## Solaar pleure carrément?

In my limited experience of French hiphop, I've gotten the impression that it's rhythmically rather "square", in the sense that the syncopations or polyrhythms that are common in the corresponding American genres are relatively rare. As a first tentative step in evaluating this (perhaps quite wrong) idea, I analyzed the word-to-beat alignments of MC Solaar's popular 2001 piece Solaar Pleure. Here's the official video on YouTube:

And there's a set of annotated lyrics at genius.com.

As with music of this kind in general, inspection with an audio editor reveals the start of the background beat sequence (in the recording that I made from the YouTube video), and the (metronomic) interval separating beats. This makes it easy to write a program to create (for example) an Audacity label track that aligns with the beat sequence throughout the piece:

 9.934470 9.934470 1
10.243700 10.243700 2
10.553000 10.553000 3
10.862300 10.862300 4
11.171600 11.171600 5
11.480800 11.480800 6
11.790100 11.790100 7
12.099400 12.099400 8
12.408700 12.408700 9
12.717900 12.717900 10

Then the syllables can be recorded on a second label track. Thus for the first couplet

Fuck la terre, si je meurs voici mon testament:
Déposez des cendres dans la bouche de tous nos opposants
"Fuck the earth, if I die this is my will:
Put ashes in the mouth of all our enemies"

we get

My practice is to mark beat-aligned syllables at exactly the time of the corresponding beat label, and just to place off-beat syllables somewhere in the interval between the times of beat labels. This is relatively fast and is good enough for the purposes exemplified below.

(I'm sure that there are some errors in my annotation, but it should be generally correct. I don't know of any automatic speech alignment software that's good enough to do an accurate job at this task automatically, alas — maybe someday.)

The resulting labels for the whole piece are here. I wrote a script to turn these labels into an html table — here's the first verse (assuming no bugs in the program):

 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + . fuck . la terre . si je meurs voi- -ci mon tes- . -ta- -ment/ dé- -po- -sez des cen- -dres dans la bouche de tous nos op- -po- -ants/ . vi- -rez_à coup de front . kick les faux ils viennent se re- -cueil- -lir . j'veux des fleurs et des gosses et ma mort . serve leur ave- . -nir/ . . peut etre com- -pren- -dront ils le sens . du sa- . -cri- -fice/ la dif- -fe- -rence . en- -tre les va- -leurs et puis l'ar- -ti- . -fice/ je sais qui pleu- -re- -ra et pour- -quoi . vous êtes les bien- . -ve- -nues/ y'au- -ra pas de par- -ve- -nues . jus- -te des gens de la rue/ . la . presse peo- -ple n'au- -ra qu'des smi- -cards et des sans pa- -piers/ . cos- -tumes mal tai- -llés meme si les mecs vou- -laient bien s'ha- -bi- -ller/ . ci git . claude . m' ba- -ra- -li i- -ni- -tiale M . C/ un p'tit qui a vou- -lu qu'la vie d'au- -trui soit comme une po- -é- -sie/ . . et sur- -tout va pas . croire . qu'y -ra dix milles . filles/ . je dis ça pour ma fa- -mille . j'n'é- -tais pas par- -ti en vrille/ . . on me jette de la terre . . on dé- -pose quel- -ques fleurs/ . seul . . . sous son saule pleu- -reur . so- . -laar . pleure/ .

The whole thing in tabular form is here, without the narrow text-gutter contraint of this blog's WordPress theme.

Then another simple script calculates the proportion of positions filled by syllables across the 84 8-beat lines in the piece:

Beat "positions" in this case are just sequential beat numbers mod 8, with the between-beat position between N and N+1 marked with N+, so that there are 16 "positions" per line in the transcript. If we take the audible beats to be quarter notes (at about MM= 194), then each line is two bars of 4/4 time, with a similar profile of syllable-association strength in each bar.

(It's possible that the audible beats should be notated as eighth notes, so that each line is one bar of 4/4 time at 97 quarter notes per minute — but this doesn't really change the analysis.)

A relevant link is Nicholas Temperley and David Temperley, "Stress-meter alignment in French vocal music", JASA 2013 — about which more later.

Does this first look at one performance confirm my overall impression of the genre? Well, the patterns of syllable placement in this piece tend to confirm my subjective impression that its vocal rhythm is pretty much binary, without much of the polyrhythmic equivocation (e.g. between 2+2+2+ and 3+3) that is prominent in American hiphop/rap. But one of my musically-literate French friends warns me that MC Solaar is not typical. And there are a few metrically-interesting lines in this piece as well . . .

So we'll see.

1. ### A Scott said,

January 30, 2016 @ 6:16 pm

Thank you for such an interesting article. French hiphop has its own rules & constraints, I'm sure. Much as UK hiphop has morphed into its own genre/genres through direct influences from African, West Indian & Indian sub-continent languages and speech patterns

2. ### David Lukeš said,

January 31, 2016 @ 5:38 am

This is a really interesting topic to investigate! I'm not sure I understand your conclusion about the predominance of binary rhythm in French hip-hop right, though — are you referring to the presence of two distinct peaks at 3 and 7? Have you tried doing the same analysis on an American hip-hop track? If not, what differences in the general shape of the plot would you expect? Flatter, to reflect the greater variety?

Also, the sense of rhythm comes about by creating local contrasts in duration or emphasis placement, but here, each syllable enters the analysis "in isolation" so to speak — the only thing that is accounted for is its position within the line, information about neighbors is discarded. I remember reading a really interesting paper by Patel et al. which tried to compare rhythm in English and French, both language and music, and they came up with a metric (the normalized pairwise variation index) which tried to capture this local variation which is the basis for rhythm. Here's a link, if you're interested: http://www.nsi.edu/~ani/Patel_Iversen_Rosenberg_2006_JASA.pdf

As the paper mentions, English and French are very different in terms of prosody. I even remember being taught that unlike in English, it doesn't make much sense to consider stress on the word level in French, it's purely an utterance-level phenomenon — it's not part of the lexicon, it manifests itself on the last word of a prosodic grouping. French poetry regularly has one or two stresses per line, at the end of each (half-)line (either at the very end or followed by one unstressed syllable). I guess that makes it difficult to create a strong polyrhythmic contrast with the underlying beat without sounding forced and unnatural, since there's no "built-in" word-level stress to exploit…

3. ### David Huggins Daines said,

February 3, 2016 @ 11:44 pm

Your impressions about the generally straightforward vocal rhythm of French hip-hop are not incorrect, but it does depend on the artist. Like French "chanson", French hip-hop is traditionally very lyrically focused, and the rapper's skill is more about rhyme, vocabulary, and rapid-fire delivery than rhythm – the boundaries between rap and slam in French can seem a bit fuzzy at times. For an example the track "Demain c'est loin" by IAM: http://genius.com/Iam-demain-cest-loin-lyrics

The lack of strong word stress in (metropolitain) French probably has a lot to do with this. It would be interesting to make a comparison with Québec hip-hop, because vernacular Canadian French has a very different prosodic stress pattern.