Cognitive and Metalinguistic Precursors of Emergent Literacy Skills: A Reexamination of the Specific

1
Cognitive and Metalinguistic Precursors of
Emergent Literacy Skills A Reexamination of the
Specific Roles Played by Syntactic Awareness and
Phonological Awareness in Early Reading Skills

• Sarah E. Torok, Ph.D.
• Psychology Department
• Mount Union College
• Alliance, OH

2
Is anything wrong with this passage?

• I twisted the foil tin around the antennae.
  Things cleared up enough to made out President
  Johnson taking his seat at a desk, people all
  around. I didnt care much for the president
  because of the way she held his beagles ears by
  the. I did admired his wife, Lady Bird, though,
  who always look like she wanted more nothing than
  to sprout wings and away flied.
• The Secret Life of Bees, Sue Monk Kidd, p. 20

3
Where are the errors? How did you know?
I twisted the foil tin around the antennae.
Things cleared up enough to made out President
Johnson taking his seat at a desk, people all
around. I didnt care much for the president
because of the way she held his beagles ears by
the. I did admired his wife, Lady Bird, though,
who always look like she wanted more nothing than
to sprout wings and away flied. The Secret
Life of Bees, Sue Monk Kidd, p. 20
4
This makes more sense...

• I twisted the tin foil around the antennae.
  Things cleared up enough to make out President
  Johnson taking his seat at a desk, people all
  around. I didnt care much for the president
  because of the way he held his beagles by the
  ears. I did admire his wife, Lady Bird, though,
  who always looked like she wanted nothing more
  than to sprout wings and fly away.
• The Secret Life of Bees, Sue Monk Kidd, p. 20

5
What processes are involved in learning to read?
(The What)
6
(No Transcript)
7
Exogenous Latent Constructs

• Verbal Working Memory
• The mental workspace in which recent
  environmental stimuli are briefly held in the
  form of linguistic codes, either for rehearsal
  and recall or for meaningful integration with
  other knowledge.
• General Linguistic Awareness
• Subsumes skills such as phonological, semantic,
  and syntactic coding. These skills are the means
  by which we store information encoded in spoken
  and written text into permanent memory.

8
Exogenous Latent Constructs

• General Analytic Ability
• (a.k.a. general intelligence) refers to the
  ability to analyze, categorize, and relate
  information in an attempt to achieve a conceptual
  grasp of objects, events, and procedures that
  exist or occur in the external world.
• It has been found to have links to reading
  comprehension skills even though it is not a
  reliable predictor of word identification and
  decoding skills.
• Operativity
• This has been used as a proxy measure of general
  analytic ability in some of the following
  analyses and refers to Piagetian concepts such as
  seriation and conservation.

9
Endogenous Latent Constructs

• Metalinguistic Awareness
• Entails manipulating spoken and written text and
  analyzing the linguistic processes associated
  with both.
• In the context of the research being examined,
  there is controversy over whether this should be
  conceptualized as a single construct encompassing
  several forms of awareness such as phonological
  awareness and syntactic awareness.

10
Endogenous Latent Constructs

• Phonological Awareness
• Explicit knowledge and conceptual grasp of the
  speech-sound systems that govern a language.
• Phonemes are the smallest units of sound in a
  given language.
• Each language uses a small and finite set of
  phonemes to form words.
• Each language uses different sets of rules for
  combining these phonemes.
• Syntactic Awareness
• The ability to reflect on and manipulate the
  words in a sentence in terms of their structural
  relationships (e.g., Subject-Verb agreement)
• Sally went to the store to bought a can of
  soda.

11
Endogenous Latent Constructs

• Phonological Decoding
• The ability to sound out words using letter
  sounds and combinations of letter sounds (-at in
  cat, fat, and hat).
• Decontextualized Word Identification
• Word identification refers to the foundational
  ability to identify whole words at sight in
  written text.
• Decontextualized word identification refers to
  the ability to identify words without the aid of
  meaningful of context.
• Reading Comprehension (terminal endogenous latent
  construct)
• Reading comprehension refers to the ability to
  extract and construct meaning from written text.
  
• All of the models discussed in the current study
  have suggested that many different cognitive and
  linguistic factors can affect childrens
  abilities to comprehend what they read.

12
Three Theories Under Investigation(The Who)
13
Theory One

• Blackmore and Pratt (1997)
• metalinguistic awareness is a single construct
  that includes phonological awareness and
  syntactic awareness

14
Blackmore and Pratts (1997) Theory Regarding the
Relationships Metalinguistic Awareness as a
Single Construct and Reading Skills
15
Theory Two

• Tunmer, Nesdale, and Herriman (1988)
• phonological and syntactic awareness both make
  separate contributions to reading achievement
  (e.g., phonological decoding, word id, reading
  comprehension)

16
Tunmer et al.s (1988) Theory Regarding the
Relationship between Forms of Metalinguistic
Awareness and Reading Skills
17
Theory Three

• Torok (2002)
• validated aspects of both theories while
  clarifying and improving on their methodologies
• added more variables to the mix
• used longer period of time to assess
  relationships
• used larger sample size

18
Toroks (2002) Theoretical Model Depicting Tunmer
et al.s Hypothesized Relationships between
Phonological Awareness, Syntactic Awareness, and
Reading Skills, While Controlling for Linguistic
and Cognitive Skills
19
Toroks (2002) Theoretical Model Depicting
Blackmore and Pratts (1997) Hypothesized
Relationships between Metalinguistic Awareness as
a Unitary Construct and Reading Skills, while
Controlling for Linguistic and Cognitive Abilities
20
Goals of Current Study

• Used structural equation modeling (SEM) to
  reexamine these relationships
• 3 objectives to test validity of
• Blackmore and Pratts (1997) theory
• Tunmer et al.s (1988) theory
• modifications to these views using constructs
  from Torok (2002)

21
Predictions

• SEM techniques provide support for
• Aspects of Blackmore and Pratts (1997) original
  model
• Aspects of Tunmer et al.s (1988) original model
• Toroks (2002) modifications to both these models

22
Methods(The How)
23
Sample

• Characteristics
• 183 children
• longitudinal study
• K-4th grade
• middle- to upper middle-class (n 1,407) Albany,
  NY area
• Selection
• battery of tests assessing cognitive abilities
  and literacy skills.
• In the middle of 1st grade, sub-samples of poor
  (n 118) and normal readers (n 65) were
  selected from the initial population (n 1,284).
  
• Based on recommendations from teachers and
  performance on intelligence and reading
  achievement tests
• At the same time the poor readers were randomly
  assigned to tutored or non-tutored groups.
• tutored group (n 76) received individual
  instruction for a ½ hour daily for up to 2
  semesters
• some in the non-tutored group received small
  group instruction (n 26)
• others in the non-tutored group received 2-3
  days of individual instruction (n 16)

24
Observed Measures
See handouts for descriptions of observed
measures.
25
Data Analysis Technique (SEM)

• What is Structural Equation Modeling (SEM)?
• Sophisticated statistical technique that allows
  researchers to evaluate hypotheses about the
  directions of influence in correlational
  relationships
• In other words, it goes beyond linear regression
  analyses because it can look at direct and
  indirect relationships between the constructs
  being investigated
• However, just like linear regression, it is still
  looking at correlational relationships so SEM
  cannot imply causal relationships
• Constructs vs. observed measures
• Constructs are the general theoretical components
  of the models (e.g. Syntactic Awareness). The
  nature of the constructs relationships are
  either the source of investigation in research
  (exploratory factor analysis) or have already
  been somewhat established in research
  (confirmatory factor analysis)
• Observed measures are the measurable variables
  that are used to operationally define the
  constructs (e.g., grammaticality judgments task)

26
More on SEM...

• Path analysis
• depicts the hypothesized direction of the
  relationships between the constructs under
  investigation (structural models)
• again, the ideas about the general direction of
  the relationships are typically established in
  previous research using other statistical
  techniques (e.g. hierarchical regression
  analyses)
• Unlike regression analysis, SEM takes account of
  measurement error by including a measurement
  model that relates the latent constructs to
  observed measures of those constructs, in
  addition to a structural model that relates the
  constructs to one another.
• my models show just the constructs because the
  observed measures I used are fairly well
  established in research as good measures of the
  constructs I am investigating
• Goodness of Fit (GFI)
• This is a statistical index of how well the model
  fits the current set of data.
• Path coefficients
• numbers at the bottom of each arrow indicate how
  much variance one construct accounts for in
  another construct

27
Statistical Indicators
28
Results SEM Analyses
29
Blackmore and Pratts (1997) Original Theoretical
Conceptions ?2(49) 155.83, p lt .0001, GFI
.88, NFI .90, CFI .93
30
Tunmer et al.s Original Theoretical Conceptions
?2(38) 175.13, p lt .0001, GFI .86, NFI
.88, CFI .90
31
Toroks Modifications to Tunmer et al.
(1988)?2(47) 164.22, p lt .0001, GFI .88,
NFI .89, CFI .92
32
Comparisons of Fit Indices for Each Specified
Model
33
General Discussion
34
Findings in Current Study Objective 1

• To use SEM techniques to test the viability of
  Blackmore and Pratts original model regarding
  the role that phonological and syntactic
  awareness play in reading development.
• This was supported because both sets of models
  provided comparable fit.
• This finding provides additional support for
  Tunmer et al.s model because it fails to provide
  strong evidence that refutes their views
  regarding the independent roles of phonological
  awareness and syntactic awareness.

35
Findings in Current Study Objective 2

• To use (SEM) to test the viability of Tunmer et
  al.s (1988) original model regarding the roles
  that phonological and syntactic awareness play in
  reading development
• This prediction was ultimately supported by the
  structural models depicting variations in Tunmer
  et al.s original model, which all yielded
  acceptable fit indices and statistically
  significant path coefficients for direct,
  indirect, and total effects for hypothesized
  relationships involving phonological awareness
  and syntactic awareness as predictors of
  phonological decoding, word identification, and
  reading comprehension.

36
Findings in Current Study Objective 3

• To test the viability of Tunmer et al.s (1988)
  and Blackmore and Pratts (1997) models, using
  SEM techniques, after incorporating the
  constructs that were added in the Torok (2002)
  study.
• This prediction was supported. The modifications
  prompted by Toroks (2002) study did generate
  several models that produced acceptable fit
  statistics.
• Results can be interpreted as additional support
  for Toroks (2002) suggestion that neither Tunmer
  et al. nor Blackmore and Pratt adequately
  controlled for all of the cognitive abilities
  that may underlie metalinguistic awareness.

37
Toroks (2005) Revised Model Combining Findings
from Previous Research by Tunmer et. al (1988),
Blackmore and Pratt (1997), and Torok (2002)
38
Theoretical Implications

• The role of phonological and syntactic awareness
  in early literacy.
• In essence, the current study revealed that both
  theoretical views regarding the role of
  phonological awareness and syntactic awareness
  are viable and relevant.
• The relationships specified in the models
  evaluating Tunmer et al.s theory produced a
  pattern of results, that were not incompatible
  with relationships specified in the models
  evaluating Blackmore and Pratts theory.

39
Practical Implications (Answering the So what,
who cares? question.)

• Findings suggest that instruction in phonological
  and syntactic awareness skills, along with joint
  use of phonological decoding and text-based
  strategies for word identification, is a better
  approach to teaching how to read than reading
  instruction that places more emphasis on only one
  of these metalinguistic skills
• whole-word contextual approaches (uses only
  syntactic awareness)
• strict phonemic analysis approaches (uses only
  phonological awareness)
• Some evidence has been found for this in
  intervention studies.

40
Limitations

• Overlapping variance among measures
• Possible collinearity existed among some of the
  observed measures used in the structural equation
  models which could potentially misrepresent the
  true relationships that exist between theoretical
  constructs.
• Generalizability of results
• Another limitation of the current study is that
  the research sample was taken from a clinical
  population sampled for purposes of evaluating the
  cause(s) and correlates of reading disability.

41
Future Research

• Randomized experimental design
• Assess different effects of formalized training
• Four experimental groups consisting of four sets
  of classrooms would be compared.
• Group 1 receive classroom instruction that
  emphasizes phonological awareness and the use of
  phonological decoding strategies for identifying
  unknown words during text reading.
• Group 2 receive classroom instruction that
  emphasizes syntactic awareness and the use of
  context-based strategies for word identification,
  during text reading.
• Group 3 receive classroom instruction that
  emphasizes both strategies for identifying
  unfamiliar words during text reading.
• Group 4 receive neither of these treatments and
  would serve as the control group.
• Classroom teachers
• randomly assigned to one of the four treatment
  groups
• Pre and post treatment outcome and control
  measures would be administered

42
Questions?