One of my underlying principles for the performance of much of the repertory I particularly admire is that musical lines should be articulated more or less as speech. This is not to suggest some broader implication about music as a "universal language" with a semantic content, but rather both that singing is the first means of musical expression and that the pronunciation & phrasing of speech provides an insight into how musical lines would have naturally been conceived. The idea certainly has both its strengths and weaknesses. As noted, the form-over-melody ideas of the 18th century tend to restrict the possibilities of phrasing from speech, whereas other systems such as Carnatic music derive all repertory directly from vocal music. So while there are clearly exceptions, and maybe subtle exceptions of which we need to be especially aware, it is my fundamental contention that in many areas individual lines should be articulated as vocal utterances, regardless of instrument. This implication may be reversed, in that those repertories & performances which I prioritize are those which accommodate such a grammar & rhetoric.
This fundamental idea provides a sort of overarching context for melody-determining-form, as indeed intersecting individual lines are my priority, as is careful & precise articulation. However, there are examples of this idea being carried rather farther than seems effective. One specific short-lived example is the measured verse of Antoine de Baïf as set by Claude Le Jeune at the end of the 16th century. In many ways, this is the apotheosis of the four century development of the French chanson, giving way to the operatic and ballet-dance conceptions of the Baroque. It is also a singularly unique idea, that syllable weight should determine rhythmic values. Perhaps this is too strong of an idea in the direction of language-determining-melody-determining-form, but it can certainly be an intriguing one. Otherwise, it is seldom the case that raw phonemes have dictated melodic structure, but rather that melody has acted more as a commentary on the language, perhaps particularly obliquely. Phonemes would then serve to color melodic lines, and finding this sort of internal color is a large part of the art of phrasing and articulation. More vaguely, the idea of language structure providing an impetus for general melodic structure in different cultures is a pregnant one. One can speculate, for instance, that the rough regular phrasing of German contributed to the regular bars of the classical period, or that the intricate grammar and lengthy word constructions of Sanskrit has spurred on the inflections of Indian classical music. Chinese and other languages using pitch movements with semantic content present another intriguing case, and of course it is easy to speculate that this sort of built-in nuance influenced the care with which each note is inflected in the ancient repertory.
Returning to the West, after the 18th century fascination with formalism, the next generations of composers began to acquire a new interest in allowing melodies to more thoroughly influence formal aspects of large-scale pieces. In some ways, this was rather distant from earlier approaches which allowed the contours of the melody to dictate the larger motion of the piece in a straightforward way, such as that cultivated by Obrecht and others. So many elements of form had become entrenched, that it was difficult to innovate melodically from within the tradition, and consequently difficult to allow melodic ideas to expand formally. Of course, the 19th century is also the point at which various surrounding folk melodies were cultivated and used more explicitly, particularly from Eastern Europe. The less typical melodies dictated some different harmonic usage, and played a definite role in the re-expansion of vertical possibilities in the West. This is a phenomenon I would characterize as one of the narrowest forms of fusion. By the time of Wagner, the idea of "never-ending melody" became an important rallying cry, and with it some rather pointed formal implications. This has been one of the central principles of post-Romantic composition, and so one can broadly state that they fall well within the basic concept of "melody determining form." This has also meant something of a bifurcation for the tradition (one of many, of course) in that post-Romantic willingness to let melody determine form is not radical enough for some ideas.
One question underlying this entire discussion remains relatively neglected. What makes something a melody at all? Last time, I wrote about the significance of phrasing in forming context for a melody, and will note now in passing that I reject the idea of melody necessarily linked to harmonic implications. Other periods of Western music, let alone monophonic world traditions, make this point clear. However, there has been an implication all along that there is a tune of some sort, an identifiable sequence one can distinguish and sing as such. In some cases, the piece might be all tune... a monophonic never-ending melody. Usually there are accompanying elements and reprises of some sort. What about music without such tunes? At some point, by suggesting that notes cannot be strung together to form a tune per se, there is an implication of pointillism. One example on this point and how it relates to "melody" is that of the rhythmic solo in Carnatic music. In an extensive elaboration, the vocalist will place a melody in increasingly smaller diminution with ornament, and at some point will leave the process in the hands of the percussionist. At that point, the melody is essentially only rhythm, yet the process has a continuity and the melodic implication remains in the rhythmic units. This is one extreme to pointillism, but approached gradually.
Without a discernible "melody" per se, the relationship between melody and form obviously becomes more difficult to address. One might suggest that the form of a serial composition dictates a lack of melody. In the sense of a "tune" that is basically true, although some serial composers have managed to include some references to tunes and tonality. For the others, it is not a priority and one must then ask whether the pieces can be said to have a disjointed melody in an abstract sense. My preference is to view "melody" more generally than "tune" in this way, and to say that most of the modern works do have melodies even if they are not presented in compact or fully contextualized forms. The scope of the latter is conceived more inclusively and serves partly to broaden the implications of the melody as such. This is one justification for attempting to disassociate melody fragments from narrower contexts such as triad placement. Techniques involving generating movement from interval structure can be seen as an application of melody determining form, rather than the reverse, and so combine with the idea of never-ending development to produce a sort of radical post-Romanticism. The most radical step in the direction of form producing melody would be note-specific chance compositions such as Cage's middle works, where the melody comes note by note from an algorithm. Perhaps this discussion can be continued at a later date.