Wednesday 12 January 2011

Attention, Agility and Poetic Effects

The idea of cognition being a process where an "executive function" determines how attention will be deployed isn't new. Here are two descriptions

  • "Bialystok (1990, 1994), in her classic discussion of cognitive development describes learning in terms of two cognitive processing components - analysis and control. Analysis changes the way knowledge is represented in the mind of the learner. Through the process of analysis, language knowledge changes from implicit knowledge organised at the level of meanings, to explicit knowledge organised at the level of formal or symbolic knowledge. Control involves a development in the learner's ability to selectively focus on relevant and appropriate information. Control, in this sense, means the process of allocating attention to specific representations of knowledge and the ability to move between representations (or particular aspects of these representations) in a manner which allows the fluent completion of the task" (condensed from Hanauer).
  • "Miller and Cohen draw explicitly upon an earlier theory of visual attention which conceptualises perception of visual scenes in terms of competition among multiple representations - such as colors, individuals, or objects. Selective visual attention acts to 'bias' this competition in favour of certain selected features or representations. ... According to Miller and Cohen, this selective attention mechanism is in fact just a special case of cognitive control - one in which the biasing occurs in the sensory domain. ... Within their approach, thus, the term 'cognitive control' is applied to any situation where a biasing signal is used to promote task-appropriate responding, and control thus becomes a crucial component of a wide range of psychological constructs such as selective attention, error monitoring, decision-making, memory inhibition and response inhibition" (from Wikipedia)

Hanauer has applied this theory to literature. As proposed by Graves (1996), the study of expert and novice readers of literature is a useful methodology for investigating the workings of the literary system. Hanauer summarise the empirical studies as follows

  • Experts analyse the literary text on multiple levels and integrate this information into their interpretations; novices relate to the local level of the text.
  • Experts analyse the communicative context of the literary text and the function of various literary patterns within this context; novices follow the narrative and dialogue structure of the literary text.
  • Experts manipulate and focus on specific information in the text in order to produce literary interpretations; novices were very influenced by the local level of the text.
  • Experts can explicitly discuss the role of formal schematic and textual features in the construction of an interpretation; novices paraphrase the meaning of the text.

I'm going to look more closely at the "levels" aspect of these conclusions, incorporating the factor of speed.

Stratified Literary Features

A literary work has many features, some of which might be considered as "layers". For example, Roman Ingarden developed Aristotle's concept that a literary work of art has at least 4 layers, starting with sound, then sense. Perceived properties (like beauty, difficulty, etc) can slide from one level to another. In an experiment by Song and Schwarz where people were shown recipes in different fonts, a recipe in a font that's hard to read was thought to be harder to execute than the same recipe in an easy font. Similarly, speed of reading (controlled by layout) can affect the perceived speed of the narrated events, and a surprising layout can be synchronised with a narrative surprise. I suspect that experts are more aware of the transference of such characteristics. In the "recipe" experiment the effect disappeared if the experimenter apologised to the subject for the readability of the font. An expert reader might not have needed such an apology in order to compensate.

Upper layers can have emergent features absent from lower ones (emotions only exist at higher levels) or can have effects that contradict those of lower layers (a story full of jokes can be sad). Interpretation is not a straightforward progression from lowest to highest level. A higher level interpretation can provoke a re-interpretation of a lower level. Years ago there was a "Truth to Materials" credo in sculpture - a belief that wooden sculpture should exploit knots and the grain rather than try to gloss them over. Many poems seem not to acknowledge the "graininess" of language - the lower levels (the choice of font, color, etc)are soon ignored. Low-level features are more likely to "show through" palimpsestically in poetry than in prose, though many prose examples exists. Some examples:

  1. If in a hand-written letter one reads "this is written with my blood", the physical medium becomes significant again.
  2. When reading handwriting one might reinterpret letters after having failed to deduce a satisfactory meaning.
  3. "Janet likes John" is a simple enough sentence. The interpretation of "Janet likes John" is slightly different.
  4. "love's sore return" - The letters painlessly form into words that in turn combine to form a sentence. When appended by "(4)" and seen as a cryptic crossword clue however, words are broken down into letters, and the meaning of the apostrophe changes.
  5. "Flower: exploding star in retreat" - An imagist poem? No, another clue. An exploding star is a nova, which reversed spells the solution, which is "Avon" - something that flows and hence is a 'flower'. In this case we need to regress back to the "letters" level and re-create an alternative sound and meaning for 'flower'

In the table below I attempt to describe the effects of some layered poetic features on attention. The up/down directions mentioned below pertain to the hierarchy formed by letters, words/sound, localized meaning, and general meaning. The in/out direction is relative to the text. I also consider the narrowness of the attention.

MeterInDown to soundWide
RhymeInDown to soundMedium
AcrosticInDown to lettersMedium
Internal referenceInSame levelNarrow
Intertextual referenceOut (text)Same levelNarrow
NounsOut (World)Up out of languageMedium
Proper nounsOut (World)Up out of languageNarrow

I hope these entries are not far from your subjective impressions. Meter, for example, is a field effect, a wide dissipated awareness of sound emerging from below. End-rhyme has a more local affect. On the other hand if someone called James Lawson was reading a poem that said "James Lawson is a prat", attention would be narrowly focussed away from the poem and away from language. Wherever the attention is dragged, sensations like "difficulty" might be carried along too. Some points to note

  • The fewer "in" effects, the more transparent the text appears to be (the fewer layers it seems to have)
  • Something like a proper noun that can distract attention from the text can be anchored to the text by making it rhyme, etc.

Rather than use the analogy of layers, this situation might be described in terms of processes with feedback.


In this document I'll stick with the "layer" metaphor for the sake of argument (and because it helps with the "palimpsest" metaphor).

Layer confusion and compression

The layers are not always clearly distinguished. For example,

  • Cryptic crossword clues conflate the layers
  • Phrases like "love is a four letter world" make readers change depth
  • Finnegan's Wake exhibits layer instability, as do poems like 'What's in a Homophone' (Josephine Abbott, Staple 25) which begins
    How can I bare it?
    My idle,
    My bridle partner
    Left me last weak
    For a made.
    What a waist
    Of ours
    And ours.

Gérard Genette used the term 'metalepsis' for when boundaries between layers are crossed by characters or other textual elements. For example, in Coleman Dowell's novel "Island People" a low level framed story becomes the top level, taking over the narrative, creating a kind of Mobius band.

Sometimes, especially during reading, the concept of layers is ignored; layers are merged together. This is also done in Cartography, in painting (perspective), and in Photoshop (to save storage space). It's often done to "fix" the interpretation from a particular viewpoint. The disadvantage is that the process is irreversable and makes alternative interpretations difficult. Making the reader do this might be the intent of the author, lulling the reader into a false sense of security.


Within a layer each new element can suggest meanings or eliminate possibilities (for example, consider the effect of each new fact in a whodunit). There can be an association with other phrases in the same layer or with elements in other layers. E.g.

  • 2 ambiguous phrases taken together can produces an unambiguous result.
  • A line-break (a lower layer feature) can change the interpretation of a phrase.

Reading often doesn't proceed linearly through the text. Even in prose there's typically 10% of saccades (eye-movements) backwards. In poetry this percentage is likely to be higher. In particular ambiguity and confusion will provoke backtracking or regression to lower layers. The amount of backtracking needed will influence the reading strategy. Options include

  • serial processing (if readers pick the wrong alternative they backtrack to the last fork - usually on the same layer)
  • parallel processing (keeping all options open - multi-tasking)
  • minimal commitment (choosing an option but accepting that the chosen option is provisional - using peripheral vision)

Chapter-ends and poetic line-breaks tend to force a decision/resolution. The sentence above in the 'layer confusion' section works better with line-breaks

   love is
   a four letter

Rather than just using the emergent meaning one can have a richer reading experience by retaining the lower-level meaning too, leaving that up/down dimension open for traversal. End-rhymes force an awareness of the lower level of sound, though it's only one of many poetic effects that do this. In general, the path through the text is less linear for the poetry reader than for the prose reader. Poetry readers are more likely to bob between layers and move backwards and forwards in the text - they're more agile, and they might read prose in a similarly agile way.

Mental Agility

When it's said that someone lacks mental agility, what is meant? Clearly it's something to do with speed (which hasn't yet been mentioned) as well as movement. I think the movements alluded to fall into 3 main types

  • Moving forwards and backwards in the text
  • Changing between 2 modes of attention (for example, from visual effects to plot-following, or seeing an issue from others' viewpoints)
  • Zooming in/out while staying in the same mode.

Control of this movement is the role of what's often called the "central executive function". A related notion to this movement is multi-tasking (or dual tasking). There's quite a lot of research about this so although it's not strictly relevant, it's worth considering the material because it too involves task-switching. There are situations where we perform multi-tasking - when a task is automated (e.g. driving), or when the tasks can be clumped into a fewer activities (a pianist doesn't have to worry about each finger or hand in isolation) but as Earl Miller has shown, we really only focus on one or two items at a time. Multitasking is essentially delegation or fast task-switching. Autistic people tend to lack this ability. It's harder when the tasks are similar, or use the same part of the brain; easier when the tasks are easily interruptable.

Expert readers have fewer difficulties

  • Expert readers are likely to use more parts of their brain because they look for more features
  • Expert readers have more opportunies to clump, putting less load on working memory.

Plus points

This mixture of tightening and loosening the reader's attention, of turning their attention inward, upward, then outward, of rocking them backwards and forwards through rhyme, can be choreographed by the poet, and presents a challenge to readers with inflexible attention strategies, resulting in simplifications and tunnel vision. In some poems inflexibility isn't important because there's a uniformity of effect. For example

  • With Dada "sound poetry" attention is always "down to sound"
  • A list poem (of regions used in BBC weather forecasts, for example) predominantly uses one effect
  • Confessional poetry has a narrow focus

During reading, some people resist the movement between layers. They become stuck in a layer, trying to read Finnegan's Wake as if it were a poorly spelt straight story. They lose the possibility of seeing how the 2 movements (forwards and backwards; up and down) interact. It's as if an archaeologist ignored the depth at which artifacts were found. For this reason (and because they notice peripheral clues) agile readers might be less prone to whodunnit punchlines.

At other times (e.g when reading an ironic text), not only is flexibility required but speed is too. Using a cinema analogy, what's required is not only a good editor to combine the viewpoints afterwards, but a good director too, otherwise important data may not be collected in the first place. An effective director might perform an initial scan through the modes to see if anything stands out (much as one might scan the horizon or flick through TV channels, or size up a poem) - a shallow, wide search. In a live TV show a director needs to be an editor too. They may have a bank of screens showing the output from various cameras, and like a reader will choose which to prioritise whilst retaining a periperal awareness of happenings on other screens. This is analogous to the situation a poetry listener is in - speed of control and task-switching become important. Once interesting features are identified, attention might be centred on those modes that recognise the features. To be effective, people require

  • quick processing (so the director can decide quickly where to focus attention next)
  • sufficient working memory storage so that a task can be continued from where it was left off when task-switching.

With a good director, the person can still emulate someone without agility (they can narrowly focus on a single task), but they can also rapidly juxtapose 2 viewpoints (hence produce comedy, insight, dramatic irony, etc) producing responses that are hard for non-agile readers to appreciate (like not getting a joke)

Minus Points

Uncontrolled flitting can be a problem. It can look like

  • inappropriate detachment - withdrawal; lack of response
  • multiple personality
  • confusion
  • attention deficiency
  • shallowness
  • dilution of the poem's impact by considering too many types of features.

By breaking "agility" down, problems and diagnoses can be more accurately diagnosed. A fast director (an imprecise one especially) can increase the workload for the "editor" stage. Continuing the film analogy

  • If a good director has few cameras they'll be good at a few tasks, but they'll have blindspots
  • If a good director has bad cameras they'll not miss the obvious (common sense rather than insight)
  • If a bad director has good cameras someone else can use their work. it might be saved by a good editor

What can be taught?

Both speed of task-switching and range of tasks can be increased by training. Readers can practise switching, superimposition, etc against the clock, and they can become more sensitive to issues like Form, Implied Addressee, etc.

According to Hanauer both implicit (theory) and explicit (examples) teaching methods should be used. The former allows the development of individual literary patterns and the latter widens the options of types of literary pattern that can be considered.

An Example

               Going Down
   Cycles cluck past, two boys walk with each girl,
   And upstairs James, not Jim, strums his sitar,
   Making words tadpole in your desklamp's pool,
   Breaking the concentration of the hour.

Rather than analysing a single factor in any depth, an expert reader will briefly consider layout, sound, the title and the first few lines of a poem to see which tools might be most appropriate for the task and genre. This pieces look like a sonnet, so the reader's attention might be attracted towards sound - particularly line-endings. Later there are enough references for the reader to realise the poems about Oxbridge life. Once they've finished a first reading, the title might be revisited ("Going Down" means leaving University). But the title's also a clue that the poem's an acrostic, a realisation that leads the reader to re-visit the lower levels. Readers need to go backwards and forwards as well as up and down. They can't afford to forget about a layer once it's been interpreted.

Schematically (and somewhat fancifully) here's a trajectory through the poem. The reader begins at the title (1) and perhaps guesses a theme, then does a low-level scan of the whole piece (2 to 3) then returns to a fairly high level (4) to read the piece, their attention draw back and down to the sounds of previous end-rhymes. At the end (5) there's another dip into lower levels (6) to read the initial letters of the lines, then finally (7) a high-level conclusion. A similar graph of a prose reader's journey would be a near-horizontal line.


  • Bialystok, E. (1990), "Communication Strategies: A Psychological Analysis of Second-Language Use". Oxford: Basil Blackwell Inc.
  • Bialystok, E. (1994), "Analysis and control in the development of second language proficiency". Studies in Second Language Acquisition 23, 157-68.
  • Graves, B. (1996). "The study of literary expertise as a research strategy". Poetics 23, 385-403.
  • Hanauer, D. (1999), "Attention and Literary Education: A Model of Literary Knowledge Development". Language Awareness, 8 (1), 15-29. (online)
  • Miller, E.K. and Cohen, J.D. (2001), "An integrative theory of prefrontal cortex function". Annual Review of Neuroscience 24, 167-202.
  • Song, H. and Schwarz, N. (2010), "If it's easy to read, it's easy to do, pretty, good, and true", The Psychologist, V23, February 2010

No comments:

Post a Comment