Poetry, rhythm, and meter
Knowledge and culture subproject 4: "Poetry, rhythm, and meter" of Leiden University Centre for Linguistics
- 2013 - 2017
- NWO Horizon grant
Rhythm is a form of temporal organization that plays a role in any systematic human activity that involves sound. Rhythm is a dimension of musical structure, but it also plays an important role in poetic meter (i.e. the regular alternation of prominent and less prominent syllables), as well as in the opposition between 'syllable-timed' and 'stress-timed' languages. This project investigates the universal aspects of poetic and phonological meter; the relation between rhythm in music and language; and the relation between rhythm, meter, and core knowledge of number.
In music, rhythm is considered one of the dimensions of musical structure, alongside melody and color (Honing et al. 2009; Honing 2009). In natural language, rhythm is a property that distinguishes types of languages (French is ‘syllable-timed’, English ‘stress-timed’). In poetry, the meter of a verse can be described as a regular alternation of prominent and less prominent syllables, where prominence is defined differently in different poetic traditions (in terms of syllable length in Classical Greek or Arabic, in terms of stress in English or Dutch) (Dell & Halle 2005, Fabb & Halle 2008, forthcoming, Hayes 2009, Hayes & Kaun 1996, Halle & Lerdahl 1993, Lerdahl & Jackendoff 1983, Dresher & Friedman 2006).
Metrical poetry is typically assumed to display hierarchical organization: a poem has a number of stanzas, which organize lines, which organize feet, which organize syllables. Classical traditions distinguish three bisyllabic foot types: trochees ( ¯ ˘ ), iambs ( ˘ ¯ ) and spondees ( ¯ ¯ ), alongside a range of ternary foot types: the dactyl ( ¯ ˘ ˘ ), amphibrach ( ˘ ¯ ˘ ) and anapest ( ˘ ˘ ¯ ) as well as a number less familiar bisyllabic and ternary foot types. In most Western cultures the iamb has become the dominant foot type for literary poetry, while folk and nursery rhymes tend to be written in trochees. Ternary foot types, in particular the dactyl, tend to be associated with ‘classicist’ traditions.
What is striking about this inventory is that any metrical line is thus subdivided into a number of smaller subgroups which each has 2 or 3 members. This is reminiscent of the distinction between OTS and ANS as formulated in theories of core knowledge systems: the large numbers are divided into binary or ternary groups. And just like in some experiments on OTS in the number domain it turns out that 4 can marginally still be considered a ‘small’ number, classical metrical theory also assigns a marginal status to feet of four syllables, such as the choriamb; standardized feet of five syllables or more do not exist.
The questions to be addressed in this subproject are the following:
(1) a. What are the invariant, universal aspects of poetic and phonological meter once abstraction is made of language-specific variables?
b. To what extent is meter a rhythmic ability that is properly linguistic or part of a mode fundamental human ability that is shared with music?
c. How can the universal aspects of meter throw light on a possible
cognitive ability of rhythm? Is the connection to the core knowledge system of number, as outlined above, real?
These questions will be tackled in a Postdoc project and a PhD project.
Postdoc project (Teresa Proto): Text setting
For the implementation of this subproject, see Teresa Proto's page.
The problem of textsetting has been described in detail by the musicologists Halle & Lehrdahl (1993), see also Hayes & Kaun 1996, Dell & Halle 2005, Hayes 2005). When people see a new text for a tune they know, they have intuitions about how to align the new text to the known tune. Halle & Lerdahl point out from this that people have a productive textsetting ability for which they have never had any specific training. This is surprising, because textsetting involves the rather complicated task of mapping structures of various sorts onto each other. At least three levels of rhythmic structure are involved: the linguistic level involving the inherent stress pattern of individual words; the metrical level of poetic trochees and iambs; and the musical level involving a system of beats and off-beats. In one way or another, test subjects are able to choose a specific mapping that effortlessly aligns all these systems in the best possible way. It has been observed that people routinely agree on the mappings chosen.
Hayes (2005) gives the following example: "while not every English speaker who knows the song ‘What Shall We Do with a Drunken Sailor?’ knows that ‘Stick on his back a mustard plaster’ is a line of this song, even speakers who don’t know this line will set it on the music in one very specific way:
Although the claims made in the literature are suggestive, the literature on this topic has so far been very sparse. Text setting requires a hierarchical ordering between the three levels. The linguistic level of word stress is often made subordinate to the metrical and musical levels: unstressed syllables may become stressed because the meter requires it. However, although much is known about the internal organization of each of these rhythmic levels, the exact nature of the hierarchy between rhythmic levels is largely unknown.
From the point of view of the main research questions in (2) and (3), this research are is interesting precisely because it involves the interaction of three different levels of rhythmic structure in a highly constrained context. The results of this research will throw light on the question (2), i.e. to what extent meter is a properly linguistic ability or not and (3) which aspects of rhythm should be ascribed to a ‘core knowledge in this domain.
The subproject will address this issue both theoretically and empirically. Empirically, a number of experiments will be conducted in tandem with the PhD student. Subjects will be asked to sing new texts on a number of lullabies and other well-known songs. These data will be taken from at least one Western tradition (probably Dutch) and at least one non-Western tradition (possibly Turkish; it is important that the language in question has at least a stress system and a metrical tradition in poetry). At least one experiment will involve second language learners. The question arises how a native speaker of Turkish having learned Dutch as a second language will set a text to music. Does the speaker resort to the metric system of Dutch or Turkish to complete this task?
Theoretically, the work will be based on the tradition of which the key references have just been mentioned. The focus will be on the interaction between musical, textual and linguistic phrasing, a topic that has not received a lot of attention in the literature so far. Like relative prominence, all three levels also have a concept of ‘phrasing’ or hierarchical grouping. The issue is how aligning the relative types of phrasing, and what the relative weight is that should be assigned to the success of this alignment for the evaluation of text setting as a whole.
PhD project (Varun deCastro-Arrazola): Symmetry and asymmetry in folk poetry
For the implementation of this subproject, see Varun deCastro-Arrazola's page.
In order to answer the questions in (1), this subproject studies the metrical structure of poetry at several different levels in parallel: that of the metrical foot, that of the line of metrical poetry, and that of the stanza (a unit of metrical lines in a poem or a song). The central hypothesis is that ‘natural’ divisions in terms of the OTS/ ANS distinction observed in the core knowledge system for number play a role in this kind of organization.
We will study questions of this type on the basis of ‘folk poetry’. Given that these have gone through a process of oral transmission, they are more likely to have adapted to a ‘natural’ feeling of the appropriate grouping than literary texts. (Hayes & MacEacher 1996, 1998, Kopies & Brink, 1998, Burling 1966). Trochees are the preferred foot type also in the linguistic (non-poetic) structuring of words (English water, father, bottle, warden). Furthermore, we possess several large corpora of folk poetry and folk song which are already tagged for metrical structure. This makes it possible to perform statistic analysis on a large scale. For this project we will use the Liederenbank (Song database, www.liederenbank.nl) of the Meertens Institute, which contains several thousands Dutch song texts from the past centuries.
Kiparsky (2008) is an intriguing study of a number of corpora of American folk poetry. The central organizing principles of organization in this poetry are the same at all levels, such as symmetry and shortest-last. (See also Wilson & Hayes 2008.)
To illustrate, let us take the organization of the stanza. If we denote a line with four feet with the number 4 and a line with three feet with 3, we logically have 16 ways to mix line types in a stanza of four lines:
(2) 3333, 3334, 3343, 3344, 3433, 3434, 3443, 3444, …
In actual practice, we find large asymmetries in the corpus, which can be described by assuming that stanza structure obeys symmetry (they consist of two couplets of two line each, and these mirror each other: 3434, 3333, 4343 are preferred over 3334, 4344, etc.) and shortest-last (a 3 should follow a 4 rather than the other way around: 4443 is preferred over 3334). These principles sometimes conflict with each other (4443 is the favourite stanza shape with respect to shortest-last, but non symmetrical). It can be shown that these principles also play a role internal to the line and even at the level of the foot (Dell & Halle 2005, Fabb & Halle 2008).
We will explore our central hypothesis on the Liederenbank using the work just mentioned as our guideline. Furthermore, we try to see to what extent principles of metrical organization can be reduced to core knowledge systems and/or to linguistic principles, and furthermore to supplement them with a culture-historic background which may make them less 'naïve' to the outside observer. One such point may be that many songs have the same rhythmic structure simply because they used to be sung on the same popular melody º– a practice (contrafact) which was very common until the 19th Century.