At the outset I wish to mention, for general orientation, four versification systems that had a major impact on Western literatures: (1) Syllabic meter (predominant, e.g., in French and some other Romance languages) specifies the number of syllables in a verse line. (2) Tonic (accentual) meter (said to be predominant, e.g., in Biblical Hebrew verse) specifies the number of stresses in a verse line. (3) Quantitative meter (in classical Greek and Latin poetry, as well as in Mediaeval Arab and Hebrew poetry) specifies the order of longer and shorter events in a metric unit (foot), and the number and sequence of feet in a verse line. (4) Syllabotonic (syllabic accentual) meter (dominant in English poetry from, roughly, Chaucer to Yeats, and in some other modern languages, as in Russian, German, and Hebrew) specifies both the number of syllables in a verse line, and the sequence of stressed and unstressed events in a foot. In fact, it is the number of metrical positions rather than the number of syllables that are specified in syllabotonic verse. The occurrence of certain metres in certain languages may be determined by a wide variety of factors: the constraints of the specific languages, the aesthetic demand for unity and complexity which I assume (though cannot prove) to be universal, and many accidental factors which may be cultural, historical, or I don't know what. Now I cannot tell (though I am sure there are experts who can) why in the first place, e.g., classical Greek and Latin and Mediaeval Arab poetry resorted to quantitative metres. But we can learn very interesting things from conscious efforts to import metric systems from one language to another, and also from the immediate exposure of poets in one language to the poetry in another. English is a stress-timed language, French is syllable-timed. Poets in both languages made efforts to import the quantitative metres from classical Greek and Latin. In French these attempts failed in a very short time, and became mere historical curiosities. French poetry remained with the syllabic versification system, which is congenial to a syllable-timed language. English Renaissance poets thought they succeeded in the adaptation of the quantitative metre. But they were doing something that was very different from what they thought they were doing: working in a stress timed language, they based their metre on the more or less regular alternation of stressed and unstressed syllables, and not as they thought, on the regular alternation of longer and shorter syllables. They used the same names and graphic notation for the various metres, but the system was utterly different, and well- suited to the nature of a stress-timed language. Thanks to this state of affairs, English prosodists in the past few hundred years can spend their time in attempts to discover the intuitive metric rules followed by English poets. The Hungarian attempts to import the quantitative metres from classical Greek and Latin yielded quite different results. In the iambic, which is the metre most tolerant of deviations in all languages I know, there are frequently conflicting prominences: a stressed syllable in a weak position, and a long syllable in a strong position, or the other way around. In the ternary metres (dactyl, anapaest and amphibrach) Hungarian poets are more successful in preserving pure quantitative metre; the trochaic is somewhere in- between, nearer to their practice in the ternary metres. In modern Hebrew poetry the dominant versification system is syllabotonic. Eleventh-Century Hebrew poets in Moslem Spain made a conscious effort to import Arabic quantitative metre, based on a complex system of systematically alternating longer and shorter syllables. They came, however, to the conclusion that such a system would not be suitable to Hebrew grammar; so they decided to found their metre on the system of full vowels ("chords") systematically alternating with units consisting of a vowel + schwa mobile ("pegs").
Syllabic metre would be well-suited to any language I know of, and in quite a few of them there have been relatively short periods in which the syllabic system was prevalent. Downright dominance of the syllabic versification system, however, is surprisingly rare in European languages, mainly in the Romance languages. We may account for the relative frequency of metres other than syllabic by the aesthetic demand for unity and complexity. These two demands control each other; and seem to exert some pressure on versification systems to become more complex than mere syllabic metre. In French, where attempts to accommodate other versification systems failed, complexity was imposed by means compatible with the syllabic system: in the French alexandrine there is an obligatory caesura after the sixth syllable, and there is a systematic alteration of "feminine" and "masculine" rhymes. In this latter respect too convention was restricted by the constraints of the French language. In English, German, Hebrew and Hungarian, and some other languages, in some words the main stress is on their last syllable, in some words on their last-but-one syllable. In French, by contrast, there are no such contrasts of stress. The only way in which French rhymes can be systematically varied is to contrast words which end with an e muet ("feminine" rhyme), and those which don' t ("masculine" rhyme). When, in Dante' s time, the "pegs-and-chords" metre was exported from Spain to Italy, the Hebrew poets were exposed to Italian syllabic metre; and their "pegs-and-chords" metre began to resemble, more and more, the Italian syllabic metre. Now what may appear to be quite surprising is that this new metre gradually assumed the characteristics of the syllabotonic iambic. It was the great Hebrew poet and scholar, the late Dan Pagis, who demonstrated in great detail this development in Hebrew versification in Italy. Obviously, the iambic conformed better with the constraints of the Hebrew language, than with those of the Italian; and thus, the Hebrew versification system (but not the Italian) gradually yielded to the pressure of the aesthetic demands of unity and complexity. So far I have accounted for the differences between metrical systems ("cultural artifacts") by the interaction of three factors: (1) the attempt to import foreign metric models of authoritative status (in the case of English, French and Hungarian poetry, classical Greek and Latin models; in the case of Mediaeval Hebrew poetry, classical Arab models); (2) the constraints of the importing languages; (3) the pressure of the aesthetic demands of unity and complexity. From this interaction resulted the culture-specific and language-specific metric conventions in which the intercultural principles have been individuated. These intercultural principles, in turn, took, in the process of repeated social transmission, forms which have a good fit to the natural capacities of the human brain. The two most conspicuous relevant "natural capacities of the human brain" are the limitations and capacities of short-term memory, and the Gestalt rules of perception. Verse lines can be perceived as perceptual wholes if they can be contained in short-term memory, which functions in the acoustic mode like an echo box, and the number of units it can hold is limited, according to George Miller, at "the magical number seven, plus or minus two." The Gestalt rules of perception refer to the conditions that maximize our tendency to perceive a stimulus pattern as an integrated whole. The "better," or "simpler," the Gestalt of a stimulus pattern, the less mental processing space it occupies and is, therefore, more likely to be contained within the scope of short term memory and perceived as rhythmical. At the same time, according to Gestalt Theory, greater simplicity of the parts impairs the integrity of the whole, and the simplicity of, e.g., the metric feet must be modified so as to make them dependent on, and integrated with, e.g., the hemistich which is the perceptual whole.
In Mediaeval Hebrew poetry there is a huge number of rather complex classical metres. Some present-day scholars are inclined to praise certain poets for having recourse to a big number of metric structures. Consulting the tables of metres in academic editions of the corpus, I have shifted the focus of interest: I have checked the relative frequency of each metric structure, and found surprising results. While traditional scholars treated the various metres as equal, I have found that some metric structures are more equal than others. The index of metres in Dov Yarden's edition of Ibn Gabirol's Secular Poetry reveals, for instance, that the most common metre in this volume ("hamerube") occurs in 139 poems. To this, one may add a related structure that occurs in 24 poems. The second most common metre ("hashalem") occurs in 29 poems. The distribution of the other metres in this volume is between 1 to 12 poems. This distribution is fairly typical of the whole corpus. It has no trace in the explicit poetics of the period, and one must assume that this overwhelming preference was entirely intuitive.
In a paper published about ten years ago I pointed out that the two most frequent versification structures are those that strike the best balance between regularly recurring near-symmetric feet, and the modification of the simplicity of the last foot, generating rhythmically recurring units on the one hand, and integrating them in a percepual whole on the other. More recently I published a paper with Yehosheva Bentov in which we tried to account for the enormous difference between the relative frequency of the first and second most frequent metric structures. We invoked Woodrow's experiments in the nineteen-twenties, which demonstrated that if the organizing principle in an endless series of tick-tocks is differences of duration, the tick-tocks are grouped into end-accented groups; if differences of amplitude, they are grouped into beginning- accented groups. Pitch differences are neutral in this respect. Woodrow's findings can be explained by the limitations of short-term memory. Short-term memory must preserve the beginning of a message, while new information is still coming in and is being processed. The shorter the unit burdening memory, the greater is the space available for the mental processing of incoming new information. Thus, duration differences organize stimulus patterns into groups that leave their main weight to the end.
The first two most frequent verse structures are based on feet that consist of three long units (vowels) and one short unit (schwa mobile). They differ in the placement of the short unit. In this context, longer duration means greater accent. The Mediaeval Classical Hebrew versification system is based on duration differences. Consequently, one should expect end- accented metric feet to be more natural, and therefore more frequent. Indeed, "hamerube," the most frequent versification structure, is based on a metric foot in which the short unit comes before the first long unit (that is, its weight is in its second half). "Hashalem," the second most frequent versification structure is based on a metric foot in which the short unit comes before the last long unit (that is, its weight is in its first half). In a similar way we have accounted for the relative frequency of the other metres in the corpus. The relative frequency of metres in the corpus is publicly verifiable data that reflect the rhyhmic intuitions of Eleventh-Century poets, writing in a different cultural tradition. With the help of observations based on "the natural constraints of the human brain" we have accounted systematically for those intuitions.
The same battery of cognitive principles may account for another "mystery" as well. In classical Greek and Latin metres, Aristotle and Horace observed that the iambic is more natural than the trochaic. This asymmetry can be accounted for, again, by the principle that these metres are based on differences of duration, and that in the iambic foot the longer syllable comes last (that is, is end-accented), whereas in the trochaic it comes first (that is, is beginning-accented). With reference to the syllabotonic system too, many critics in many languages have claimed that the iambic is more flexible, more tolerant, whereas the trochaic is more rigid, "exerts more relentlessly its will." This observation appears to be inconsistent with the foregoing speculations, since syllabotonic metre is based on differences of stress, and until recently even such outstanding phonologists as Trager and Smith identified stress with amplitude. Consequnetly, one should expect beginning-accented feet to be more natural in this versification system. D. B. Fry's experiments on stress perception, however, have demonstrated that perceived stress is a mixture of pitch, duration and amplitude, in this decreasing order of effectivity. Pitch change, the most effective acoustic cue for stress is, as Woodrow found, neutral regarding accent placement in grouping; and the second most effective cue for stress is duration. Thus, again, one should expect the end-accented iambic to be more natural, more flexible than the beginning-accented trochaic.