One question arises immediately. Is this Generative Grammar then applicable to metrical lines? Is it not more likely that the poets just arranged their phrases according to the metre, and did not bother much about the grammaticality of the word order created in this way? The word orders as we find them in lines of verse then should be caused by accident, or their mere phonology. Many scholars are of the opinion that word order and syntactic structure of metrical texts are quite free and arbitrary. Therefore it seems impossible to derive something general from them: e.g. Wackernagel (1968:88) for Indo-Iranian, Canedo (1937:11) for Old-Indian, and Macdonell (1958:283ff) for Vedic. On the other hand, in recent times word order as occurring in the Rigveda is taken as a basis for defining rules on word order for Vedic, and by extension, Sanskrit. See, for example, Holland (1991:22-32), Jamison (1991:40-57), Klein (1991b:66-80).
According to many scholars, metrical phenomena are phenomena of the spoken language. Already Meillet (1923:19) observed that in every language, the discourse naturally tends to a certain rhythm. Metrics only stylizes and normalizes the natural rhythm on which the whole language reposes. This observation leads Lotz (1960:137) to the conclusion that verse cannot be defined as an aspect of art or literature: ..., it is impossible to define verse from any other single point of view, such as art or literature. The observation that poems are based on spoken language appears to be sufficient reason to him to exclude them. He seems to neglect the fact that poetry is certainly not the same as spoken language, just as Lotz (1960:138), too, observed: the language used in verse might differ from the "normal'' use of the language.
An interesting question is the one concerning the difference between prose and poetry. Lotz (1960:137) assumes that the only feature that is missing in prose is the metre. That observation consequently leads to the conclusion that prose is only based on the daily language in a lesser degree. Prose shares only the words with daily language, poetry both words and metre. The metre is related to the language, and according to Olson (1966:15-26 esp.19) depends even on it, since form is never more than an extension of content. His statement disregards the importance form sometimes has, since form can have a considerable influence on the statement, as is evident from jokes, satires and last but not least poetry. The two elements are interwoven, and therefore cannot be divided properly, as if they were independent features, coinciding by accident. Often they are equal attributes to the whole of the utterance.
For instance, in Old-Arabic only little difference in word order can be found between prosaic and poetic texts. Those anomalous orders in poetic works which cannot be found in prosaic texts, are generally caused by rhyme, sometimes by the metre; they are, however, scarce, and do not give the poet's language a principally deviant character, compared with prosaic texts (Bloch 1946:30ff). According to Bloch (1951:214) the Arabic metres seem to have been based directly on normal Arabic discourse sentences.
This compatibility of spoken language with metrical phenomena may be caused by the fact that spontaneous speech (including improvised declamation) takes place in waves (balancements, see Jousse 1925). The length of these waves seldom exceeds 16 syllables and shows a marked preference for roughly half that number, as may be observed in different speech milieus in different languages. The number of 16 syllables appears to be just about the maximum the average speaker is able to produce in one breath under normal circumstances.
Although natural speech is in a way metrical or rhythmical, writers of language are more bound to the metre than the speakers are. The variation in written poems cannot be of the same order as that of the spoken or sung texts. Poetical metre is a fixed stylistic device, while the metricallity of speech is not. The metre of speech is allowed to vary from one sentence into the following, or even within the sentence. As a result, a poet sometimes has to struggle to a certain extent to fix the normal syntactic structures within the framework of the metre. Many an omission (this emerges from passages which occur twice or more) is closely connected to the circumstance that there was simply no room left for it in the stanza (Bloomfield 1916:164). Just as on other occasions repetition and parallel thoughts will have been stimulated by room in the stanza (Gonda 1975e:289).