[Smt-talk] semantics

Nicolas Meeùs nicolas.meeus at paris-sorbonne.fr
Wed Apr 8 00:32:19 PDT 2009

mmorse at ca.inter.net a écrit :
> [...] what I recall is that Benveniste struggled mightily with the 
> phonetic size of linguistically meaningful, ie semantic units. The 
> same phoneme or handful of phonemes could amount to a full statement 
> in one context, and an purely dependent part of an utterance in 
> another. Thus part of his problem was or became a quantitive dimension 
> of how sounds or concatenations of sounds become significant, ie how 
> "much" sound is needed for an utterance. That's what I meant by the 
> (admittedly hazy, and I apologize) expression "semantic quanta."
More precisely, Benveniste proposes a distinction between what he calls 
"semiotic meaning", the meaning at the level of elementary signs, i.e. 
the couple Signifier/Signified, and the "semantinc meaning", the meaning 
at the level of phrases, which cannot be described in the same manner. 
Semiotic meaning is stable and can be listed – as is made evident by the 
existence of dictionaries, which are lists of such meanings – while 
semantic meaning is unpredictable and cannot be considered the sum of 
the individual meanings of the semiotic units in the phrase. What 
Benveniste does, in other words, is add a third level of articulation 
after those of Martinet's double articulation: (1) the level of 
phonemes, i.e. individual sounds, devoid of meaning; (2) that of 
morphemes, the "semiotic" units; (3) that of phrases, "semantic" units.
    Benveniste adds the most interesting remark (especially for us) 
that, in music, level (2) of this articulation is lacking and one passes 
without transition from level (1) to level (3). What Benveniste did not 
add, but is pertinent to the present discussion, is that there seems to 
be no upper limit to the ever larger spans of meaningful articulation in 
music, up at least to the level of entire movements. (More about that in 
my paper "Musical articulation", Music Analysis 21/2, 2002.)

This all has to do with the "issue between music and language" that 
Dmitri (rightly) says remains open. It reminds of the matter of general 
semiotics, a few decades ago, when it was claimed that we needed a 
general theory of semiotics and that language would form the necessary 
model for it. Since then, the idea remained that language must form the 
yardstick for the evaluation of any semiotic system [the word "semiotic" 
being here used in a more general sense than in Benveniste]. Dmitri 
proposes a clear instance of this, when he claims that music cannot know 
levels of recursion larger that those of language, with an argumentation 
that boils down to saying that this is so because language is the model 
of all semiotic systems.
    Music, or at least the type(s) of music that we all have in mind 
during this discussion, is a semiotic system, i.e. a system of units 
ruled by a syntax. Of all semiotic systems, music is the closest to 
language because of the temporal dimension of their organization – the 
temporality of visual systems, for instance, does not really compare 
with that of music or language. I do not want to raise once more the 
question whether music is a language; let's merely agree on the fact 
that there exist various semiotic systems, including those of music and 
language, and that it cannot seriously be maintained that any one of 
them necessarily must be the model for all others. In several respects, 
music might be taken as a model for language – especially as soon as one 
is dealing with artistic aspects of language.
    I cannot figure out what Chomsky had in mind when he claimed that 
(the possibility of infinite) recursion is a defining characteristic of 
language (if he, Chomsky, really claimed this). This probably is another 
case of linguistic arrogance, i.e. of the idea that any characteristic 
of semiotic systems at large must have its origin in language. Recursion 
probably is a characteristic of any semiotic system; a most telling 
example of visual recursion is available at 

Even if the possibility of recursion may be a defining characteristic of 
semiotic systems, it does not follow that any semiotic utterance must be 
recursive. Yet, specific types of utterances might need to be recursive. 
What would you think of the following statement (it's just an example): 
"Recursion is a defining characteristic of sonata form"?


Nicolas Meeùs
nicolas.meeus at paris-sorbonne.fr

More information about the Smt-talk mailing list