Reading Is the Means by Which the World Does a Large Part of Its Work

An Introduction to LiveInk

Manual For Scientific Investigators

An Introduction to Live Ink Technology

A Method and Apparatus for Improved Human Reading

US Patents 5,802,533 and 6,279,017 With Additional US and Global Patents Pending

Randall C. Walker, MD Stan D. Walker, MD

Walker Reading Technologies, Inc. 923 Tenth Street SW Rochester, MN 55902

1-507 292-0566 1-877-liveink www.liveink.com

- 1 - An Introduction to LiveInk

“Reading is the means by which the world does a large part of its work. The slightest improvement either in the page or in the method of reading means a great service to the human race.”

E.B. Huey (1908) in The Psychology and Pedagogy of Reading.1

In the century since Dr. Huey, the Harvard pioneer in reading psychology, wrote this statement, a lot has been learned about reading, language, and vision. In the past decade alone, there has been a revolution in the understanding of the brain, and of the visual and linguistic processing entailed in reading. There are now powerful tools, such as functional brain imaging, to advance this understanding even further.

There has also been a revolution in the access to and affordability of computational power and electronic media that can be used to analyze and represent natural language symbols.

And, in the past several decades, reading has become increasingly important economically for the individual and for society.

Our project aspires to that “slightest improvement” which we hope will be of service to today’s readers.

Reading is complex, and has comprised a large part of psychology research. Our project bridges three dimensions of reading described by Legge:

"A critical part of normal reading is the integration of visual data with lexical knowledge and eye-movement control."2

As one reads conventional block text, almost all but the shortest words are directly examined, with the eyes fixating on one or two words spanning about nine characters at a time.3

The size of the small zone of globular visual information just to the right of the fixated word can be used in saccade strategies to skip over smaller words. 4

The shorter the word, the more likely it is to be skipped over. 5

In conventional block text, parafoveal information that is above and below the fixated word does not contribute meaningfully to reading, even though the macula captures this information.6 Instead, this visual information above and below the line of fixation clutters the visual scene, with features that compete for attention.

 This parafoveal spatial information to the right of the fixated word is created by the insertion of spaces between words. These inter-word spaces are taken for granted today, but they were not invented until 900 AD, when scribes realized the spaces enabled them to read text silently. Prior to this invention, all but the most exceptional readers, (Julius Caesar and St. Ambrose, the mentor of St. Augustine) had to read aloud to be able to read at all. Alberto Manguel reviews this in The History of Reading.

Reading is an active visual task, and the sequential movement of eye fixations over lines of text requires an adaptation of the visual search mechanism.7 This adaptation must be learned because the natural use of the visual search mechanism is more random: it is naturally used, for example, when one drops a coin on the floor. Moreover, visual search does not ordinarily keep track of the locations that have been inspected.8

Selective attention mechanisms in visual search create an “attentional spotlight” in which the discrimination of an object is associated with the suppression of processing of previously observed objects. The cluttered visual scene in conventional block text presents many objects that compete for limited attentional resources. For this reason, reading constitutes a very special way in which the attentional spotlighting process is being used, and “may well be one of the most challenging tasks for (the pathways that mediate this process) in modern civilization.”9

The visual search and attentional spotlight mechanisms in reading conventional block text are further challenged by the fact that, for many readers and for many kinds of texts, visual regressions are necessary to understand the text.

Regressions are frequent when reading complex, difficult texts, and are particularly important to re-analyze previously read segments as a reader reconciles new words in the text string with a hypothetical syntactic structure.10 A study of very large English corpora revealed that over half of the sentences people write have complex structures (i.e., not simple right-branching syntax) which require re-analysis by most readers.11

When regressions are necessary, the visual search mechanism (which typically does not keep track of locations that have already been inspected) and the narrowly focused attentional spotlight (which suppresses processing of previously observed objects in order to devote attention to the directly examined object) -- both of which are imposed by the conditions of the crowded visual scene in conventional block text -- become self-defeating for reading. When regressions are necessary, it would instead be preferred to have kept track of where previous inspections occurred, and to have the processing of words that were recently inspected left un-suppressed.

The regressions in reading demonstrate a fundamental difference between writing and speech. Listeners generally do not request speakers to zigzag with multiple repetitions of individually pronounced words (though they may request repetition of sentences). There is additional structure in the speaker-listener system that readers of conventional block text do not have.

Obviously, there are many advantages of writing over speech; moreover, text, when it includes punctuation, is generally an accurate and complete spatial representation of an author’s utterance, and is usually capable of ultimately being fully resuscitated by another reader besides the author. Nevertheless, there are dimensions in speech that appear to enhance the efficiency of communication.

Human Speaker In The Sound Pattern of English, Chomsky and Halle outline Pitch: multiple dimensions in the High prosody of speech: Stress, Pitch (with High, Low, Elevated, Stress Low Rising, Falling, and Concave Elevated Length variations), and Length.12 Rising Rhythm Falling An obvious function of prosody Cadence Concave is to denote segments that correspond to syntactic units. However, the dimensions of prosody are not merely single- purpose and concatenated-- Human Listener rather, they can combine and commute in musical ways to enhance speech. For example, the pronunciation of a phrase can be at once low, and elongated, and concave, and this profile, in turn, can be embedded in a larger pattern that is created with multiple phrases with varying profiles.

There may be some performance limitations in how well prosodic structure can represent syntactic structure. For example, “This is the cat that ate the rat that ate the cheese.” is cited as a case where several embedded clauses are often inappropriately pronounced as three separate sentences. Moreover, not all prosody denotes syntax, nor is all syntax represented in prosody.

However, the full complexity of prosodic structure and its impact on the efficiency of listening is only now being appreciated. For example, with the following two sentences:

The city council argued the mayor’s position forcefully.

(1) The city council argued the mayor’s position was incorrect.

--listeners can accurately predict, based on prosodic information heard through “position”, the syntactic category of the phrase that is about to be spoken.13 Moreover, even a minor increase in the stress of pronunciation of a word will have a cataphorical impact on the processing of that word.14

If listeners carefully listen to a knowledgeable speaker who is speaking spontaneously but eloquently about a complex subject, they will appreciate a prosodic structure that is much richer than simple segment-to-segment boundaries. Instead, if they have a musical ear, listeners will discern a subtle tonal fabric, with each segment having several prosodic attributes that combine and commute across the sentence as a whole to denote the relationships of each segment to all of the others.15

We hypothesize a principle of syntactic harmonics in prosody, which contributes to the greater communication efficiency of listening over reading, by generating a structure for the dynamic, cataphorical modulation of the listener’s linguistic processing. Recent research from the Max Planck Institute supports this hypothesis.16

In summary, for complex and abstract writing, which typically requires frequent regressions to be understood, the crowded visual scene of conventional block text is self-defeating. Moreover, the structure of conventional block text is relatively impoverished compared to the prosodic structure of speech, making reading much less efficient than listening--even if an equivalent level of comprehension is eventually possible with reading. Finally, conventional block text is quite limited compared to the powerful visual processing capabilities that we use all the time to build visual representations of the world around us.

One of the principles of Visual Grammar is that different processing methods are automatically adopted depending on the attributes of the scene observed, and the context and conditions of the perception.17 We hypothesize that it is possible, by redesigning the visual structure of text, to enlist perceptual processes that may be more favorable to

reading and re-examining complex sentences, and to building the conceptual structures that language represents.

We have patented and developed the Text Processor, a Method and Apparatus for Improved Human Reading. (US Patent No. 5,802,533 with additional US patents and global patents pending, copy enclosed).

A current embodiment of the invention is a software engine (the LiveInk® Text Enhancement Engine) that executes multi-tiered, recursive algorithms to transform any English text in real-time for improved human reading; over 1 million computer calculations for a typical sentence are performed.

For example, the current engine transforms this18:

How Ann Salisbury can claim that Pam Dawbee’s anger at not receiving her fair share of acclaim for Mork and Mindy’s success derives from a fragile ego escapes me.

Into this:

How Ann Salisbury can claim that Pam Dawbee’s anger at not receiving her fair share of acclaim for Mork and Mindy’s success derives from a fragile ego escapes me.

40 Independent studies19, in which normal, college-level readers read 35 passages from the Federal Register 30 25 that were transformed in real-time Conventional 20 by the Live Ink engine, found an Live Ink 15 10 5 0 Average Reading Comprehension Test Score (percent correct,s.e.) for Each Tex t Presentation Method

immediate and significant increase in reading comprehension compared to conventional text.

Moreover, it appears some readers may benefit more than others from this transformed text presentation. Poorer readers, who were otherwise intellectually equal to the better readers, appeared to benefit more than the better readers from Live Ink text. Reading time was not significantly different in these studies.

0 Poorest Lower Upper Best Readers Readers Middle Middle

Studies are now planned in: younger readers; college students with reading disabilities; and readers of English as a Second Language.

We are also very interested in more basic research in the mechanisms of reading Live Ink text. This method appears to impact on each of the 3 dimensions of reading: linguistic processing; eye-movement control; and visual perception.

In linguistic processing, evidence supporting the impact of this method on syntactic processing includes that fact that earlier

versions of our software, which employed common function words but which had minimal intelligence for verb phrases, had very little impact on reading comprehension. In other words, simply fragmenting the text into smaller visual bundles had no effect on comprehension. We found that what was needed was a more detailed, and fairly accurate extraction and representation of syntactic units (particularly verb phrases), and their relationships to each other, in order to obtain the increase in comprehension.

The Live Ink text structure is multi-layered and hierarchical, thus differing from mere “chunking” or segmentation of text into concatenated “thought units”. Moreover, the Live Ink text uses segmentation and relative displacement of segments to depict sentence structure transparently, rather than using annotative marks and diagrams.

As the above sample illustrates, the current engine does not (yet) generate a precise or entirely accurate syntactic structure. Nevertheless, this intermediate “protosyntax” structure may be an aid to human reading--irrespective of the innate mechanisms by which syntactic building occurs in reading.20

In other words, transforming this:

Four score and seven years ago, our fathers brought forth on this continent a new nation, conceived in liberty and dedicated to the proposition that all men are created equal.

Into this:

Four score and seven years ago, our fathers brought forth on this continent a new nation, conceived in liberty and dedicated to the proposition that all men

are created equal.

May assist in achieving this21:

brought forth a new nation

Four score conceived in liberty and seven our years ago and dedicated to fathers the proposition on this that continent all men are created equal

Prose Paint: A method and apparatus to generate and depict syntactic relations using commutable chromatic vectors for improved human reading. Protected under US Pat. 5,802,533 and patents pending. In this diagram, Bluing vectors denote entities, Reddening vectors denote actions/verbs, and Yellowing vectors denote modification elements , whose attributes commute through to subordinate elements in the nested syntactic structure shown.

Recent functional MRI brain-imaging studies of sentence comprehension demonstrate a heightened requirement for dynamic collaboration between lexical and syntactic processing when words are infrequent and when syntax is more complex.22

The researchers concluded that these lexical and grammatical processes are “so highly interactive that altering the demands placed on one process indirectly alters the demands placed on the other. Specifically, lexical processes may feed into and affect the

thematic/syntactic processes and vice versa.”

A visual syntactic context may therefore enhance word decoding and free-up working memory for other higher-order reading tasks.23 24 25 The visual protosyntactic cues of Live Ink text may also improve the efficiency of reading, in a way that is functionally similar to the cataphorical cues of aural prosody in speech.

Instead of a self-defeating cycle as seen in conventional block text, the spatial-linguistic cues in LiveInk text could create a self- reinforcing result. It has been observed, for example, that the perceptual span in conventional text is variable and attentionally constrained, with less parafoveal information acquired with greater syntactic complexity.26 Therefore, reducing the “clutter” of the visual scene of the text, and replacing it with spatial-linguistic cues, may expand the visual eye span and prolong the attentional spotlight.

In addition, the differentiated spatial appearance of a sentence--the contour of the sentence as a whole and the shapes it contains--may assist in directing saccades.27 This shape-directed mechanism of eye-movements, in turn, may permit a modified visual search mechanism that includes some visual scanning attributes.

Four score and seven years ago, our fathers brought forth on this continent a new nation, conceived in liberty and dedicated to the proposition that all men are created equal.

In other words, with: a) a modified visual search with scanning attributes, b) a widened perceptual span, and c) a longer-lasting attentional spotlight--the reader could better discern the varied displacement of sentence segments in LiveInk text, in parallel with the word-by-word reading of the segments. The effect would be similar to, but more powerful than, the current use of parafoveal information just to the right of the fixated word in conventional block format. In this way, regressions and re-examination across a wider span of previously read words and phrases may be possible.

Because each of the fixations over a Live Ink sentence has a coherent, visual-meaning relationship to the others, it is possible that an entire “sentence image” may emerge in visual representation. There is recent evidence that synaptic integration of the information in the visual cortex occurs for 20, not just 3, as was previously thought.28

It is this larger sentence image, built up from integrate-able fixations, each bound to the other with larger perceptual eyespans and longer attentional spotlights, which in turn permits sentence examination and a new dynamic in linguistic processing, even without visually perfect syntax.

A 1 2 3 How Ann Salisbury can claim The pre-reading spatial information that Pam Dawbee's 4 anger 5 (fixations A,B,C) will have created a at not “sentence map” that helps the reader direct B receiving her fair the eyes down from “anger” (fixation 5) share of acclaim to “derives from” (fixation 6), even before reading the subordinate, nested segments 6 for Mork and Mindy's success immediately following “anger”. derives from a fragile ego C escapes me.

Moreover, we are currently encoding algorithms that will enable our next generation LiveInk engine to create this:

How Ann Salisbury can claim that Pam Dawbee's anger at not receiving her fair share of acclaim for Mork and Mindy's success derives from a fragile ego escapes me.

Because LiveInk text is a new paradigm in reading, there is very little literature that relates directly to it. In Language & Space, Jackendoff outlines in “The Architecture of the Linguistic-Spatial

Interface” how the experience of space ultimately enters a conceptual realm, and it is from this conceptual realm that linguistic representations of the spatial concepts emerge.29 However, writing itself is a spatial system of language representation--but the topic of writing and reading is not reviewed in Language & Space.

Aristotle observed, in On Interpretation, that:

“Spoken words are the symbols of mental experience, and written words are the symbols of spoken words.”

We propose, because language (and the conceptual structure it represents) is multi-layered and multi-dimensional, that text--the spatial system that represents spoken language--also be multi-layered and multi-dimensional.

In other words, space and language can both be used to represent conceptual structure, and can be more than interfaced--they can be joined in the same medium. The relationship of language to space may reside at a deeper level of information processing:

The known processes of computational vision are capable of recovering the syntactic structure of visual scenes. Thus, visual perception is capable of providing a repertoire of cognitive processes instantiating precisely abstract (catastrophe-theoretic) structures in terms of which the structure of sentences is understood. 30

With more powerful extraction techniques and appropriate structural mapping across visual dimensions, Live Ink technology aims to

more fully exploit the capacity of human visual perception for multi- dimensional reading.

Individual Differences.

That some readers would benefit more than others from Live Ink text is compatible with the current understanding of reading as an unnatural, multi-factorial process. It is possible that Live Ink text could ameliorate some of the innate difficulties that some readers have with conventional text. There is evidence that reading ability varies widely across the population over and above the variation that would be attributable to differences in intelligence alone.31 Some of this variance is due to differences in linguistic processing, with word decoding being the most significant factor.32 Live Ink text could help these readers by enhancing the representation of the phrase context in which words are decoded.

Live Ink text could also help those readers who appear to perform word decoding in working memory, rather than with more automated processes in the occipital, parietal, and temporal cortex. 33

As Pinker observes in How the Mind Works:

“the frontal lobes, which house the circuitry for decision making, are not directly connected to the brain areas that process raw sensory input (the mosaic of edges and colors and the ribbon of changing pitches). Instead, most of their input fibers carry what neuroscientists call “highly processed” input coming from regions one or more steps

downstream from the first sensory areas. The input consists of code for objects, words, and other complex concepts.”34

By partially parsing the text (a process which must otherwise be performed in working memory)--and creating “highly processed input” that visually represents the “complex concepts” of syntactic relationships--Live Ink text may free-up working memory resources for both word-decoding and other, higher-level reading tasks for these poorer readers.

Some of the individual variation in reading also appears to be due to differences in visual motion processing. In reading, this motion processing entails gating mechanisms that are required to transform the continual flow of retinal data into an image of stillness. There are both psychological and functional brain imaging studies demonstrating that differences in processing visual motion correlate with differences in reading ability--across a wide spectrum of reading abilities, and not just among the poorest readers.35

Better Reader Poorer Reader

Live Ink text may help readers with impaired visual motion processing by reducing the number of regressions in reading, and making any requisite regressions easier to perform.

Examples of Future Research.

Practice and Accrued Benefits. Research to date has involved naïve subjects, who had not previously encountered this varied text presentation method. It is possible that with increasing use, readers can learn better how to recognize and integrate the shape information with linguistic content.36

Metalinguistic awareness. In addition, continued practice with LiveInk text may increase “metalinguistic awareness” -- specifically syntactic awareness -- in certain groups of readers. Among younger readers, such syntactic awareness has been shown to predict a student’s reading comprehension skills several years later.37 Among adolescent and adult readers, there is evidence that syntactic knowledge differentiates better from poorer readers.38 39 Previous work has also been done in which modifying texts, by phrase-cueing, and by simplification of syntactic structure, has, for some students and certain types of reading material, improved reading proficiency and performance.40 41

Reading in a Second Language. There is a growing number of people who need to learn, work, and acquire information through reading in a second language. In Europe, where many persons can read English as a second language, as much as 80% of Internet text content is presented in English. There is evidence that differences in syntactic structure between native and second languages contribute significantly to a person’s reading performance in the second language.42

Eyestrain. Surveys of subjects in LiveInk studies have 45 revealed that the frequency of 40 computer-reading eyestrain 35 (after 1 hour of on-line 30 reading) is substantially Conventional Text 25 reduced with LiveInk 20 Live Ink® Text compared to block text—from Presentation 41 to 11%. 15 10 Computer-reading eyestrain, 5 affects up to 80% of the 70 0 million US workers who spend Percent of subjects experiencing 2 or more hours a day working eyestrain from reading text in either at a computer, and has now text presentation method. become a work-related symptom that is monitored by OSHA.43 44 Research with LiveInk text could examine the relationship between eye-movements, fixation

times, linguistic processing, and objective markers for eyestrain symptoms (e.g., accommodation or blink rates), in order to characterize this benefit from LiveInk. Reading Span, Reading Practice, and Learning. A more durable spatial representation of the sentence image in LiveInk may permit more readers to maintain textual information in short-term memory and attention -- that is, it may help increase a person’s reading span.45 This increase in the reading span may assist the higher order processes that occur in reading. These higher-order processes are more abstract and conceptual than the actual surface form of the sentence, and include: relating text content to the reader’s world knowledge; building inferences that are not explicitly stated in the text; and building a coherent and enduring representation of what is understood in long-term memory. 46 A greater reading span (in short-term memory and attention) from LiveInk may enable readers to generate elaborations that are at once useful but not overly restrictive of their interpretation of the text.

Reading practice improves reading skills, but the opportunity for practice is limited by the ability to read. 47 This paradox is a major contributor to individual differences in reading, and is called the “Matthew effect”, from the biblical reference: “For unto every one that hath shall be given, and he shall have abundance; but from him that hath not shall be taken away even that which he hath.” Matthew 25: 29). LiveInk may help increase reading practice by enabling those who are poorer readers, and those who experience eyestrain from extended computer reading, to read more and to read for longer periods. In this way, LiveInk could ameliorate the “rich get richer” paradox of reading skills.

The principles of attribute extraction and varied presentation can be similarly applied to higher order discourse, rhetorical, and conceptual structures in text documents. These principles of document design can be integrated with and build upon the presentation dimensions that are used by LiveInk at the sentence level. For example, because LiveInk cascaded sentences use some of the dimensions that conventional text uses to denote paragraphs,

LiveInk presentation uses varying background color to denote changes in paragraphs.

Additional integrated dimensions can similarly be generated to denote the rhetorical structures that are otherwise embedded in the text string. For example, attributes such as “for example”, “moreover”, or “on the other hand”, can be used to vary the patterns of horizontal and vertical boundaries between sequential panels of text, or the direction of movement of text panels as they enter the reading field on the display. By enhancing the spatial representation of concepts that are embedded in the language of the linear text string, LiveInk technology may assist learning for those who learn better through spatial relations that through written language.48

In conclusion, over the centuries, the medium of text has been developed and used by a minority of self-selected users who were good readers, and until only recently, this was acceptable:

“Mass literacy is a relatively new social goal. A hundred years ago people didn’t need to be good readers in order to earn a living. But in the Information Age, no one can get by without knowing how to read well and understand increasingly complex material.”49

Fortunately, there is today a broad understanding of the reading process, new tools for better understanding the reading process, and affordable technology that can automate the algorithmic methods which will enable more readers to read at their full intellectual capacity.

We look forward to continued independent examination of LiveInk by experts in diverse fields. LiveInk technology, and the multidimensional LiveInk reading method, present many testable hypotheses in reading, language, spatial cognition, and learning.

- 22 - 1 Huey EB. The Psychology and Pedagogy of Reading. MIT Press. Cambridge, MA. 1908/1968. 2 Legge GE, Klitz TS, Tjan BS. Mr. Chips: An ideal-observer model of reading. Psychological Review (1997). 104:524-553. 3 Legge GE, Ahn SJ, Klitz TS, Luebker A. “Psychophysics of Reading—XVI. The Visual Span in Normal and Low Vision.” Vision Research 14: 1999-2010, 1997. 4 Legge GE, Klitz TS, Tjan BS . Ibid. 5 Manguel, Alberto. A History of Reading. pp 41-53. Penguin Books. New York. 1996. 6 Pollatsek A, Raney GE, La Gasse L, and Rayner K, The use of information below fixation in reading and in visual search. Can J. Experiment. Psychol. 47:179-200, 1993. 7 Andrews TJ, Coppola DM. Idiosyncratic characteristics of saccadic eye movements when viewing different visual environments. Vision Research 39 (1999): 2947-2853. 8 Horowitz TS and Wolfe JM. Visual search has no memory. Nature 394: 505-507 (1998). 9 TR Vidyasagar and K Pammer. Impaired visual search in dyslexia relates to the role of the magnocellular pathway in attention. NeuroReport 1999 10:1283-1287 10 Carpenter PA, Just MA “What your eyes do while your mind is reading. In K. Rayner, Eye movements in reading: perceptual and language processes. (pp. 275-301). Academic Press. New York. 1983. 11 Hindle D, Rooth M. Structural ambiguity and lexical relations. Computational Linguistics 19: 103-120. 1993. 12 Chomsky N, Halle M. The Sound Pattern of English. (pp 299-300, 371-372). MIT Press. Cambridge MA, 1991. 13 Ferreira, Fernanda and Anes, Michael, “Why Study Spoken Language?”, Chapter 2. In Handbook of Psycholinguistics, Mortan Ann Gernsbacher, ed., (pp 34-36).Academic Press, San Diego. 1994. 14 Gernsbacher MA, Language Comprehension as Structure Building.(pp 145-160). Lawrence Erlbaum Associates, Hillsdale, NJ 1990 15 Jackendoff, Ray. Languages of the Mind: Essays on Mental Representation. (pp 6-7). MIT Press. Cambridge MA. 1992. Jackendoff cites evidence that prosodic structure can be hierarchichal and states that “there must also be principles of correspondence that establish, for example, how the intonational phrasing of a sentence corresponds to its syntactic phrasing.” We suggest that the structure of prosody is more musical, and that some of the principles of musical parsing that Jackendoff describes in this volume and in others, will also have correspondences with syntactic structure. 16 Steinhauer K, Alter K, Friederici AD. Brain potentials indicate immediate use of prosodic cues in natural speech processing. Nature Neuroscience. 2 (2): 191-196, February, 1999. 17 Hoffman, Donald. Visual Intelligence: How We Create What We See. W.W. Norton, New York. 1998 18 Pinker S, The Language Instinct: How the Mind Creates Language. (pg 97). W. Morrow & Co., New York, 1994. This sentence is cited as being particularly difficult to comprehend because of its multiple embedded clauses. 19 Linzie B, Fletcher CR, Minnesota Center for Cognitive Sciences, (manuscript in preparation). 20 Pickering MJ, Clifton C, Crocker MW. Architectures and Mechanisms in Sentence Comprehension. (pp 1-28) in Architectures and Mechanisms in Language Processing. Crocker MW, Pickering MJ, and Clifton C, eds. Cambridge University Press. 2000. 21 Walker RC. Prose Paint: A method and apparatus to generate and depict syntactic relations using commutable chromatic vectors for improved human reading. Protected under US Pat. 5,802,533 and patents pending. 22 Timothy A. Keller, Patricia A. Carpenter and Marcel Adam Just. The Neural Bases of Sentence Comprehension: a fMRI Examination of Syntactic and Lexical Processing. Cerebral Cortex, Vol. 11, No. 3, 223-237, March 2001 23 Just MA and Carpenter PA. A Capacity Theory of Comprehension: Individual Differences in Working Memory. Psychological Review, 99:122-149, 1992. 24 Norman S, Kemper S, Kynette D. Adults’ reading comprehension: Effects of syntactic complexity and working memory. Journals of Gerontology. 47: 258-265. 1992. 25 Ni W, Constable RT, Mencle WE, et al. An Event-related Neuroimaging Study Distinguishing Form and Content in Sentence Processing. Journal of Cognitive Neuroscience. 12: (1), 120-133. January 2000. 26 Henderson JM, Ferreira F. Effects of foveal processing difficulty on the perceptual span in reading: Implications for attention and eye movement control. J of Exper Psych: Learning, Memory, and Cognition. 16: 417-429. 1990. 27 Melcher D, Kowler E. Shapes, Surfaces and Saccades. Vison Research. 39: 2929-2946, 1999. 28 Bringuier V, Chavave F, Glaeser L, Fregnac Y. Horizontal propagation of visual activity in the synaptic integration field of area 17 neurons. Science 283: 695-699.. 29 Jan 99.

29 Jackendoff R. “The Architecture of the Linguistic-Spatial Interface” in Language & Space, (pp 1-40) Bloom P, Peterson MA, Nadel L, and Garret MF, eds. MIT Press, Cambridge. 1996. 30 Petitot, Jean. “Morphodynamics and Attractor Syntax: Constituency in Visual Perception and Cognitive Grammar”. In Mind as Motion: Explorations in the Dynamics of Cognition. (pp.228-280) MIT Press, Cambridge,1995.

31 Shaywitz SE et al. Evidence that dyslexia may represent the lower tail of a normal distribution of reading ability. N Eng J Med 1992; 326:145-50. 32 Shaywitz, SE et al, “Functional disruption in the organization of the brain for reading in dyslexia” Proc. Nat. Acac. Sci. Vol. 95, Issue 5 Neurobiology, 2636-2641, March 3, 1998 33 Shaywitz, SE et al. Proc. Nat. Acac. Sci Ibid. 34 Pinker, Stephen. How the Mind Works. (p.90) W. Morrow & Co. 1996. 35 Demb JB, et al “Brain activity in visual cortex predicts individual differences in reading performance.” Proc. Natl. Acad. Sci. 1997 (Nov.25); 94: 13363-13366.

36 Sigman M, Giblert CD. Learning to find a shape. Nature Neuroscience. 3: 264-269, March 2000. 37 Demont E, Gombert JE. Phonological awareness as a predictor of recoding skills and syntactic awareness as a predictor of comprehension skills. British Journal of Educational Psychology. Vol. 66: 315-332, Sep 1996. 38 Abrahamsen EP, Shelton KC. Reading comprehension in adolescents with learning disabilities: Semantic and syntactic effects. Journal of Learning Disabilites. Vol 22 (9): 569-572, 1989. 39 Cupples L, Holmes VM. Evidence for a difference in syntactic knowledge between skilled and less skilled adult readers. Journal of Psycholinguistic Research. Vol. 21(4): 249-274, 1992. 40 Rasinski TV. Developing syntactic sensitivity in reading through phase-cued texts. In: Reading Research Anthology: the Why? Of Reading Instruction,. by Consortium on Reading Excellence (CORE), et al. (pp 94-97) Arena Press. Novato, CA. 1999. 41 Conte B, Menyuk P, Bashir A. Facilitating reading comprehension in middle school students with language disorders. International Jounal of Psycholinguistics. Vol. 10 (3): 273-280, 1994. 42 Frenck-Mestre C, Pynte J. Syntactic ambiguity resolution while reading in second and native languages. Quarterly Journal of Experimental Psychology. A. Human Experimental Psychology. Vol 50A (1): 119-148, Feb 1997. 43 Lim S-Y, Sauter SL, Schnorr TM. Occupational Health Aspects of Work with Video Display Terminals. In: NIOSH Pubications on Video Display Terminals. US Dept of Health and Human Services. DHHS (NIOSH) Pub. No. 99-135. (pp 3-14). Sept 1999. 44 Working Safely with Video Display Terminals (VDT). Occupational Safety and Health Administration. US Dept of Labor. OSHA 3092. 1996. “NIOSH recommends a 10-minute rest break after 2 hours of continuous VDT work for operators under moderate visual demands; and a 15-minute rest break after 1-hour of continuous work where there is a high visual demand or repetitive work task.” 45 Van Den Broek, Paul. “Comprehension and memory of narrative texts.” In Handbook of Psycholinguistics, Mortan Ann Gernsbacher, ed., (pp 539-588) Academic Press, San Diego. 1994. 46 Fletcher, Charles R. “Levels of Representation in memory for discourse.” In Handbook of Psycholinguistics, Mortan Ann Gernsbacher, ed., (pp 589-607) Academic Press, San Diego. 1994. 47 Perfetti, Charles A. “Psycholinguistics and Reading Ability.” In Handbook of Psycholinguistics, Mortan Ann Gernsbacher, ed., (pp 849-894) Academic Press, San Diego. 1994. 48 Gardner, Howard. Multiple Intelligences: The theory in practice. Basic Books. New York. 1993. 49Kantrowitz B, Underwood A. Dyslexia and the New Science of Reading. Newsweek. (pp 72-79). November 22, 1999.