How do you think we read?
How do you think we read?
-memorizing words on the page
-extracting just the meanings of the words
-playing a mental movie in our heads of
what the text describes
-some combination of these???
An answer?
• Evidence suggests that we can use all of
these “codes” or levels of representation.
However, some are more important that
In the Beginning: Reading
• Stage 0 (prior to 1st grade)
– Discriminate letters (i.e., Pepsi vs. Coke)
• Stage 1 (first year formal instruction)
Phonological recoding skills are learned
• Stage 2 (2nd & 3rd grades)
– Children are reading fluently but it is effortful and they
don’t comprehend much
• Stage 3 (grades 4-8)
– Reading as a tool to gather knowledge, switch to
reading individual words rather than sounding
everything out.
How could we study Reading
• Four levels of analysis:
– Phonology: Study of production and perception of
language sounds.
– Syntax: The study of the structure of sentences, and
of rules determining the order of words and phrases
in those sentences.
– Semantics: The study of the meaning of words
– Pragmatics: Context and social interaction coupled
with semantics
• Morpheme: The smallest language unit
that carries meaning. Morphemes are
conveyed by sounds called phonemes.
English has 46 different phonemes. There
are only 200 across all languages.
Languages vary they may have as few as
20 or more than 80.
Hooked on Phonics, DIBELS,
Reading First
• Sound-letter correspondence is critical to
decoding words and retrieving their
• Direct instruction targets teaching children
how to sound out words.
• There are diagnostic tools used to test
student progress (i.e., DIBELS)
Levels of Representation
• Surface level: Memory for veridical
wording, typeface, color.
• Textbase: Memory for the meaning of
words used in the text and their explicit
• Macrostructure/Situation level: Memory
for the “gist” can include information that
wasn’t even in the text.
Word Identification
• Direct access: Use visual representation
to identify. See word go directly to
“meaning dictionary” lexicon.
• E.g., DOG- access without sounding it out
Word Identification
• Direct access: Use visual representation
to identify. See word go directly to
“meaning dictionary” lexicon.
• E.g., DOG- access without sounding it out
• Problem with this view: SLOM can not be
accessed because this letter string is not
in our lexicon.
Word Identification
• Indirect access: use a words sound to identify it.
• See word and sound it out using grapheme-to-phoneme
correspondence (GPC) rules.
• Slom can be said using this method.
• Dual access: We use both direct and indirect methods.
• Familiar words use direct access.
• Unknown/uncommon words (Slom) use indirect access.
• Horse race model (Rayner and Pollatsek, 1989) we use
both at the same time. One gets there faster.
Beyond the Word (Discourse)
• Propositional representations: A collection of
conceptual nodes labeled by pathways, where the entire
structure represents the meaning of the sentence.
• Strength of this type of representation: Reflects the
meanings of sentences but is not sensitive to changes in
surface features (e.g., paraphrases).
Evidence: Kintsch (1974)
The crowded passengers squirmed uncomfortably. (2
The horse stumbled and broke a leg. (3 propositions)
Discourse Structures
• Kintsch and van dijk’s Model: Posits a
distinction between Microstructure and
• Microstructure: The level of discourse in which
propositions (smallest unit of meaning that can
have a truth-value) are linked together.
• Propositions have two elements:
– Argument (concept) usually a noun or some object
– Predicate (focus) usually a verb or some relational
Discourse Structures
• Macrostructure: The gist of the text (what we
walk away from the text remembering).
• Primary goal of this model is to explain the
coherence of a text (i.e., how well a text makes
sense). Coherence is achieved by an overlap of
arguments in propositions.
• This model also accounts for the “bottle neck” of
STM. We process in cycles where the most
recent and most important propositions are kept
Kintsch’s CI Theory
Construction Integration Model
• Readers break down text into propositions
• Understanding the text is the process of linking
propositions together into a coherence graph
(this is the microstructure)
• The macrostructure is then built, which consists
of prior real-world knowledge (schema) and an
edited version of the microstructure.
Kintsch’s CI Theory continued
Problems w/ the model:
• Too many details of the processes
(forming propositions) are not well worked
• Understanding a text (coherence) is more
than simply linking a series of
An Alternative View to Propositions:
Perceptual Symbols (Barsalou,
Glenberg, Zwaan)
• The amodal argument
• Readers understand the text as if they are
in the story world (embodiment).
Propositional theories don’t capture this.
• In this case understanding text is the
process (re)activating parts of the brain
associated with experiences.

Reading Comprehension Three divisions within the