Re: Musical conlangs

On Wed, Oct 16, 2002 at 11:17:29PM +1000, Adrian Morgan wrote:
> The possibility of a true musical language occured to me a long time
> ago, but I haven't known - and don't know - if it's really possible.
> I can't prove that it isn't, though.
Interesting idea.

> By "true musical language" I mean to imply far more than a code that
> just happens to use notes as phonemes. A language where the word for
> "thief" happens to be mi-fa-re-do, for example, and every word in the
> lexicon has a similar mapping, is not a true musical language in this
> sense. I have in mind a language where morphemes correspond to
> collections of notes defined as obeying certain parameters, rather
> than as following precise sequences.
Very good point. Most people don't have perfect pitch, and therefore
would have trouble knowing the precise pitch of an isolated note. On the
other hand, most people are able to tell the *relative* pitch differences
between a series of notes. I.e., most people can hear the difference
between, say, a perfect 5th and a major 3rd, even if they can't name it.
(However, it becomes less clear when dealing with pitches in the same
class -- not many untrained people would be able to tell the difference
between a minor 2nd and a major 2nd.)

Hence, I propose that such a language would need to be isomorphic w.r.t.
transposition -- i.e., it's not the exact pitches that matter, but the
intervals between them. The exact pitches themselves may vary greatly.
(Even in tonal languages, this is true -- e.g., in Mandarin, you have a
mid-level and low tone, but one person's mid-level may coincide in pitch
with another person's low. It's the context surrounding the pitch, which
determines that it is a low tone; since the higher-pitched person would
pronounce the other tones higher as well.)

> These parameters would be flexible enough that any sentence could
> actually be tuneful, with a skilled player being able to make a more
> tuneful version of exactly the same sentence via ornamentations and so
> on, without changing the meaning, or changing it only in subtle ways
> such as emphasis. A study of music theory would be essential for
> designing such a language.
Interesting idea. I suppose, as a working hypothesis, we can assume that
ordinary people can't distinguish between 4ths and 5ths, major and minor
2nd's, major and minor 3rds, etc. So that leaves us with a few "interval
classes" to work with:

- unison (same pitch as previous)
- maj/min 2nd
- maj/min 3rd
- 4th/5th
- 6ths & 7ths.

I'm not an acoustician, but I suspect that people would have trouble
telling apart octave intervals esp. if in quick succession. Similarly with
compound intervals (larger than octave). So, to be safe, we can stick with
the above 5 interval classes and build the language from there. Given this
constraint, a more musically-inclined speaker/player of this language
would probably be able to produce a beautiful melody out of a given
phrase, by carefully selecting which exact interval to use for each
interval class that occurs in the phrase.

> I've had some ideas for a grammar:
>
> Let's suppose that a "stressed beat" is one in which the left hand
> plays a do-so cord (e.g. E and B if playing in the key of E), that a
> stressed beat is usually the first beat of a bar, but that not every
> first beat of a bar is stressed.
>
> Then perhaps: Two quick identical notes on a stressed beat (OSB)
> indicate the beginning of the subject phrase, two quick notes the
> second a tone above the first OSB indicates the beginning of the verb
> phrase, and two quick notes the second a tone below the first
> indicates the beginning of the object phrase.
[snip]

This could work, but personally, I think explicitly marking grammatical
features this way may not work very well. But of course, I could be wrong.

> Or perhaps: A verb can be made negative by delaying the first stress
> in its object phrase until the the third beat, or using a dummy object
> for object-less verbs.
Now there's a slight problem with this. Just as most people can't tell
exact pitches, most people also can't count beats without a frame of
reference. Take, for example, Beethoven's "Fur Elise": the opening phrase
does not have a prominent stress on the downbeat; so most casual listeners
wouldn't even realize there was a downbeat there. Unless this language has
an unfailing rhythm of stressing every downbeat, it would be quite hard
to notice if a verb begins on a downbeat or an upbeat. Especially if the
previous phrase doesn't end on a downbeat.

> Or perhaps: A return to the key note (do) OSB indicates a pronoun
> similar to "it", referring to the object.
Most people, although they notice key changes, have trouble identifying
the key note. If such exact points of reference are used, the speakers of
this language would need mandatory ear-training. :-)

> These are miscellaneous ideas, and I'm not convinced of the
> possibility of such a project - you'd surely end up with morphemes
> being very long if they're to be flexible. It's just that: if it
> *could* be done, and could be done well, it would be a magnificent
> achievement.
[snip]

Well, after what you said, I *think* it's possible. Here's my idea for
addressing the objections I raised above:

Basically, my observation is that in general, people who aren't musically
trained have difficulty recognizing (1) exact pitches, (2) exact
intervals, (3) which scale degree a given note belongs to. However, people
*do* readily identify (a) rhythms, (b) melodic shapes, (c) approximate
intervals.

So, my idea is that each morpheme in this language would be a musical
"gesture". A gesture is sorta like a motif, that one can readily identify
even if there are variations in it. For example, we hear b-a-g#-a as being
"related" or "similar" to e-d-c#-d, and similar to f-e-d#-e. It can be
transposed to any key, or even slightly embellished, and we can still
easily recognize it. A gesture has a unique rhythm, which can be
immediately recognized. Perhaps variations in the rhythm can be allowed,
perhaps used as a means of inflecting a word, etc..

In a sequence of "gestures", each gesture can be placed in different
relative pitches; for example, gesture A can differ from gesture B by a
4th or 5th. This difference can serve some sort of grammatical meaning;
perhaps we "stress" a gesture by raising its pitch relative to the
gestures surrounding it; or maybe the function of a word can be indicated
by its relative position: subject a 5th above, object a 3rd below, etc.,
etc..

We may say that roughly, a gesture is equal to a "word" in the language.
End of sentence can be a long rest, a comma a short rest, etc..

Now, if you allow harmonic gestures as well, we can even use cadences for
punctuation, and modulations for subordinate phrases, etc., like Matthew
suggested. (Although most people won't be able to tell you what exact
modulation has happened, usually people "sense" that something has
changed.) Again, the key here is (IMHO) not to assign meanings to exact
chords, but to think in terms of "harmonic gestures", which is a sequence
of chords which may be placed at any pitch.

Sequence is the keyword here; those who have studied music probably
realize that, say, a major triad, can sound very different depending on
where it occurs and what surrounds it. For example, a C major chord
following a G major chord sounds like a cadence; but a C major chord
following an A-flat major chord sounds like a modulation to F minor. Or,
to account for larger-scale context, a C major chord occurring in the key
of C has a firm, "home" sound; but a C major chord occurring in the key of
A-flat has a peculiarly minor-key tinge to it, because relative to its
context, it behaves like the dominant of F minor instead of a tonic chord.
The same C major chord occurring in the key of F# major takes on a totally
different flavor, because of its different relationship with its context.

So this conlang can assign meaning to easily-recognized chord sequences,
perhaps marking sentence boundary, perhaps marking word function like case
or tense, etc..

As far as grammar is concerned, if you treat these melodic and harmonic
"gestures" as "morphemes", then the grammar can be completely independent
of the exact gestures used.

What do you guys think of this idea? ;-)


T

--
This sentence is false.