire paragraph, regardless of the length of prosodic units.
29 Despite the instructions, some of the speakers omitted the title and the last sen-
tence in the text. Tiredness and informality with the ¿ eldworker (speakers were
volunteers and knew the ¿ eldworker well) might be among the reasons why not all
speakers stayed on task until the end of the recording session.
30 The two speakers who omitted the reading of the title and/or the last sentence
came from the AF group.
31 The graphical representation of means and standard errors in the Ramus et al.
(1999: 273) ar ticle, however, suggests relatively nor mal dist ributions, without sub-
stantial skewing to higher or lower %V and ǻC values, close to the distribution of
measurement points for AF and EF speakers.