Go to Addendum abstract
Back to home page
Back to "Occasional Papers"

Reuven Tsur, Joseph Glicksohn and Chanita Goodblatt

Perceptual Organization, Absorption and Aesthetic Qualities of Poetry

We report an attempt to obtain empirical evidence for a theoretical analysis of Gestalt-qualities associated with varying rhyme-patterns offered by Tsur (1988). Consider the following three quatrains:

1. Think, in this battered Caravanserai
   Whose Portals are alternate Night and Day,
     How Sultán after Sultán with his Pomp
   Abode his destined Hour, and went his way.

2. Think, in this battered Caravanserai
   Whose Portals are alternate Night and Day,
     How Sultán after Sultán did sojourn,
   And went his way then never to return.

3. Think, in this battered Caravanserai
   Whose Portals are alternate Night and Day,
     How Sultán after Sultán came to stay,
   Abode his destined Hour, and went his way.

Excerpt 1 (aaba) is one of Omar Khayyám's Rubáiyáths, in Edward Fitzgerald's English version, excerpts 2 (aabb) and 3 (aaaa) are versions of it, with its rhyme-pattern manipulated.

The respective perceptual organizations of the three versions can be accounted for in terms of two Gestalt principles, the "Law of Good Continuation", and the "Law of Return". According to the former, "a shape or pattern will, other things being equal, tend to be continued in its initial mode of operation" (Meyer, 1956: 92). The "Law of Return", on the other hand, states "that, other things being equal, it is better to return to any starting point whatsoever than not to return" (Meyer, 1956: 151). These two laws impose different characteristics upon different strophic organizations.

The first couplet of excerpt 1 actualizes the "Law of Good Continuation": the end of line 2 continues the sound pattern initiated at the end of line 1. The rest of the quatrain actualizes the "Law of Return": its third line deviates from the rhyme established in the first two lines; the fourth line returns to it. Excerpt 2 actualizes the "Law of Good Continuation". The first couplet constitutes a "strong" symmetrical shape with a single rhyme, and may be described by a single structural principle (the second line rhymes with the first). The couplet-pattern "perpetuates" itself, recurring in the next two lines as well; and had the poem contained ten, or twenty, or one hundred lines, the same "Good Continuation" would have gone on, indefinitely.

The simplicity of the couplets is considerable in excerpt 2; what is more, they constitute a larger unit, divided into two parts with identical structures; as a result, they clearly stand out in perception as two semi-independent units. This is not so in excerpt 1. Here, the smallest unit that may recur with regularity is the whole stanza. The third line, that is not part of any rhyme, is perceived in the quatrain as one that weakens the simplicity of the parts, and so increases their dependence on the whole. Consequently, there is a feeling that the quatrain constitutes a single unit that is closed with a sharp "click", that is, has strong poetic closure. In excerpt 2, the quatrain tends to separate into two couplets; and if it can be said to close with a "click", it is each one of the two couplets that closes with such a "click". This effect depends on the degree of simplicity of the whole as compared with the degree of simplicity of the parts. Greater simplicity of the whole makes for greater unity. The simpler the parts, when compared to the whole, the more clearly they tend to stand out as independent entities.

When we compare excerpt 3 to 2, we find that the aabb version is symmetrical, and well articulated into two equal parts; it displays good continuation both on the specific and on a more abstract level. The aaaa version has a less good Gestalt: it displays good continuation on the specific (but not on the abstract) level, and can be divided into two identical halves, though this division is projective, and not controlled by poetic structure as in the aabb version.

The source of unity in excerpts 1 and 3 is of different kinds. One may distinguish between them in terms of grouping and differentiation. The unchanging sequence of one rhyme in four or more lines is perceived as a homogeneous sequence, that lacks sufficient differentiation. This kind of unity is relatively unstructured, its effect is mainly through accumulation. The original Rubáiyáth, with the initial symmetrical couplet-pattern, followed by the third, "deviant" line, and the fourth line "returning" to the rhyme established at the beginning, constitutes a coherent structure, imposing unity upon the sequence.

We hypothesized that the aesthetic qualities of the text would be related to the degree of perceived closure. Thus, the more "closed" the text is, it is perceived as more tense, interesting and dynamic. In this respect, closure is a dynamic principle. As Kreitler & Kreitler (1972: 91) suggest, "we may expect that people will sometimes prefer gestalts which are not maximally good and regular, precisely because they arouse tension". These three versions of the Rubáiyáth were employed in two empirical studies investigating the relationship between aesthetic qualities and perceptual organization. We assumed that the perceived effects of poetry are a function of the degree of perceptual organization that is inherent in, or can be imposed on, the poetic text. In addition to poetic structure, the reader has to be receptive to such effects in order to experience them. The trait of absorption which may be defined as the propensity to adopt an experiential set, that is, "a state of receptivity or openness to undergo whatever experiential events, sensory or imaginal, that may occur, with a tendency to dwell on, rather than go beyond, the experiences themselves and the objects they represent" (Tellegen, 1981: 222) should be predictive here.

In the first study that we report (Tsur, Glicksohn & Goodblatt, 1991), a select group of qualified readers was employed to test, the relationships among perceptual organization, absorption and aesthetic qualities. The goal of the second study that we report (Glicksohn, Tsur & Goodblatt, 1989) was to investigate the relationship between aesthetic ratings and perceived degree of closure. We shall first present the results of both of these studies, and then discuss and compare them.

Sudy 1

Subjects Eighteen subjects participated in this study. Nine of the subjects were formally qualified readers of literature, as indicated by their current teaching and literary scholarship, and of these 6 possessed a PhD (including 3 tenured associate and full professors of literature) and 3 an MA. A comparison group of nine formally nonqualified readers of literature was drawn from scholars in the fields of psychology (2), sociology (1), education (1), linguistics (2), history (2) and Biblical studies (1), 5 of whom possessed a PhD (all tenured associate or full professors, in their fields) and 4 an MA,

Questionnaire Each subject was given a questionnaire to complete in his/her own free time. The three versions of the Rubáiyáth were printed on separate pages. The subject was asked to read the texts carefully, and evaluate each version along seven 7-point scales, anchored by the following terms: TENSE~RELAXED, BORING~INERESTING, STATIC~DYNAMIC, UNEMOTIONAL~EMOTIONAL, UNPLEASANT~PLEASANT, COMPLEX~SIMPLE, OPEN~CLOSED. Similar scales have previously been employed in the rating of poetry (Hasenfus, Martindale & Birnbaum, 1983). Order of presentation of both the texts and the scales was counterbalanced across subjects. The subjects were requested to read the questionnaire, and reply to the questions therein in order, and to refrain from "correcting" or referring back to earlier sections of the questionnaire. All subjects then completed the Absorption Scale, taken from the Multidimensional Personality Questionnaire (Tellegen, 1982).

In addition to the grouping factor of formal qualification (group), the subjects were divided into "low" and "high" scorers on the Absorption Scale, by median split. The scales were then submitted to separate three-way analyses of variance (group x absorption x version), with repeated measures for the last factor. A significant (at p < .05 or better, both here and below) group x absorption interaction was found for two scales: boring~interesting and unemotional~emotional. A significant version x absorption interaction was found for the scale open~closed. All other two-way and three-way interactions were nonsignificant. A significant main effect for version was found for the scale TENSE~RELAXED; the respective means were 4.7, 4.6, 3.6, indicating that the aaba version (the original) was viewed as relatively more tense than the other 2 versions (as predicted from the theory). In addition, the main effect for group was significant for a number of scales, but these are not of present interest. Turning to the version x absorption interaction (Figure 1) one notes that relative to the other versions, version aaba was perceived by low-absorption subjects as being relatively closed (M=5.9), whereas the high-absorption subjects viewed it as being relatively open (M=3.0).

Sudy 2

Subjects Our sample consisted of 83 subjects, studying introductory psychology (n = 39), Hebrew literature (n = 14) and English composition or literature (n = 30). The former 53 students related to Hebrew versions of the English texts employed. However, 30% of our subjects did not differentiate among the different versions along the critical scale of OPEN~CLOSED (which tapped the effectiveness of the experimental manipulation). When these subjects are removed, we are left wih n = 54, whose data we report.

Questionnaire Each subject was given a questionnaire to complete in his/her own free time. The three versions of the Rubáiyáth appeared side-by-side on one page. The subject was instructed to read the texts carefully, to compare them one to the other, and then evaluate each version along the 7-point scales. Each set of seven scales was printed immediately beneath each text, and a comparative rating was required along each scale. Order of presentation of both the texts and the scales was counterbalanced. The subject then completed the Absorption scale.

In this study, absorption level (defined by median split) had neither a main effect, nor an interactive effect with language and/or version. Thus, a 2x3 (language by version) analysis of variance (ANOVA), with repeated measures on the last variable, was conducted for each of the seven scales. These analyses were supplemented by trend analyses. The quadratic component of the main effect for Version was found to be significant for the following scales: STATIC~DYNAMIC, OPEN~CLOSED, BORING~INERESTING and TENSE~RELAXED. The linear component of the main effect for version was found to be significant for UNPLEASANT~PLEASANT.

Apart from the scale UNPLEASANT~PLEASANT, the same quadratic trend was found for the other scales (where an effect is found). For all of these, the version aabb is judged to be more active/energetic (i.e., interesting, or tense) than the other two versions. Furthermore, language of the text does not make a difference here. In order to ascertain whether subjects studying literature (n = 20) differed from those not studying literature (n = 34) in their evaluation of the texts, two groups were formed. A series of 2x3 ANOVAs with supplementary trend analysis was conducted, as before. The quadratic component of the version x group interaction was found to be significant for the scales UNEMOTIONAL~EMOTIONAL, andBORING~INERESTING. In addition, all previously found effects were robust. Figure 2 presents the respective means for the two version x group interactions revealed by this series of analyses. As can be seen from figure 2, when group interacts with version, one obtains a more pronounced quadratic trend for students of literature than for the other subjects.

Figure 2. Mean Ratings along Scales as a Function of Group and Version

Overall, then, the version aabb is judged by these subjects to be more dynamic, more closed, more interesting and more tense than the other versions. In addition, for the students of literature, this version is judged to be relatively more emotional. In contrast, when the three versions are judged for pleasantness, the aabb version is rated midway between the version aaaa and aaba.


The results of the two studies significantly differ in two respects. In the first study, but not in the second, we have found a significant absorption x version interaction. And in the first study the aaba version was found to be most tense, whereas in the second study the aabb version was found to be the most active-energetic.

In study 2 we found distinct U-shaped curves relating aesthetic qualities to regularity for 5 scales: STATIC~DYNAMIC, OPEN~CLOSED, BORING~INERESTING, TENSE~RELAXED, UNEMOTIONAL~EMOTIONAL. The more closed the text is perceived to be, the more dynamic, interesting, tense, and emotional it is judged to be. The more open the text is perceived to be, the more static, boring, relaxed and unemotional it is judged to be. Of course, these are all relative judgments, and the size of the effect is low. The trends, however, are consistent, and lend support to the Gestalt-approach to empirical aesthetics that we have adopted here. Closure has an aesthetic effect on these readers, revealed in such dynamic-energetic scales. These relationships are not affected by the language of the text or by the field of study of the reader, at least as far as our student-subjects are concerned. This latter finding is an important one, because it highlights the structural factor of the rhyme-scheme (i.e., Gestalt) in determining aesthetic judgment.

Note that it is the aabb version which is viewed in this study to be more closed than the (original) aaba version, rather than the opposite case, as we had hypothesized. Thus, a major contribution of the second study is in highlighting the aesthetic qualities associated with such a good Gestalt in its literary use. It is symmetrical and is well-articulated into two equal parts; and it displays good continuation, that creates good closure. The aaba version, on the other hand, is perceived by these subjects as a "bad" version of both the aabb and the aaaa ones. Thus, within this intuitive scheme, the aaba version may receive similar ratings to that of the aaaa one. Accordingly, differences in the versions will be eliminated by either leveling or sharpening. Sharpening, according to Arnheim (1957: 57), refers to the "changing ... [of] a figure in which two structural patterns compete for dominance into another that shows clear dominance of one of them". "Leveling" attempts to minimize or even eliminate (under conditions that keep the stimulus control weak enough to leave the observer with a margin of freedom) the unfitting detail. Thus, when the reader notes the deviating "b"-line in the aaba version, its significance may be minimized, in comparison with the aaaa version. The two versions will then receive similar ratings. On the other hand, the weight of this deviant "b"-line may be exaggerated (sharpened).

Thus, the aaba structure is double-edged. Double-edgedness is the phenomenon, that a certain literary device or structure may have different, or even opposite, effects in varying contexts or performances. There is a wide range of double-edged phenomena in literature (cf. Tsur, 1985). The aaba version, for instance, may be perceived as having a weak shape in which two structural principles compete for dominance. Thus it will be rated low on the UNPLEASANT~PLEASANT scale; or there may be a (frustrated) attempt to "level out" the deviating line, in which case the rating will be again low on the same scale; or there may be a (successful) attempt to sharpen the disturbance of the deviating line, followed by a highly gratifying return to the initial rhyme-pattern, in which case the rating of pleasantness will be high. The ratings of closure, too, will tend to be low in the first two cases, and high in the last.

Figure 2 reveals another illuminating finding. On the two dimensions in which the two groups of our student-population differed from one another, the students of literature sharpened the contrast between the preferred version and the rest, whereas the others appear to have attempted to level it. This may indicate that our students of literature did have recourse to sharpening, but instead of using it to "impose" structure upon the aaba form, they used it to "impose" structure upon the set of three versions presented to them. This may explain certain differences between the results of our two experiments. In the second experiment, subjects were explicitly instructed to compare them before rating, whereas in the first one subjects were explicitly requested not to refer back to preceding pages when answering the questions. That is probably the reason why subjects in the second experiment tended to regard the three versions as constituting one Gestalt, and thus certain effects that were conspicuous in the first experiment may have been blocked in the second one.

In our first study we found that for this group of readers absorption interacted with the critical scale of OPEN~CLOSED; low-absorption subjects judged the aaaa and aaba versions to be much more closed than the aabb one, whereas high-absorption subjects judged these versions to be much more open. High-absorption persons are expected to be more tolerant of weak, ambiguous shapes than low-absorption persons. In the aaaa version the stimulus control is weak enough to leave the low-absorption observer with a margin of freedom to impose arbitrary closure on the poem. Whereas in the aaba version there are legitimate cues for sharpening the deviating "b"-line, so as to achieve an exceptionally strong closure.

When we began the evaluation of our findings, we first tended to regard the afore-mentioned experimental results as the failure of some of our respondents to recognize the "true" structure of the poem. In the light of the present analysis, however, we had to restructure the problem, and to regard the weak shape of the aaba version in which two structural principles compete for dominance, as the one that underlies all the responses recorded hitherto; one that (vainly) attempts to level discrepancies and one that (successfully) sharpens them into a good perceptual organization. Paraphrasing Arnheim (1957: 55), the foregoing analysis gives sufficient evidence that the realization of a poem involves the solution of a problem--namely, the creation of an organized whole. "Organized whole", then, is not a given fact, but rather an achievement gained by deploying certain cognitive strategies such as leveling and sharpening. Thus, the cognitive mechanism leveling-and-sharpening appears to be one of our major resources of literary performance, allowing us to account, in a systematic and principled way, for significant differences in literary response.


Arnheim, R. (1957). Art and Visual Perception. London: Faber.

Glicksohn, J., Tsur, R., & Goodblatt C. (1989) Perceptual organization and esthetic qualities of poetry. Submitted for publication.

Hasenfus, N., Martindale, C., Birnbaum D. (1983). Psychological reality and cross-media artistic styles. Journal of Experimental Psychology: Human Perception and Performance. 9 (6): 841-863.

Herrnstein Smith B. (1968). Poetic Closure: A Study of How Poems End. Chicago: Chicago UP.

Kreitler H., & Kreitler S. (1972). Psychology of the Arts. Durham NC: Duke UP.

Meyer, L.B. (1956). Emotion and Emotional Qualities in Music. Chicago: Chicago UP.

Tellegen, A. (1981). Practicing the two disciplines for relaxation and enlightenment: Comment on "Role of the feedback signal in electromyographic biofeedback: the relevance of attention" by Qualls and Sheehan. Journal of Experimental Psychology: General, 110: 217-226.

Tellegen, A. (1982). Multidimensional Personality Questionnaire. University of Minnesota.

Tsur, R. (1985). Contrast, ambiguity, double-edgedness. Poetics Today, 6: 417-445.

Tsur, R. (1988). Hebrew Hypnotic Poetry. Tel Aviv: The Katz Research Institute for Hebrew Literature (in Hebrew).

Tsur, R., Glicksohn, J., & Goodblatt, C. (1991) Gestalt Qualities in Poetry and the Reader's Absorption Style. Journal of Pragmatics 16/4: 193-206.

Published as:
Tsur, R., Glicksohn, J., & Goodblatt, C. (1990) Perceptual Organization, Absorption and Aesthetic Qualities of Poetry. Halász László (ed.), Proceedings of the 11th International Congress on Empirical Aesthetics. Budapest: Institute for Psychology of the Hungarian Academy of Sciences. 301-304.

Go to Addendum abstract
Back to home page
Back to "Occasional Papers"

Original file name: AbsorbLecture - converted on Friday, 16 January 1998, 08:09

This page was created using TextToHTML. TextToHTML is a free software for Macintosh and is (c) 1995,1996 by Kris Coppieters