I have not yet studied tonal languages, so it might be understandable, but when I listen to Chinese music, for example, I'm unable to perceive tones. This makes me think they are partially or completely dropped in music.
Are tones preserved or simply dropped in music? Do the tones often influence the melody or the other way around? How has this changed through history?
In Cantonese songs, one of two things happen — the tones are preserved in a particular way, or the tones are ignored. The paper Tone and melody in Cantonese by Marjorie K.M. Chan has some information on how often the tones are preserved, showing that the tonal contrasts are largely preserved by the melody in a small sample of 90s HK pop songs. The paper is about the relative pitches of adjacent syllables, but there is a second mechanism for preserving tones in the melody.
In my experience, in Cantonese songs which preserve contour tones, this occurs by applying a grace note to the syllable, so that it starts below the notated tone. You can hear this in, say, 男兒當自強, where the rising contour tone syllables 膽, 打, 似 are sung with a lower grace note. The grace note is on the previous note of the scale in 膽, but a semitone below the notated note on 打.
I don't know of any singer who tries to preserve the Cantonese low-falling tone (tone 4) with an upper grace note, though.
This manner of preserving tonal information can be applied even when the original melody was not written for the lyrics, as in 每天愛你多一些, which was written for a Japanese song.
Off the top of my head, I can't think of a Cantonese song which ignores contour tones altogether, although sometimes you hear Happy Birthday sung with the Mandarin lyrics in their Cantonese readings (祝你生日快樂 etc.), and I think the contours are not generally respected.
Many Mandarin songs, on the other hand, obliterate tone distinctions and don't try to preserve them in the melody. 朋友 by 周華健 is an example of this.
I suspect that the greater number of contour tones in Cantonese is the reason why singers at least try to preserve the contour/level contrast. It would be interesting to get data on whether singers in other tone systems — Thai and Vietnamese, for instance — preserve contour tones or not.
In Vietnamese songs the tones will generally follow the melodies. Of course sometimes there still be some odd, out-of-tone sounds but most of it is a product of an amateur musician.
As André Müller said in this answer:
In my experience, this is different in other tonal languages, pop songs and chansons in Vietnamese have quite clearly audible tones and especially in slow songs this becomes obvious. You still hear the melodie, but you also hear the ups and downs of the single syllables. As if you simply pitch them higher and lower on the musical scale.
I second what André said about Vietnamese songs: the tones need to match the rise and fall of the melodies (there's more leniency than would be the case for spoken Vietnamese, but not much)
Also according to this paper:
A number of studies are converging on the conclusion that in many tone languages, it is important for the musical melody and the linguistically specified pitch sequences to match in certain ways. That is, there are text-setting constraints that limit the possible combinations of text and melody. These have been found in a number of languages, including Cantonese, Shona, Dinka and Vietnamese. In these languages, roughly speaking, it appears that if the linguistically specified pitch goes up from one syllable to the next, then across those two syllables the musical melody must not go down (and vice-versa). Looked at the other way, if the musical melody goes down from one note to the next, the words must be chosen so that the linguistically specified pitch sequence does not go up on the two corresponding syllables – and vice-versa.
There's more interesting information in the paper that you can reference.
In Thai language tone is ALWAYS preserved when singing, some of the most exceptionally composed pieces will even use this to their advantage by applying a a lyric that will in turn producing a pleasing melody which add dimensions to the existing song.
From what I heard, there are about 3 generations of a class of music currently employ in Thai language. The first generation (outside of the traditional classical Thai music which is an entirely different animal) is recently adapt from western music, tone is sometime more or less adapt to the existing melody (but never to discard altogether as it will also make the lyric incomprehensible), second generation make use of lyric which bring about their own melody to the song, and the third generation (current generation) is less concern with lyric-melody harmonic relationship in the second gen and is generally sung independently from lyrical tone.
First generation music https://www.youtube.com/watch?v=vvkI7SC4e9I
Second Generation https://www.youtube.com/watch?v=GfhYatEZP94 (adopt from Tchaikovsky Violin Concerto)
The thing is that two musical systems are different, the Occidental being septatonic and the Oriental, pentatonic. Thus, 'western' music distiguishes seven tones, while 'eastern' music distiguishes five tones.
However, in both musical metanarrations the 4:3 and 3:2 ratios are used for the melos and canon settings. And, yes, there was a time when septatonic scale was also applied in Chinese music.
Here is a link to Chinese poetry transcribed in western alphabet and read by native speakers along with the Chinese music played (sorry, the translations are in Russian only). One can notice that there is a certain match between the tones in Chinese speech and those in the music.
