Handbook of Learning Disabilities

178 CAUSES AND BEHAVIORAL MANIFESTATIONS

ers in recognizing which serial position an Barwick, M. A., & Siegel, L. S. (1996). Learning
individual letter occurred in most frequent- difficulties in adolescent clients of ashelter for
ly. In this study, good and poor readers were runaway and homeless street youths. Journal of
shown two pseudo-words, one containing a Research on Adolescence, 6, 649–670.
letter in its most frequent serial position and
the other containing the letter in its least fre- Bentin, S., Deutsch, A., & Liberman, I. Y. (1990).
quent serial position. Poor readers made Syntactic competence and reading ability in chil-
more errors than good readers. Thus, poor dren. Journal of Experimental Psychology, 49(1),
readers had less orthographic knowledge 147–172.
about single letters, in contrast to groups of
letters, than did good readers Besner, D., Twilley, L., McCann, R., & Seergobin,
K. (1990). On the association between connec-
Conclusions tionism and data: Are a few words necessary?
Psychological Review, 97, 1–15.
The period of rapid acquisition of reading
skills, three processes—phonological, syn- Bradley, L., & Bryant, P. E. (1983). Categorizing
tactic, and working memory—show signifi- sounds and learning to read: A causal connec-
cant increases in development. These tion. Nature, 301, 419–421.
processes are significantly disrupted in chil-
dren who are reading disabled, but semantic Brittain, M. M. (1970). Inflectional performance
and orthographic processes are not disrupt- and early reading achievement. Reading Research
ed to the same extent. However, the under- Quarterly, 1, 34–48.
utilization of phonological processing and
the reliance almost entirely on semantics Brown, A. D. (1987). Resolving inconsistency: A
and orthographic or visual processes dis- computational model of word naming. Journal of
rupts reading. A deficit in three fundamen- Memory and Language, 26, 1–23.
tal cognitive processes—phonological pro-
cessing, syntactic awareness, and working Bruck, M. (1988). The word recognition and
memory—constitutes the basic characteris- spelling of dyslexia children. Reading Research
tics of reading disability. It is important that Quarterly, 23, 51–68.
assessment for learning disabilities reflect an
understanding of these processes and sys- Bruck, M. (1990). Word-recognition skills of adults
tematically measure them. with childhood diagnosis of dyslexia. Develop-
mental Psychology, 26, 439–454.
Acknowledgments
Bryson, S. E., & Werker, J. F. (1989). Toward un-
The preparation of this chapter was partially sup- derstanding the problem in severely disabled
ported by a grant from the Natural Sciences and readers: I. Vowel errors. Applied Psycholinguis-
Engineering Research Council of Canada and was tics, 10, 1–12.
written while I was a Scholar in Residence at The
Peter Wall Institute for Advanced Studies. I wish to Byrne, B. (1981). Deficient syntactic control in poor
thank Sarah Kontopoulos and Stephanie Vyas for readers: Is a weak phonetic memory code respon-
their secretarial assistance. sible? Applied Psycholinguistics, 2, 201–212.

References Byrne, B., & Shea, P. (1979). Semantic and phone-
mic memory in beginning readers. Memory and
Backman, J., Bruck, M., Hebert, M., & Seidenberg, Cognition, 7, 333–341.
M. (1984). Acquisition and use of spelling–sound
correspondences in reading. Journal of Experi- Calfee, R. C., Lindamood, P., & Lindamood, C.
mental Child Psychology, 38, 114–133. (1973). Acoustic–phonetic skills and reading:
Kindergarten through twelfth grade. Journal of
Backman, J. E., Mamen, M., & Ferguson, H. B. Educational Psychology, 64, 293–298.
(1984). Reading level design: Conceptual and
methodological issues in reading research. Psy- Coltheart, M. (1978). Lexical access in simple read-
chological Bulletin, 96, 560–568. ing tasks. In G. Underwood (Ed.), Strategies of
information processing (pp. 151–216). London:
Academic Press.

Cromer, W., & Wiener, M. (1966). Idiosyncratic re-
sponse patterns among good and poor readers.
Journal of Consulting Psychology, 30, 1–10.

Da Fontoura, H. A., & Siegel, L. S. (1995). Read-
ing, syntactic, and working memory skills of
bilingual Portuguese–English Canadian children.
Reading and Writing: An Interdisciplinary Jour-
nal, 7, 139–153.

Doctor, E. A., & Coltheart, M. (1980). Children’s
use phonological encoding when reading for
meaning . Memory and Cognitive, 8, 195–209.

Ehri, L. C., & Wilce, L. S. (1980). The influence of
orthography on readers’ conceptualization of the
phonemic of words. Applied Psycholinguistics, 1,
371–385.

Ehri, L. C., & Wilce, L. S. (1983). Development of
word identification speed in skilled and less-
skilled beginning readers. Journal of Educational
Psychology, 75, 3–18.

Basic Cognitive Processes and Reading Disabilities 179

Ellis, A. W. (1985). The cognitive neuropsychology ic readers. British Journal of Psychology, 73,
of developmental (and acquired) dyslexia: A criti- 455–460.
cal survey. Cognitive Neuropsychology, 2, Jorm, A. F. (1979). The nature of reading deficit in
169–205. developmental dyslexia: A reply to Ellis. Cogni-
tion, 1, 421–433.
Forster, K. I., & Chambers, S. (1973). Lexical ac- Katz, L. (1977). Reading ability and single-letter or-
cess and naming time. Journal of Verbal Learning thographic redundancy. Journal of Educational
and Verbal Behavior, 12, 627–635. Psychology, 69, 653–659.
Lennox, C., & Siegel, L. S. (1993). Visual and
Fowler, C., Liberman, I., & Shankweiler, D. (1977). phonological spelling errors in subtypes of chil-
On interpreting the error pattern in beginning dren with learning disabilities. Applied Psy-
reading. Language and Speech, 20, 162–173. cholinguistics, 14, 473–488.
Liberman, I. Y., Liberman, A. M., Mattingly, I., &
Fowler, C., Shankweiler, D., & Liberman, I. (1979). Shankweiler, D. (1980). Orthography and the be-
Apprehending spelling patterns for vowels: A de- ginning reader. Unpublished manuscript.
velopmental study. Language and Speech, 22, Lindgren, S. D., De Renzi, E., & Richman, L. C.
243–251. (1985). Cross–national comparisons of develop-
mental dyslexia in Italy and the United States.
Frith, U., & Snowling, M. (1983). Reading for Child Development, 56, 1404–1417.
meaning and reading for sound in autistic and Lovett, M. W. (1979). The selective encoding of se-
dyslexic children. British Journal of Developmen- quential information in normal reading develop-
tal Psychology, 1, 329–342. ment. Child Development, 50, 897–900.
Manis, F. R., & Morrison, F. J. (1985). Reading
Frost, R. (1998). Toward a strong phonological the- disability: A deficit in rule learning? In L. S.
ory of virtual word recognition: True issues and Siegel & F. J. Morrison (Eds.), Cognitive devel-
false trails. Psychological Bulletin, 123(1), opment in atypical children: Progress in cogni-
71–99. tive development research (pp. 1–26). New York:
Springer–Verlag.
Geva, E., & Siegel, L. S. (2000). Orthographic and Manis, F. R., Savage, P. L., Morrison, F. J., Horn, C.
cognitive factors in the concurrent development C., Howell, M. J., Szesulski, P. A., & Holt, L. K.
of basic reading skill in two languages. Reading (1987). Paired associate learning in reading-dis-
and Writing: An Interdisciplinary Journal, 12, abled children: Evidence for a rule-learning defi-
1–30. ciency. Journal of Experimental Child Psycholo-
gy, 43, 25–43.
Glass, A. L., & Perna, J. (1986). The role of syntax Manis, F. R., Szeszulski, P. A., Howell, M. J., &
in reading disability. Journal of Learning Disabil- Horn, C. C. (1986). A comparison of analogy-
ities, 19, 354–359. and rule-based decoding strategies in normal and
dyslexic children. Journal of Reading Behavior,
Glushko, R. J. (1979). The organization and activa- 18, 203–213.
tion of orthographic knowledge in reading aloud. Mann, V. A. (1984). Longitudinal prediction and
Journal of Experimental Psychology: Human prevention of early reading difficulty. Annals of
Perception and Performance, 5, 674–691. Dyslexia, 34, 117–136.
Mann, V. A., Liberman, I. Y., & Shankweiler, D.
Goldman, S. R. (1976). Reading skills and the mini- (1980). Children’s memory for sentences and
mum distance principle: A comparison of listen- word strings in relation to reading ability. Memo-
ing and reading comprehension. Journal of Ex- ry and Cognition, 8, 329–335.
perimental Child Psychology, 22, 123–142. Metsala, J., & Siegel, L. S. (1992). Patterns of atyp-
ical reading development: Attributes and under-
Gough, P. B., & Juel, C. (1991). The first stages of lying reading processes. In S. Segalowitz & I.
word recognition. In L. Rieben & C. A. Perfetti Rapin (Eds.), Handbook of neuropsychology
(Eds.), Learning to read: Basic research and its (Vol. 7, pp. 187–210). New York: Elsevier.
implications (pp. 47–56). Hillsdale, NJ: Erlbaum. Meyer, D. E., Schvanevelt, R. W., & Ruddy, M. G.
(1974). Functions of graphemic and phonemic
Gough, P. B., & Tunmer, W. E. (1986). Decoding, codes in visual word-recognition. Memory and
reading, and reading disability. Remedial and Cognition, 2, 309–321.
Special Education, 7, 6–10. Olson, R. K., Kliegl, R., Davidson, B. J., & Foltz,
G. (1985). Individual and developmental differ-
Guthrie, J. T., & Seifert, M. (1977). Letter–sound ences in reading disability. In G. E. MacKinnon
complexity in learning to identify words. Journal & T. G. Waller (Eds.), Reading research: Ad-
of Educational Psychology, 69(6), 686–696. vances in theory and practice (Vol. 4, pp. 1–64).
New York: Academic Press.
Guthrie, J. T., & Seifert, M. (1983). Profiles of Pennington, B. F., McCabe, L. L., Smith, S. D.,
reading activity in a community. Journal of Read- Lefly, D. L., Bookman, M. O., Kimberling, W. J.,
ing, 26, 498–508. & Lubs, H. A. (1986). Spelling errors in adults

Harris, M., & Coltheart, M. (1986). Language pro-
cessing in children and adults: An introduction.
London: Routledge & Kegan Paul.

Hogaboam, T. W., & Perfetti, C. A. (1978). Read-
ing skill and their role of verbal experience in de-
coding. Journal of Educational Psychology, 5,
717–729.

Humphreys, G. W., & Evett, L. J. (1985). Are there
independent lexical and nonlexical routes in the
word processing? An evaluation of the dual-route
theory of reading. Behavioral and Brain Sciences,
8, 689–740.

Johnston, R. (1982). Phonological coding in dyslex-

180 CAUSES AND BEHAVIORAL MANIFESTATIONS

with a form of familial dyslexia. Child Develop- and Writing: An Interdisciplinary Journal, 1,
ment, 57, 1001–1013. 37–52.
Pratt, A. C., & Brady, S. (1988). Relations of Siegel, L. S., & Heaven, R. K. (1986). Categoriza-
phonological awareness to reading disability in tion of learning disabilities. In S. J. Ceci (Ed.),
children and adults. Journal of Educational Psy- Handbook of cognitive, social and neuropsycho-
chology, 80, 319–323. logical aspects of learning disabilities (Vol. 1, pp.
Rack, J. P. (1985). Orthographic and phonetic cod- 95–121). Hillsdale, NJ: Erlbaum.
ing in developmental dyslexia. British Journal of Siegel, L. S., Levey, P., & Ferris, H. (1985). Sub-
Psychology, 76, 325–340. types of developmental dyslexia: Do they exist?
Seidenberg, M. S., Bruck, M., Fornarolo, G., & In F. J. Morrison, C. Lord, & D. P. Keating
Backman, J. (1985). Word recognition processes (Eds.), Applied developmental psychology (Vol.
of poor and disabled readers: Do they necessarily 2, pp. 169–190). New York: Academic Press.
differ? Applied Psycholinguistics, 6, 161–180. Siegel, L. S., & Linder, B. A. (1983). Short-term
Seidenberg, M. S., & McClelland, J. L. (1989). Vi- memory processes in children with reading and
sual word recognition and pronunciation: A com- arithmetic learning disabilities. Developmental
putational model of acquisition, skilled perfor- Psychology, 20, 200–207.
mance, and dyslexia. In A. M. Galaburda (Ed.), Siegel, L. S., & Metsala, J. (1992). An alternative to
From reading to neurons. Issues in the biology of the food processor approach to subtypes of learn-
language and cognition (pp. 255–303). Cam- ing disabilities. In N. N. Singh & I. Beale (Eds.),
bridge, MA: MIT Press. Current perspectives in learning disabilities: Na-
Seymour, P. H. K., & Elder, L. (1986). Beginning ture, theory, and treatment (pp. 44–60). New
reading without phonology. Cognitive Neuropsy- York: Springer-Verlag.
chology, 3, 1–36. Siegel, L. S., & Ryan, E. B. (1988). Development of
Seymour, P. H. K., & Porpodos, C. D. (1980). Lexi- grammatical sensitivity, phonological, and short-
cal and nonlexical processing of spelling in term memory in normally achieving and learning
dyslexia. In U. Frith (Ed.), Cognitive processes in disabled children. Developmental Psychology,
spelling (pp. 443–473). New York: Academic 24, 28–37.
Press. Siegel, L. S., & Ryan, E. B. (1989a). The develop-
Shafrir, U., & Siegel, L. S. (1991). Preference for vi- ment of working memory in normally achieving
sual scanning strategies versus phonological re- and subtypes of learning disabled children. Child
hearsal in university students with reading dis- Development, 60, 973–980.
abilities. Journal of Learning Disabilities, 27, Siegel, L. S., & Ryan, E. B. (1989b). Subtypes of de-
583–588. velopmental dyslexia: The influence of definition-
Shafrir, U., & Siegel, L. S. (1994). Subtypes of al variables. Reading and Writing: An Interdisci-
learning disabilities in adolescents and adults. plinary Journal, 2, 257–287.
Journal of Learning Disabilities, 27, 123–134. Siegel, L. S., Share, D., & Geva, E. (1995). Evidence
Shankweiler, D., & Liberman, I. (1972). Misread- for superior orthographic skills in dyslexics. Psy-
ing: A search for causes. In J. Kavanaugh & I. chological Science, 6(4), 250–254.
Mattingly (Eds.), Language by ear and by eye: Smiley, S. S., Pasquale, F . L., Chandler, C. L.
The relationship between speech and reading (pp. (1976). The pronunciation of familiar, unfamiliar
293–317). Cambridge, MA: MIT Press. and synthetic words by good and poor adolescent
Shankweiler, D., Liberman, I. Y., Mark, L. S., readers. Journal of Reading Behavior, 8(3),
Fowler, C. A., & Fischer, F. W. (1979). The 289–297.
speech code and learning to read. Journal of Ex- Snowling, M. J. (1980). The development of
perimental Psychology: Human Learning and grapheme–phoneme correspondence in normal
Memory, 5, 531–545. and dyslexic readers. Journal of Experimental
Siegel, L. S. (1985). Psycholinguistic aspects of Child Psychology, 29, 294–305.
reading disabilities. In L. S. Siegel & F. J. Mor- Snowling, M., & Frith, U. (1981). The role of
rison (Eds.), Cognitive development in atypi- sound, shape, and orthographic cues in early
cal children (pp. 45–65). New York: Springer- reading. British Journal of Psychology, 72,
Verlag. 83–87.
Siegel, L. S. (1988). Evidence that IQ scores are ir- So, D., & Siegel, L. S. (1997). Learning to read Chi-
relevant to the definition and analysis of reading nese: Semantic, syntactic, phonological and
disability. Canadian Journal of Psychology, 42, working memory skills in normally achieving and
201–215. poor Chinese readers. Reading and Writing: An
Siegel, L. S. (1992). An evaluation of the discrepan- Interdisciplinary Journal, 9, 1–21.
cy definition of dyslexia. Journal of Learning Sprenger-Charolles, L. (1991). Word–identification
Disabilities, 25, 618–629. strategies in a picture context: Comparisons be-
Siegel, L. S. (1993). Phonological processing deficits tween “good” and “poor” readers. In L. Rieben
as the basis of a reading disability. Developmen- & C. A. Perfetti (Eds.), Learning to read: Basic
tal Review, 13, 246–257. research and its implications (pp. 175–188).
Siegel, L. S., & Faux, D. (1989). Acquisition of cer- Hillsdale, NJ: Erlbaum.
tain grapheme–phoneme correspondences in nor- Stanovich, K. E. (1988a). Explaining the differences
mally achieving and disabled readers. Reading between the dyslexic and garden variety poor

Basic Cognitive Processes and Reading Disabilities 181

reader. The phonological–core variance–differ- Reading disability: An investigation of the per-
ence model. Journal of Learning Disabilities, 21, ceptual deficit hypotheses. Cortex, 8, 106–118.
590–604, 612. Venezky, R. L., & Johnson, D. (1973). Develop-
Stanovich, K. E. (1988b). The right and wrong ment of two letter–sound patterns in grades one
places to look for the cognitive locus of reading through three. Journal of Educational Psycholo-
disability. Annals of Dyslexia, 38, 154–177. gy, 64, 109–115.
Stanovich, K. E., Nathan, R. G., & Zolman, J. E. Wallach, M., & Wallach, L. (1976). Helping disad-
(1988). The developmental lag hypothesis in vantaged children learn to read by teaching them
reading: Longitudinal and matched reading–level phoneme identification skills. Paper presented at
comparisons. Child Development, 59, 71–86. the Learning Research and Development Center,
Steinhauser, R., & Guthrie, J. T. (1974). Scanning Pittsburgh University, Pittsburgh.
times through prose and word strings for various Waller, G. (1976). Children’s recognition memory
targets by normal and disabled readers. Perceptu- for written sentences: A comparison of good and
al and Motor Skills, 39, 931–938. poor readers. Child Development, 47, 90–95.
Stevenson, H. W., Stigler, J. W., Lucker, G. W., Hsu, Waters, G. S., Bruck, M., & Seidenberg, M. (1985).
C. C., & Kitamura, S. (1982). Reading disabili- Do children use similar processes to read and
ties: The case of Chinese, Japanese, and English. spell words? Journal of Experimental Child Psy-
Child Development, 53, 1164–1181. chology, 39, 511–530.
Tal, N. F., & Siegel, L. S. (1996). Pseudoword read- Waters, G. S., Seidenberg, M. S., & Bruck, M.
ing errors of poor , dyslexic and normally achiev- (1984). Children’s and adults’ use of
ing readers on multisyllable pseudo-words. Ap- spelling–sound information in three reading
plied Psycholinguistics, 17, 215–232. tasks. Memory and Cognition, 12, 293–305.
Taylor, H. G., Satz, P., & Friel, J. (1979). Develop- Weber, R. (1970). A linguistic analysis of first-grade
mental dyslexia in relation to other childhood reading errors. Reading Research Quarterly, 5,
reading disorders: Significance and clinical utility. 427–451.
Reading Research Quarterly, 15, 84–101. Werker, J. F., Bryson, S. E., & Wassenberg, K.
Temple, C. M. (1988). Red is read but eye is blue: A (1989). Toward understanding the problem in se-
case study of developmental dyslexia and follow- verely disabled readers. Part II: Consonant errors.
up report. Brain and Language, 34, 13037. Applied Psycholinguistics, 10, 13–30.
Vandervelden, M. C., & Siegel, L. S. (1995). Phono- Willows, D. M., & Ryan, E. B. (1981). Differential
logical recoding and phonemic awareness in early utilization of syntactic and semantic information
literacy: A developmental approach. Reading Re- by skilled and less skilled readers in the interme-
search Quarterly, 30, 854–875. diate grades. Journal of Educational Psychology,
Vellutino, F. R., Steger, J. A., & Kandel, G. (1972). 73, 607–615.

11

Memory Difficulties in Children and
Adults with Learning Disabilities

H. Lee Swanson
Leilani Sáez

Memory reflects the ability to encode, This chapter selectively reviews previous
process, and retrieve information to which and current working memory (WM) re-
one has been exposed. As a skill, it is insepa- search conducted by the first author (H. L.
rable from intellectual functioning and S.). The reader is also referred to Swanson
learning. Individuals deficient in memory (1989, 1991) for a comprehensive review of
skills, such as children and adults with short-term memory (STM) serial recall and
learning disabilities (LD), would be expect- neuropsychological studies spanning 10
ed to have difficulty on a variety of academ- years of research. In addition, a comprehen-
ic and cognitive tasks. Memory is linked to sive review of memory research as applied
performance in several academic (e.g., read- to LD has been provided elsewhere (e.g.,
ing) and cognitive areas (e.g., problem solv- Cooney & Swanson, 1987; Swanson et al.,
ing), and therefore, it is a critical area of fo- 1998). Before reviewing some of these stud-
cus in the field of LD for three reasons. ies, we provide the theoretical framework
and definitional criteria used for participant
1. It reflects applied cognition; that is, selection in most of the investigations.
memory functioning reflects all aspects of
learning. Theoretical Framework

2. Several studies suggest that the memory The research reported within this chapter
skills used by students with LD do not draws heavily on the tripartite view of WM
appear to exhaust, or even tap, their abil- put forth by Baddeley (e.g., Baddeley, 1986,
ities; therefore, we need to discover in- 1996; Baddeley & Logie, 1999). This view
structional procedures that can capitalize characterizes WM as comprising a central
on this underdeveloped potential. executive controlling system that interacts
with a set of two subsidiary storage systems:
3. Several cognitive intervention programs the speech-based phonological loop and the
that attempt to enhance the overall cog- visual–spatial sketch pad. The phonological
nition of persons with learning disabili- loop is responsible for the temporary stor-
ties rely on principles derived from mem-
ory research (see Swanson, Cooney, &
O’Shaughnessy, 1998, for a review).

182

Memory Difficulties in Children and Adults 183

age of verbal information; items are held lated to components of WM referred to in
within a phonological store of limited dura- Baddeley’s model (Baddeley & Logie, 1999) as
tion, maintained through the process of sub- the phonological loop and the executive sys-
vocal articulation. The visual–spatial sketch tem. Individuals with LD do not suffer all as-
pad is responsible for the storage of visu- pects of the phonological loop (e.g., they have
al–spatial information over brief periods relatively normal abilities in producing spon-
and plays a key role in the generation and taneous speech and have few difficulties in
manipulation of mental images. The central oral language comprehension) or the executive
executive is involved in the control and reg- system (e.g., they have normal abilities in
ulation of the WM system. According to planning and sustaining attention across
Baddeley and Logie (1999), it coordinates time). Those aspects of the phonological sys-
the two subordinate systems, focusing and tem that appear particularly faulty for individ-
switching attention, in addition to activat- uals with LD relates to accurate and speedy
ing representations within long-term memo- access of speech codes and those aspects of the
ry (LTM). Correlates in the neuropsycho- executive system that appear faulty are related
logical literature complement the tripartite to the concurrent monitoring of processing
structure, showing functional independence and storage demands and the suppression of
among the three systems (e.g., Smith & conflicting (e.g., irrelevant) information. Defi-
Jonides, 1999). ciencies in these operations influence perfor-
mances in academic domains (reading com-
Although assumptions we make about prehension, mathematics) that draw heavily
the WM model are consonant with Badde- upon those operations. Deficiencies in these
ley’s tripartite structure, we also incorporate operations are not due to academic achieve-
the notion that the central executive system ment or psychometric IQ because problems in
includes a mental workspace, distinct from WM capacity remain when achievement and
the two subordinate systems, with limited IQ are partialed out or controlled in a statisti-
resources. This assumption is also consis- cal analysis. In addition, our results show that
tent with Daneman and Carpenter’s model these limitations in WM are independent of
(1980), which views WM as reflecting a limitations in phonological processing. Chil-
combination of processing and storage com- dren with LD do well in some academic do-
ponents. mains because (a) those domains do not place
heavy demands on WM operations, and/or (b)
Within the aforementioned theoretical they compensate for WM limitations by in-
context, Swanson and Siegel (2001a) delin- creasing domain specific knowledge and/or
eated a causal model of LD drawing from their reliance on environmental support. (pp.
Swanson’s studies on WM, as well as those 107–108)
of others (e.g., Bull, Johnston, & Roy, 1999;
Chiappe, Hasher, & Siegel, 2000; De Beni, Definition of LD
Palladino, Pazzaglia, & Cornoldi, 1998; de
Jong, 1998; Passolunghi, Cornoldi, & De In our studies we define LD samples by their
Liberto, 1999; Siegel & Ryan, 1989). The primary academic difficulties in reading and
model states: mathematics and then attempt to isolate
problems in psychological processes. Partic-
Limitations in WM capacity have a neurologi- ipants with LD are operationally defined as
cal/biological base. These limitations are mul- those children and adults who have general
tifaceted as to the psychological operations IQ scores on standardized tests above 85
they influence. Limitations in WM capacity and who have scores below the 25th per-
cause LD. However, these limitations disrupt centile on a standardized reading and/or
only certain cognitive operations (a cognitive mathematics achievement measure. In some
operation involves manipulating, representing, studies, the criterion we have used for defin-
storing, and/or allocating of attentional re- ing low achievement is much lower than a
sources) when high demands are placed on cutoff score below the 25th percentile (i.e.,
processing. When performance demands on < 8th percentile). In general, the majority of
various tasks directly tax the WM capacity of the studies we cite involve LD samples with
individuals with LD, deficiencies related to primarily reading deficits, particularly word
the accessing of speech-based information recognition accuracy and reading compre-
and/or the monitoring of attentional processes hension. However, we recognize that read-
emerge. These two areas of deficiencies are re-

184 CAUSES AND BEHAVIORAL MANIFESTATIONS

ing problems are strongly correlated with readers with LD was at a statistically com-
other problems, such as mathematics. Thus, parable level to that of skilled readers in
LD samples with reading problems may suf- low-effort conditions. Furthermore, in the
fer problems in other academic domains lower-effort condition, a trend, in which
that share a common resource (e.g., lan- readers with LD recalled more words than
guage). did skilled readers. The results suggest that
after a difficult primary task, secondary task
Executive System performance is easier for skilled readers
than it is for LD readers.
This section reviews studies that have impli-
cated deficits in executive processing for Two additional experiments (Swanson,
children with LD. There are a number of 1984) replicated these findings. However,
cognitive activities assigned to the central the experiments primarily required the pro-
executive, including subsidiary memory sys- cessing of words. Thus, three additional ex-
tems coordination, control of encoding and periments were designed to reflect attention-
retrieval strategies, attention switching dur- al demands on both the verbal and
ing manipulation of material held in the ver- visual–spatial system. In Experiment 1
bal and visual–spatial systems, and LTM (Swanson, 1993a), a concurrent memory
knowledge retrieval (e.g., Baddeley, 1996). task, adapted from Baddeley, Lewis, El-
We hypothesize that the crucial component dridge, and Thomas (1984) was adminis-
of the central executive as it applies to LD is tered to LD and skilled readers. The task re-
controlled attention. Controlled attention is quired subjects to remember digit strings
defined as the capacity to maintain and hold (e.g., 9, 4, 1, 7, 5, and 2) while they concur-
relevant information in “the face of interfer- rently sorted blank cards, cards with pictures
ence or distraction” (Engle, Kane, & Tuhol- of nonverbal shapes, and cards with pictures
ski, 1999, p. 104). Executive processing of items that fit into semantic categories
constraints for participants with LD is in- (e.g., vehicles—car, bus, truck; clothing—
ferred from three outcomes: (1) poor perfor- dress, socks, belt). Demands on the central
mance on complex divided attention tasks, executive capacity system were manipulated
(2) weak monitoring ability, as exhibited in through the level of difficulty (three- vs. six-
the failure to suppress (inhibit) irrelevant in- digit strings) and type of sorting required
formation, and (3) depressed performance (e.g., nonverbal shapes, semantic categories,
across verbal and visual–spatial tasks that and blank cards). The results showed that
require concurrent storage and processing. LD readers could perform comparably to
chronological age (CA)–matched peers on
Complex Divided Attention verbal and visual–spatial sorting conditions
that involved low demands (i.e., three-digit
An early study by Swanson (1984) showed strings), and that only when the coordina-
that the mental allocation of attentional re- tion of tasks became more difficult (e.g., six-
sources of students with LD was more limit- digit strings) did ability group differences
ed than that of their nonlearning disabled emerge. More important, the results for the
(NLD) counterparts. Elementary-age stu- high-memory load condition indicated less
dents were given anagram problems to solve recall for LD readers than CA-matched (and
(the primary task). Upon the determination achievement-matched) peers during both
of an answer, LD and NLD participants verbal and nonverbal sorting. Because recall
were asked to recall words related to their performance was not restricted to a particu-
anagram solution (the secondary task). A lar storage system (i.e., verbal storage), one
significant group × cognitive effort interac- can infer that processes other than a lan-
tion emerged in the results. That is, no mat- guage-specific system accounted for the re-
ter what the organizational characteristics sults.
of words (i.e., semantic, nonsemantic, or
phonetic) were, words were better recalled Monitoring Activities
by skilled than by readers with LD under
high-effort conditions. However, recall of Our earlier work also investigated how ca-
pacity limits in the allocation of attention re-
sources were strategically handled. We inves-

Memory Difficulties in Children and Adults 185

tigated whether children with LD had In our laboratory we have also explored
greater trade-offs and weaker inhibition selective attention to word features within
strategies than average achievers on divided and across the cerebral hemispheres for chil-
attention tasks. Swanson (1989a) compared dren with LD. An abundance of experimen-
the performance of slow learners, children tal evidence points to an association be-
with LD, average achievers, and intellectual- tween left and right cerebral hemispheres
ly gifted children on a primary and sec- and variations in capacity demands. For ex-
ondary recall task. Children were asked to ample, the targeting of information in one
select one of two nouns (e.g., foot or dress) ear is assumed to consume resources that
presented in the context of two different would normally be used in processing in-
types of sentences, “base” sentences (e.g., formation in the competing ear (e.g., see
The woman wore a pretty _______) and Friedman & Polson, 1981). Given this as-
“elaborative” sentences (e.g., The woman sumption, Swanson and Cochran (1991)
wore a pretty _____ at the dance). The miss- compared 10-year-old average-achieving
ing word varied as a function of the mental children and same-age peers with LD on a
effort needed to ascertain a correct response. dichotic listening task.
For example, consider the base sentence
“The _____ went to school.” The response Participants were asked to recall words or-
set included either an easy- or low-effort ganized by semantic (e.g., red, black, green,
choice (e.g., children vs. house) or a hard- or and orange), phonological (e.g., sit, pit, and
high-effort choice (e.g., friends vs. children). hit), and orthographic (e.g., sun, same, seal,
Thus, children were provided with two and soft) features presented to either the left
words from which to choose the best word or right ear. The study included two experi-
for sentence completion (the primary task). ments. Experiment 1 compared free recall
Recall for chosen words, nonselected words, with different orienting instructions to word
and targeted adjectives was measured (the lists. For example, in the orienting condition,
secondary task). The results produced three children were told about the organizational
noteworthy findings. To simplify the report- structure of the words to be presented, such
ing of the results, only findings related to LD as to remember all of the words heard, “but
and average achievers will be highlighted. to specifically remember words that go with
_______” (e.g., colors–semantic feature ori-
First, as found in the previous studies entation), or “words that rhyme with ____”
(Swanson, 1984), no differences occurred (e.g., it–phonological feature orientation), or
between ability groups in secondary recall “words that start with the letter _____” (e.g.,
for the low-effort condition. However, high- s–orthographic feature orientation). For the
effort conditions favored the recall of sec- nonorienting condition, children were told
ondary words by average-achieving children to remember all words, but no mention was
when compared to children with LD. made of the distinctive organizational fea-
tures of the words. Experiment 2 extended
Second, recall insertions (the proportion Experiment 1 by implementing a cued-recall
of nontargeted words incorrectly recalled condition. In both experiments, children
between the secondary and the central task) were told they would hear someone talking
were significantly higher in children with through a set of earphones but that they
LD than in average achievers. Thus, chil- should only pay attention to what was said
dren with LD had greater difficulty inhibit- in one of the ears (i.e., the targeted ear). The
ing nontargeted words than did average children were told that when they stopped
achievers. hearing the information in both ears, they
were to tell the experimenter all the words
Finally, clear differences emerged between they could remember.
ability groups in the prioritization of re-
sources (i.e., in the direction of the correla- In both experiments, NLD children had
tions between the primary and secondary higher levels of targeted and nontargeted re-
tasks). Trade-offs, in the form of low posi- call compared to children with LD. More im-
tive or negative correlations, emerged for portant, ability group differences emerged in
students with LD between the primary and how specific word features were selectively
secondary task, whereas there was a sharing attended to. The selective attention index fo-
of resources (i.e., positive correlations) for cused on the targeted words in comparison
average children.

186 CAUSES AND BEHAVIORAL MANIFESTATIONS

to the background words (targeted word re- without cues). The study also explored
call minus background word recall from oth- whether ability groups vary in their WM
er lists within the targeted ear), as well as spans as a function of the type of WM task
background items in the contralateral ear. across age.
Regardless of word features, whether com-
peting word features were presented (within- This study included two verbal WM mea-
ear or across-ear conditions), or whether re- sures that required the processing of acousti-
trieval conditions were cued or noncued, cally familiar rhyming words (phonological
selective attention scores of readers with LD task, e.g., car, star, bar, and far) or the pro-
were smaller (the difference score between cessing of semantically related words (se-
targeted items and nontargeted items was mantic task, e.g., pear, apple, prune; car, bus,
closer to zero) than that of NLD readers. and truck), and a visual–spatial WM mea-
Thus, when compared with children with sure (visual-matrix task) that required the se-
LD, NLD children were more likely to ignore quencing of dots on a matrix. The general
irrelevant information in the competing con- findings of the Swanson and Sachse-Lee
ditions. Taken together, the results of this (2001a) study were that (1) young adults
study, as well as those of three earlier dichot- (i.e., 20 and 35 years old) performed better
ic listening studies (Swanson, 1986; Swan- than did children and older adults (i.e., 55-
son & Mullen, 1983) suggest that children year-olds), (2) skilled readers performed bet-
with disabilities suffer processing deficits re- ter than readers with LD in all processing
lated to resource monitoring, regardless of conditions, and (3) the gain condition im-
the type of word features, retrieval condi- proved span performance from initial condi-
tions, or ear presentation. tions, but performance declined when main-
tenance conditions were administered.
Combined Processing and Storage Demands However, these findings were qualified by
age × ability group interactions related to
Recent studies (e.g., Swanson, 1994; Swan- memory conditions (initial, gain, mainte-
son & Ashbaker, 2000; Swanson, Ashbaker, nance) and type of WM task (verbal vs. visu-
& Lee, 1996; Swanson & Sachse-Lee, al–spatial). Both skilled readers and those
2001b) on executive processing have includ- with LD showed continuous growth in ver-
ed tasks that follow the format of Daneman bal and visual–spatial WM that peaked at
and Carpenter’s Sentence Span measure, a approximately 20 and 35 years of age. The
task strongly related to student achievement results clearly showed that the readers with
(see Daneman & Merikle, 1996, for a re- LD had less WM recall than did skilled read-
view). These studies have consistently found ers across a range of age groups on tasks that
readers with LD to be more deficient than involved the processing of phonological, vi-
skilled readers in WM performance using sual–spatial, and semantic information. Al-
this task format, which presumably taps though WM performance levels of skilled
central executive processes related to “up- readers and those with LD were comparable
dating” (Miyake, Friedman, Emerson, Witz- at some adult ages on the phonological and
ki, & Howerter, 2000). Updating requires visual–spatial WM measures, comparable
monitoring and coding of information for performance levels were not sustained across
relevance to the task at hand and then ap- all adult age groups. Thus, the study provid-
propriately revising items held in WM. A ed little evidence that WM skills of readers
cross-sectional study (Swanson & Sachse- with LD “catch up” with skilled readers as
Lee, 2001a) compared skilled readers and they age, suggesting that a deficit model
readers with LD across a broad age span. rather than a developmental lag model best
The study compared six age groups (7, 10, captures such readers’ age-related perfor-
13, 20, 35, 55) on phonological, semantic, mance.
and visual–spatial WM measures adminis-
tered under conditions referred to in Swan- From the foregoing findings, as well as
son and colleagues (1996): initial (no others (Swanson, 1992, 1993d; Swanson et
probes or cues), gain (cues that bring per- al., 1996), we have found evidence of do-
formance to an asymptotic level), and main- main-general processing deficits in children
tenance conditions (asymptotic conditions and adults with LD, suggestive of executive
system involvement. Because of the poten-
tial for misinterpretations related to these

Memory Difficulties in Children and Adults 187

findings (e.g., the paradox between domain- to a general system from that related to a
specific deficits commonly attributed to LD specific WM system we have found results,
with the finding that they have domain gen- which substantiate the notion of between-
eral processing deficits), we will clarify our systems independence (Swanson & Alexan-
interpretation of the results (also see Swan- der, 1997; Wilson & Swanson, 2001). For
son & Siegel, 2001a, p. 111, for discussion). example, Wilson and Swanson (2001) sta-
The perplexity in adequately linking a do- tistically controlled the domain-specific
main-general processing deficit to LD is re- variance from verbal and visual–spatial
lated to the confusion in the literature as to WM performance of participants with math
what such a system entails. Cognitive opera- disabilities and participants without math
tions independent of, or not directly moder- disabilities across a broad age span. We par-
ated by, verbal or visual–spatial skills have titioned the variance in WM performance,
been referred to as domain-general process- via structural equation modeling, by creat-
es or the central executive system. This sys- ing two first-order factors (the verbal WM
tem reflects a diversity of activities (12 are tasks reflected factor 1 and the visual–spa-
listed in Swanson & Siegel, 2001b, such as tial WM tasks reflected factor 2) to capture
planning, allocating attention, accessing in- unique variance, and a single second higher-
formation from LTM, etc.). These processes order factor that reflected shared variance
draw from several regions of the brain but or domain-general performance among all
are associated primarily with the prefrontal the tasks. When the ability groups were
cortex (e.g., Smith & Jonides, 1999). Thus, compared on these factor scores, groups
domain-general processing is perhaps a mis- without math disabilities were superior to
nomer because operations that cut across those with math disabilities on factor scores
verbal and visual spatial skills are multifac- that included variance partitioned into do-
eted. main-general WM, verbal WM, and visu-
al–spatial WM. Thus, it appears, at least in
We have addressed some of the alterna- the area of mathematics, that ability group
tive explanations to our findings on execu- differences emerge in domain-specific sys-
tive processing—for example, deficits are tems and in a general executive system of
due to attention-deficit/hyperactivity disor- WM.
der (ADHD), low intelligence, domain-
specific knowledge, low-order processes In terms of interdependence among do-
(such as phonological coding), and so on main-general and isolated processes, Swan-
(see Swanson & Siegel, 2001a, for a review son and Alexander (1997) examined the in-
of studies). We find (as do independent lab- terrelationship among cognitive processes in
oratories) that (1) children with normal IQ predicting word recognition and reading
can have executive processing deficits, (2) comprehension performance of readers with
some readers with LD suffer executive pro- LD. The correlation among phonological,
cessing deficits that do not overlap with the orthographic, semantic, metacognitive, ver-
deficits attributed to children with ADHD bal/visual–spatial WM measures, and read-
(e.g., WM deficits emerge in readers with ing performance were examined in readers
LD but not in children with ADHD of nor- with LD and skilled readers, ages 7 to 12.
mal intelligence), (3) significant differences The study yielded the following important
in WM remain between LD and NLD par- results: (1) readers with LD were deficient
ticipants when achievement, domain-specif- in all cognitive processes when compared to
ic knowledge, and psychometric intelligence skilled readers, but these differences were
are partialed from the analysis, and (4) the not a reflection of IQ scores; (2) readers
causal basis of attention between children with LD were deficient compared to skilled
with LD and ADHD (as well as manifesta- readers in a general factor primarily com-
tions) differs. posed of verbal and visual–spatial WM
measures and unique components, suggest-
Of course, the foregoing comments raise ing that reading ability group differences
the issue of whether deficits in a domain- emerge in both general and specific (modu-
general system can operate independently of lar) processing; (3) the general WM factor
deficits in a specific system, such as the best predicted reading comprehension for
phonological loop (to be discussed later). both skilled and LD readers’ groups; and (4)
When we have partitioned variance related

188 CAUSES AND BEHAVIORAL MANIFESTATIONS

phonological awareness best predicted culty in forming and accessing phonological
skilled readers’ pseudo-word reading, representations impairs the ability to re-
whereas the general WM factor best pre- trieve verbal information. Interestingly, this
dicted pseudo-word performance of readers phonological impairment does not appear
with LD. Overall, Swanson and Alexander’s to have broad effects on general ability
study showed that verbal and visual–spatial apart from the developmental consequences
WM tasks share variance with a common on language-related functions (Hohen &
system but also have some unique variance Stevenson, 1999). One language function
related to a specific system. Furthermore, alluded to in the literature is verbal memory.
both the general system and specific phono- Before reviewing the evidence on verbal
logical system predicted reading. memory, the overlap and distinctions be-
tween verbal STM and verbal WM must be
Summary addressed. Most studies that compare the
performance of skilled readers and those
We have selectively reviewed studies sug- with LD assume that verbal STM measures
gesting that WM deficits of children with capture a subset of WM performance (i.e.,
LD may, depending on the task and materi- the use and/or operation of the phonologi-
als, reflect problems in the executive system. cal loop). Some authors have even suggested
These problems appear to be related to at- that the phonological loop may be referred
tention allocation, and the shifting and up- to as verbal STM (e.g., Dempster, 1985) be-
dating of information in WM. These prob- cause it involves two major components dis-
lems are not isolated to the verbal domain. cussed in the STM literature: a speech-based
phonological input store and a rehearsal
It is important to note that students with process (see Baddeley, 1986, for review).
LD are not deficient on all executive pro- Our research has addressed some of this
cessing activities. For example, although confusion. We briefly review our research
planning (such as mapping out a sequence here, which suggests that some distinction
of moves) is considered a component of the between the two concepts is necessary.
executive system (however, see, e.g., Miyake
et al., 2000, p. 90), we have not found abili- STM-A Review
ty group differences between LD and NLD
students on such tasks (see Swanson, 1988, A 1998 meta-analysis was conducted to
1993c, for studies that examine perfor- quantitatively summarize the experimental
mance on other executive processing tasks, literature (O’Shaughnessy & Swanson,
such as the Tower of Hanoi, Combinatorial, 1998) for studies within the last 30 years
Picture Arrangement or Pendulum tasks). that met the following criteria: (1) directly
compared readers with LD with average
The Phonological System readers, as identified on a standardized
reading measure and at least one STM mea-
In Baddeley’s model (1986), the articulatory sure; (2) reported standardized reading
or phonological loop is specialized for the scores, which indicated that students with
retention of verbal information over short LD were at least 1 year below grade level;
periods. It is composed of both a phonolog- and (3) reported intelligence scores for stu-
ical store, which holds information in dents with LD that were in the average
phonological form, and a rehearsal process, range (i.e., 85–115). Although the search
which maintains representations in the resulted in approximately 155 articles on
phonological store (see Baddeley, Gather- immediate memory and learning disabili-
cole, & Papagano, 1998, for an extensive ties, only 38 studies (24.5%) met the crite-
review). A substantial number of studies ria for inclusion. For comparisons in this
support the notion that children with LD synthesis, an effect size (ES) magnitude of
experience memory deficits in processes re- –0.20, was considered small, –0.50 was
lated to the phonological loop (e.g., see moderate, and –0.90 was considered a large
Siegel, 1993, for a review of studies show- ES difference in favor of NLD when com-
ing deficits in readers with LD related to pared to LD participants. A summary of the
phonological representations). That is, diffi- results follows:

Memory Difficulties in Children and Adults 189

1. The group with LD performed poorly –0.68). That is, studies that used either
on STM tasks requiring memorization of word recognition alone or reading compre-
verbal information compared to the NLD hension alone resulted in similar effect sizes
group (ES = –0.68). STM tasks that em- of ES = –0.59 and ES = –0.56, respectively.
ployed stimuli that could not easily be Thus, studies that used both word recogni-
named, such as abstract shapes, did not pro- tion and reading comprehension as the cri-
duce large differences between NLD and LD teria for assessing reading skills resulted in a
readers (ES = –0.15). sample of subjects with LD who demon-
strated more severe immediate memory
2. STM tasks requiring readers with LD problems.
to recall exact sequences of verbal stimuli,
such as words or digits, immediately after a In summary, this quantitative analysis of
series presentation yielded a much greater the literature clearly showed that children
overall mean ES (ES = –0.80) than nonver- and adults with LD are inferior to their
bal serial recall tasks (ES = –0.17). Thus, se- counterparts on measures of STM in which
rial recall performance with verbal material familiar items such as letters, words, and
of students with LD was over three-quarters numbers, and unfamiliar items such as ab-
of a standard deviation below that of aver- stract shapes were recalled.
age readers. However, when memory per-
formance with nonverbal stimuli was com- Verbal STM versus Verbal WM
pared, the difference equaled less than
one-quarter of a standard deviation in favor Are verbal STM deficits synonymous with
of the average readers. deficits in verbal WM? Results from our lab
have suggested that the tasks differ in subtle
3. The overall mean ES for studies that ways. Simply stated, some children with LD
provided instructions in mnemonic strate- perform poorly on tasks that require accu-
gies (e.g., rehearsal and sorting items into rate and/or speedy recognition/recall of let-
groups) prior to recall and used verbal stim- ter and number strings or real words and
uli was –0.54; the overall mean ES for stud- pseudo-words. Tasks, such as these, which
ies using verbal stimuli without mnemonic have a “read in and read out” quality to
strategy instruction was –0.71. These results them (i.e., place few demands on LTM to
indicate a lingering difference in memory infer or transform the information) reflect
performance between LD and NLD stu- STM. One common link among these tasks
dents in spite of mnemonic training. That is, is the ability to store and/or access the
although the memory performance of stu- sound structure of language (phonological
dents with reading disabilities improved processing). However, some children with
with training in mnemonic strategies, LD also do poorly on tasks that place de-
70.5% of average readers still scored above mands on attentional capacity. Everyday ex-
the mean of the group with LD. amples of verbal WM processing include
holding a person’s address in mind while lis-
4. STM tasks that involved auditory pre- tening to directions regarding location, lis-
sentation of verbal stimuli resulted in an tening to a sequence of events in a story
overall mean ES of –0.70, whereas those while trying to understand what the story
that involved visual presentation of verbal means, and locating a sequence of land-
stimuli resulted in an overall mean ES of marks on a map while determining the cor-
–0.66. Thus, the inferior verbal STM per- rect route.
formance of readers with LD appears to be
unrelated to the modality in which a stimu- We tested whether the operations related
lus is received. to STM and WM operated independently of
one another. A study by Swanson and Ash-
5. STM tasks that involved the visual baker (2000) compared readers with LD and
presentation of nonverbal stimuli, such as skilled readers and younger achievement-
abstract shapes, resulted in an overall mean matched children on a battery of WM and
ES of –0.15. STM tests to assess executive and phonolog-
ical processing. Measures of the executive
The largest overall ES was exhibited in system were modeled after Daneman and
studies that used both word recognition and Carpenter’s (1980) WM tasks (i.e., tasks de-
reading comprehension as a means of distin-
guishing between subjects with LD (ES =

190 CAUSES AND BEHAVIORAL MANIFESTATIONS

manding the coordination of both process- measures. Participants were divided into
ing and storage), whereas measures of the four ability groups: High Comprehen-
phonological system included those that re- sion/High Word Recognition, Low Compre-
lated to articulation speed, digit span, and hension/High Word Recognition, High
word span. The Swanson and Ashbaker Comprehension/Low Word Recognition,
study yielded two important results. First, al- and Low Comprehension/Low Word Rec-
though the reading group with LD was infe- ognition. The results were straightforward:
rior to skilled readers in WM, verbal STM, WM measures were related primarily to
and articulation speed, the differences in ver- reading comprehension, whereas phonolog-
bal STM and WM revealed little relation ical STM measures were related primarily to
with articulation speed. That is, reading- reading recognition. Most critically, because
related differences on WM and STM mea- no significant interaction emerged, the re-
sures remained when articulation speed was sults further indicated that the comorbid
partialed from the analysis. These reading- group (i.e., children low in both comprehen-
group differences were pervasive across ver- sion and word recognition) had combined
bal and visual–spatial WM tasks, even when memory deficits. That is, WM deficits were
the influence of verbal STM was removed, reflective of the poor comprehension–only
suggesting that reading-group differences are group and STM deficits were reflective of
domain general. Second, WM tasks and ver- the poor recognition–only group.
bal STM tasks contributed unique, or inde-
pendent, variance to word recognition and Why the distinction between STM and
reading comprehension beyond articulation WM? We argue that WM tasks require the
speed. These results are consistent with those active monitoring of events. Monitoring of
of Daneman and Carpenter (1980) and oth- events within memory is distinguishable
ers (e.g., Engle, Tuholski, Laughlin, & Con- from simple attention to stimuli held in
way, 1999), who have argued that verbal STM. There are many mnemonic situations
STM tasks and WM tasks are inherently dif- in which a stimulus in memory is attended
ferent, and although phonological coding to and the other stimuli exist as a back-
might be important to recall in STM, it may ground—that is, they are not the center of
not be a critical factor in WM tasks. current awareness. These situations, in our
opinion, do not challenge monitoring. Mon-
The foregoing findings from Swanson itoring within WM implies attention to the
and Ashbaker’s study are consistent with stimulus that is currently under considera-
early work on samples with LD (Swanson, tion together with active consideration (i.e.,
1994; Swanson & Berninger, 1995). In a attention) of several other stimuli whose
1994 study, Swanson tested whether STM current status is essential for the decision to
and WM contributed unique variance to be made.
academic achievement in children and
adults with learning disabilities. Swanson Summary
found that STM and WM tasks loaded on
different factors. Further, both of these fac- There is evidence that participants with LD
tors contributed unique variance to reading suffer deficits in the phonological system. A
and mathematics performance. A study by substrate of this system may contribute to
Swanson and Berninger also examined po- problems in verbal WM that are indepen-
tential differences between STM and WM dent of problems in verbal STM. In addi-
by testing whether STM and WM account- tion, these problems in verbal WM are not
ed for different cognitive profiles in readers removed by partialing out the influence of
with LD. Swanson and Berninger used a verbal articulation speed, reading compre-
double-dissociation design to compare chil- hension, or IQ scores.
dren deficient in reading comprehension
(based on scores from the Passage Compre- Visual–Spatial Processing
hension subtest of the Woodcock Reading
Mastery Test) and/or word recognition In Baddeley’s (1986) model, the visual–
(based on scores from the Word Identifica- spatial sketch pad is specialized for the pro-
tion subtest of the Woodcock Reading Mas- cessing and storage of visual material, spatial
tery Test) on WM and phonological STM

Memory Difficulties in Children and Adults 191

material, or both, and for linguistic informa- with LD on visual–spatial WM tasks rela-
tion that can be recoded into imaginal forms. tive to verbal WM tasks. The Swanson and
The literature linking LD to visual–spatial colleagues results showed demonstrations of
memory deficits is mixed. For example, sev- deficiency in both verbal and visual–spatial
eral studies in the STM literature suggest WM performance for readers with LD for
that visual STM of children with LD is intact gain and maintenance conditions (i.e., after
(see Swanson et al., 1998, for a comprehen- receiving cues to aid recall) that were not
sive review). Some studies have found that evident for initial condition performance
visual–spatial WM in students with LD is in- (i.e., without cue assistance). Furthermore,
tact when compared with their same-age these performance differences held up when
counterparts (e.g., Swanson et al., 1996, Ex- verbal STM scores (see Experiment 2) were
periment 1), whereas others suggest prob- partialed from the analysis, providing addi-
lems in various visual–spatial tasks (Swan- tional support for a general system involve-
son et al., 1996, Experiment 2). Most studies ment that cuts across both verbal and visu-
indicate, however, that greater problems in al–spatial WM tasks, which influences the
performance are more likely to occur on ver- performance of readers with LD when pro-
bal than visual–spatial WM tasks. For exam- cessing demands are high.
ple, Swanson, Mink, and Bocian (1999)
found by partialing out the influence of ver- Summary
bal IQ via regression analysis, that students
with reading disabilities were inferior in per- The evidence for whether children with LD
formance to slow learners (i.e., garden-vari- have any particular advantage on visual–
ety poor readers) on visual–spatial and ver- spatial WM tasks, when compared to their
bal WM measures. That is, although normal-achieving counterparts, appears to
children with a specific reading disability fluctuate with processing demands. Swan-
demonstrated a greater deficit on the verbal son (2000) proposed a model that may ac-
WM task than the visual–spatial WM task, count for these mixed findings.
performance on both types of tasks was infe-
rior to other poor learning groups when ver- WM and Achievement
bal IQ was statistically controlled.
The importance of the executive and phono-
It may not be the case, however that dif- logical system in predicting reading perfor-
ferentiation occurs between reading and mance is related to age. As children age, the
math subgroups when visual–spatial WM executive system may play a more primary
measures are used (Swanson, 1993b, role in separating good and poor readers
1993d). For example, Swanson (1993d) than at the younger ages. Furthermore, dif-
found that 10-year-old children with LD ficulties in the executive system may devel-
who suffered either math problems or read- op independent of the specific difficulties in
ing problems could not be clearly differenti- reading of readers with LD. That is, based
ated by performance for verbal or on the foregoing observations, we speculate
visual–spatial WM measures. This study in- that as children age, skilled readers have rel-
cluded six tasks that assessed verbal WM atively higher WM capacity than do readers
(Rhyming, Story Retelling, Auditory Digit with LD, and therefore will have more
Sequence, Phrase Sequencing, Semantic As- available resources related to the executive
sociation, Semantic Categorization) and five system with which to perform tasks, regard-
tasks that assessed visual–spatial WM (Vi- less of the nature of the task. We hold that
sual Matrix, Picture Sequence, Mapping & people are poor readers because they have a
Directions, Spatial Organization, Nonver- small WM capacity, and this capacity is not
bal Sequencing). In general, Swanson found entirely specific to reading. That is, poor
that children with arithmetic disabilities readers have more limited WM than skilled
performed as low as children with reading readers, not as a consequence of poor read-
disabilities across verbal and visual–spatial ing skills but because they have less WM ca-
WM tasks. pacity available for performing reading and
nonreading tasks. Of course, individuals
In a later study, Swanson and colleagues
(1996) tested whether there might be poten-
tial performance advantages for readers

192 CAUSES AND BEHAVIORAL MANIFESTATIONS

will vary in the efficiency (e.g., speed of acti- from the analysis. Furthermore, the attenu-
vation) of their mental operations on some ating effect of executive processing on read-
specific tasks, but other things being equal, ing comprehension did not appear to be due
it is our belief that high-WM-capacity indi- to phonological processing speed or LTM.
viduals still have more attentional resources
available to them than do low-WM-capaci- Swanson was also interested in determin-
ty individuals. ing whether there were some fundamental
processing differences between readers with
Our research shows that WM plays an im- LD and skilled readers that superseded their
portant role in predicting academic perfor- problems in reading comprehension. He an-
mance. In general, we find that both the ex- alyzed the processing variables as a function
ecutive system and the phonological loop of reading conditions by reframing the com-
predict performance for complex domains parison groups in terms of the regression-
(e.g., reading comprehension) and low-order based design outlined by Stanovich and
domains (e.g., calculation). We briefly re- Siegel (1994). When reading comprehension
view studies here that support these conclu- was statistically controlled, the results indi-
sions in the areas of reading comprehension, cated significant differences in WM and pro-
problem solving, writing, and computation. cessing speed for phonological information
between LD and skilled readers, independent
Complex Cognition of their reading comprehension levels.

READING COMPREHENSION WRITING

Several of our studies have shown that WM Unfortunately, only one of our studies fo-
accounts for significant variance in compre- cuses specifically on the relationship be-
hension performance of readers with LD tween WM and writing in samples with LD.
(e.g., Swanson, 1999a; Swanson & Alexan- In this study (Swanson & Berninger, 1996),
der, 1997; Swanson & Sachse-Lee, 2001b). children from a university clinic were ad-
One study (Swanson, 1999a) in particular ministered a standardized WM battery
identifies those components of WM that are (Swanson Cognitive Performance Test
most important to reading comprehension. [S-CPT]; Swanson, 1995), Test of Written
In this study, Swanson (1999a) found signif- Language, Wide Range Achievement Test,
icant differences between students with LD and Peabody Achievement Test. The corre-
and peers matched for age and nonverbal lation analysis yielded three important out-
IQ on measures of phonological processing comes. First, visual–spatial and verbal WM
accuracy (i.e., phonemic deletion, digit re- were significantly correlated with writing,
call, phonological choice, and pseudo-word reading recognition, and reading compre-
repetition), phonological processing speed hension (r ranged from .39 to .79 across
(i.e., timed responses from phonemic dele- measures). Second, the influence of both vi-
tion, digit recall, phonological choice, and sual–spatial and verbal WM was pervasive
pseudo-word repetition task), LTM accura- across all writing tasks. These relationships
cy (i.e., orthographic choice, semantic remained even when the influence of vocab-
choice, and vocabulary), LTM time (i.e., ulary was removed from the analysis. Final-
timed response from orthographic choice, ly, the contribution of WM tasks to writing
semantic choice, and vocabulary), and exec- was not a function of reading ability. That
utive processing (i.e., Sentence Span, Count- is, the correlations between WM and writ-
ing Span, and Visual-matrix tasks). The re- ing were maintained when reading was en-
sults showed that the CA-matched group tered first into a regression equation.
outperformed the reading group with LD,
whereas the readers with LD were compara- WORD PROBLEM SOLVING
ble to reading level–matched children. The
important findings, however, were that the There is limited information on the contri-
significant relationship between executive bution of WM to the problem-solving accu-
processing and reading comprehension was racy for students with LD. In one of the few
maintained when LTM and phonological studies conducted (Swanson & Sachse-Lee,
processing composite scores were partialed 2001b), children with LD, approximately

Memory Difficulties in Children and Adults 193

12 years of age, were compared with CA- processes (i.e., semantic and orthographic),
matched and younger achievement-matched may be qualitatively different in predicting
(matched for reading comprehension and academic performance of participants with
math computation skill) children on mea- LD in some age groups (Swanson &
sures of verbal and visual–spatial WM, Alexander, 1997). We summarize our obser-
phonological processing, components of vations across various age ranges as fol-
problem solving, and word problem-solving lows:
accuracy. In this study, children were pre-
sented arithmetic word problems orally and For skilled readers and those with LD,
asked a series of questions about the various ages 6 to 75: (1) both domain-general WM
processes they would use to solve the task. and domain-specific WM are related to
They also solved problems that required word recognition (Swanson, 1996; Swanson
them to apply algorithms related to subtrac- & Sachse, 2001a, 2001b); (2) age-related
tion, addition, and multiplication. The changes in WM in skilled readers are best
study produced a number of important find- explained by a capacity rather than a pro-
ings. First, phonological processing, verbal cessing-efficiency model (Swanson, 1999b);
WM, and visual WM each contributed (3) readers with LD, defined by word-recog-
unique variance to solution accuracy. More nition deficits, experience WM deficits into
important, both verbal and visual–spatial adulthood (Swanson & Sachse-Lee, 2001a);
WM performance predicted solution accu- and (4) WM performance of readers with
racy when phonological processing was LD is changeable via probing or cued proce-
controlled by entering it first in the regres- dures; however, significant differences re-
sion model. Furthermore, the results main between reading groups because of
showed that performance on phonological, greater domain-general capacity limitations
verbal WM, and visual–spatial WM mea- in readers with LD (Swanson et al., 1996;
sures were statistically comparable in their Swanson & Sachse-Lee, 2001b).
contribution to ability group differences in
solution accuracy. For children ages 9 to 15, we find that (1)
domain-general WM differences between
Low-Order Tasks: Arithmetic Computation skilled readers and those with LD are not
eliminated when reading comprehension is
Wilson and Swanson (2001) examined the partialed from the analysis (Swanson,
relationship between verbal and visual– 1999a); (2) phonological and executive
spatial WM and mathematical computation processes are equally important in predict-
skill in children and adults. Participants, ing reading comprehension, as well as prob-
skilled and disabled in mathematics, ranged lem solving (Swanson, 1999a; Swanson &
in age from 11 to 52 years. Our major find- Alexander, 1997; Swanson & Sachse-Lee,
ing was that groups without math disabili- 2001b); (3) deficits in executive processing
ties were superior to those with math dis- and reading comprehension are only par-
abilities on factor scores that included tially mediated by the phonological system
variance partitioned into domain-general or LTM (Swanson, 1999a); (4) domain-
and domain-specific verbal and visual– specific deficits emerge on verbal WM tasks
spatial WM. These results were the same on initial (noncued) conditions, but deficits
when both the linear and quadratic compo- in both verbal and visual–spatial WM
nents of age were partialed from the analy- emerge as processing demands increase un-
sis. The results also showed that both the der gain (cued) and maintenance (high de-
verbal and visual–spatial WM composite mand) conditions (Swanson et al., 1996);
scores predicted mathematics performance. and (5) WM deficits related to poor readers
Furthermore, these results held when read- are best attributed to a capacity, not pro-
ing ability was partialed from the analysis. cessing-efficiency, model (Swanson, 1994;
Swanson & Sachse-Lee, 2001a).
Lifespan WM Development and Achievement
Based on these observations, it appears to
The interrelationship between phonological us that the phonological system may play its
and executive processes, as well as related primary role in predicting reading recogni-
tion and comprehension (accuracy and flu-
ency) for early elementary school learning.
Between the ages of 9 and 16, the executive

194 CAUSES AND BEHAVIORAL MANIFESTATIONS

system and the phonological system play 1998; Passolunghi et al., 1999; Siegel &
equal, as well as independent, roles in pre- Ryan, 1989). For example, Stanovich and
dicting word reading and reading compre- Siegel (1994) showed that readers with LD
hension accuracy and fluency (Swanson, (as well as poor readers) suffer deficits in
1999a; Swanson & Alexander, 1997). We both verbal WM and verbal STM, even af-
also suggest that for adult poor readers ter reading ability was controlled. In their
(drawing on our pilot work, and Engle, study, Stanovich and Siegel amalgamated a
Cantor, & Carullo, 1992), executive sample that consisted of more than 1,500
processes play a more important role in pre- children from 7 to 16 years of age. Children
dicting reading, especially reading compre- were classified into poor readers who also
hension, than does the phonological system. had low IQs (< 86), poor readers who had
That is, we find with older participants (ju- average IQs (> 90), and those children who
nior high through adult ages) that a general had average scores in IQ and reading. They
WM system, not specific to reading, sepa- compared these groups on various process-
rates skilled readers and those with LD ing measures when reading scores were en-
(Swanson, 1999a; Swanson et al., 1999). tered first into a regression model. The
Our previous research with older samples processing variables of interest were perfor-
shows that skilled readers have relatively mance on STM rhyming and nonrhyming
higher WM capacity than do readers with tasks and WM tasks that included words or
LD, even when reading comprehension numbers. The STM tasks included letter
(Swanson, 1999a), word recognition (Swan- strings that rhymed (e.g., B, C, D, G, P, T,
son et al., 1999), word recognition and IQ and V) and those that did not rhyme (e.g.,
(Ransby & Swanson, 2001; Swanson & H, K, L, Q, R, S, and W). The verbal WM
Sachse-Lee, 2001b), and articulation speed task (WM sentences) included an oral pre-
(Swanson & Ashbaker, 2000) are removed sentation of sentences, which increased in
from analysis. Depending on age and read- set size. Children were asked to supply miss-
ing fluency, we assume that although the ex- ing words (e.g., People go see monkeys in a
ecutive system plays a role relaying the re- ______) and later recall those words sup-
sults of lower-level phonological analyses plied. The number WM task required chil-
upward through the language system, it also dren to count yellow dots from a series of
serves as a monitoring system independent blue and yellow dots arranged in irregular
of those skills (e.g., Baddeley, 1986). patterns. The patterns were on cards, and
the number of cards within each set in-
Summary creased in number. For each set, the child
was to recall the count of yellow dots for
We have selectively reviewed studies related each card after all cards had been presented.
to various components of WM. There is evi- When readers with LD (those children with
dence in the literature indicating both the a discrepancy between IQ and reading) and
phonological loop and the executive system poor readers (those children whose IQ
as sources of deficit for participants with matched their low reading level) were com-
LD. Either one or both of these components pared to skilled readers, a significant advan-
play a significant role in predicting complex tage was found in favor of skilled readers’
cognitive activities such as reading compre- recall on the verbal WM and STM tasks.
hension, arithmetic problem solving, and No differences in recall were found between
writing, as well as some basic skills (e.g., the two poor reading groups.
arithmetic computation).
Further support that readers with LD suf-
Independent Researchers fer problems in both verbal STM and verbal
WM is found in a life-span study of Siegel
There is empirical support, independent (1994). Her study included some 1,200 in-
from our lab, that an LD may reflect a fun- dividuals, ages 6 to 49. These individuals
damental deficit in WM for both children were presented tasks related to word recog-
and adults (e.g., Bull et al., 1999; Chiappe nition, pseudo-word decoding, reading
et al., 2000; De Beni et al., 1998; de Jong, comprehension, and WM, as well as a STM
task requiring the recall of rhyming and
nonrhyming letters. The results indicated

Memory Difficulties in Children and Adults 195

gradual growth in WM skills from ages 6 to strained by his or her lack of knowledge rel-
19, with a gradual decline after adolescence. evant to the task. Thus, different children
On the memory tasks, across most age lev- may follow different developmental routes
els, individuals with reading disabilities per- to overcome their utilization deficiencies.
formed at a significantly lower level than in-
dividuals with normal reading skills. Thus, Some practical concepts and principles
readers with LD experienced deficits on from memory research can serve as guide-
WM and verbal STM tasks across child- lines for the instruction of students with
hood, adolescence, and adulthood. learning disabilities (see Swanson et al.,
1998, for a review of eight instructional
Practical Applications principles). Three major principles are par-
ticularly germane to our discussion.
Prior to 1989, memory research in LD was
strongly influenced by the hypothesis that Strategy Instruction Must Operate on the Law
variations in memory performance are root- of Parsimony
ed in the children’s acquisition of mnemonic
strategies (Cooney & Swanson, 1987; Because of capacity demands, particular at-
Swanson et al., 1998). Strategies are deliber- tention must be paid to developing strate-
ate, consciously applied procedures that aid gies that are parsimonious and not placing
in the storage and subsequent retrieval of excessive demands on children’s attentional
information. Research in the last 10 years resources. A number of multiple-component
has moved in a different direction, toward packages of strategy instruction have been
an analysis of nonstrategic processes that suggested to improve functioning of chil-
are not necessarily consciously applied. The dren with LD. These components have usu-
major motivation behind this movement has ally encompassed some of the following:
been that important aspects of memory skimming, imagining, drawing, elaborating,
performance are often disassociated with paraphrasing, using mnemonics, accessing
changes in mnemonic strategies. The most prior knowledge, reviewing, orienting to
striking evidence has come from strategy- critical features, and so on. No doubt, there
oriented research, which shows that differ- are some positive aspects to these strategy
ences between children with and without packages in that:
learning disabilities remain after the use of
an optimal strategy (i.e., a strategy shown 1. These programs are an advance over
to be advantageous in the majority of some of the studies that are seen in the
studies). LD literature as rather simple or “quick-
fix” strategies (e.g., rehearsal or catego-
It is clear from our synthesis of the litera- rization to improve performances).
ture (Swanson et al., 1998), however, that
children with LD can benefit from mnemon- 2. These programs promote a domain skill
ic instruction when training is sufficiently and have a certain metacognitive embell-
rigorous. However, strategy training does ishment about them.
not eliminate ability group differences be-
tween students with LD and their peers with 3. The best of these programs involved (a)
disabilities in a multitude of situations. teaching a few strategies well rather than
Some of the causes of strategy ineffective- superficially, (b) teaching students to
ness or utilization deficiencies may be relat- monitor their performance, (c) teaching
ed to individual differences in information students when and where to use the
processing capacity (i.e., children without strategy to enhance generalization, (d)
LD benefit more from the strategy than do teaching strategies as an integrated part
children with LD) and/or a particular level of an existing curriculum, and (e) teach-
of strategy effectiveness may have different ing that included a great deal of super-
causes in different children. A child with vised student practice and feedback.
LD, for example, may be unable to benefit
from a strategy because of his or her limited The difficulty of such packages, however,
capacity, whereas another child may be con- at least in terms of theory, is that little is
known about which components best pre-
dict student performance, nor do they readi-
ly permit one to determine why the strategy

196 CAUSES AND BEHAVIORAL MANIFESTATIONS

worked. The multiple-component ap- fundamental problems of children and
proaches that are typically found in a num- adults with LD. Further, these WM prob-
ber of strategy intervention studies on LD lems are related to difficulties in reading,
must be carefully contrasted with a compo- mathematics, and perhaps writing. Students
nent-analysis approach that involves the with LD in reading and/or math demon-
systematic combination of instructional strate WM deficits related to the phonologi-
components known to have an additive ef- cal loop, a component of WM that special-
fect on performance. izes in the retention of speech-based
information. This system is of service in
Memory Strategies in Relation to a Student’s complex cognition, such as reading compre-
Knowledge Base and Capacity hension, problem solving, and writing. Re-
search over the last decade also finds that
Memory capacity seems to increase with de- children and adults with LD are clearly dis-
velopment; a number of factors have the po- advantaged in situations that place high de-
tential to contribute to this overall effect. mands on a limited-capacity system. These
With development, the number of compo- constraints on the limited capacity of chil-
nent processes increase in speed, with faster dren and adults with LD mainifest them-
processes generally consuming less effort selves as deficits in controlled attentional
than slower processes, thereby increasing processing (e.g., monitoring limited re-
capacity (i.e., there is a functional gain in sources, suppressing conflicting informa-
capacity with increasing efficiency of pro- tion, and updating information). Further,
cessing). Older children are likely to have these deficits are sustained when articula-
more and better organized prior knowledge, tion speed, phonological processing, and
which can reduce the total number of infor- verbal STM are removed from analyses.
mation chunks to be processed and decrease
the amount of effort to retrieve information Acknowledgments
from LTM. Because of these developmental
relationships, as well as the constraints that This chapter draws primarily from Swanson (1991,
underlie development, this could play a role 1992), Swanson and Siegel (2001a, 2001b), Swan-
in strategy effectiveness. son, Cooney, and O’Shaughnessy (1998), and
Swanson and Sachse-Lee (2001b), and the reader is
Comparable Memory Strategy May Not referred to those sources for more complete infor-
Eliminate Performance Differences mation. Appreciation is given to Joel Levin for his
critique of this earlier work.
Several studies have indicated that residual
differences remain between ability groups References
even when groups are instructed and/or pre-
vented from strategy use (Swanson et al., Baddeley, A. D. (1986). Working memory. London:
1998, for a review). For example, Swanson Oxford University Press.
(1983) found that the recall of a group with
LD did not improve from baseline level Baddeley, A. D. (1996). Exploring the central exec-
when trained with rehearsal strategies. They utive. Quarterly Journal of Experimental Psy-
recalled less than normally-achieving peers, chology: Human Experimental Psychology,
although the groups were comparable in the 49(1), 5–28.
various types of strategies used. The results
support the notion that groups of children Baddeley, A. D., Gathercole, S. E., & Papagano, C.
with different learning histories may contin- (1998). The phonological loop as a language
ue to learn differently, even when the groups learning device. Psychological Review, 105(1),
are equated in terms of strategy use. 158–173.

Conclusions Baddeley, A. D., & Logie, R. H. (1999). Working
memory: The multiple component model. In A.
Our conclusions from approximately two Miyake & P. Shah (Eds.), Models of working
decades of research are that WM deficits are memory: Mechanisms of active maintenance and
executive control (pp. 28–61). New York: Cam-
bridge University Press.

Baddeley, A. D., Lewis, V., Eldridge, M., & Thom-
son, N. (1984). Attention and retrieval from
long-term memory. Journal of Experimental Psy-
chology: General, 113(4), 518–540.

Memory Difficulties in Children and Adults 197

Bull, R., Johnston, R. S., & Roy, J. A. (1999). Ex- variable analysis. Cognitive Psychology, 41(1),
ploring the roles of the visual–spatial sketch pad 49–100.
and central executive in children’s arithmetical O’Shaughnessy, T., & Swanson, H. L. (1998). Do
skills: Views from cognition and developmental immediate memory deficits in students with
neuropsychology. Developmental Neuropsychol- learning disabilities in reading reflect a develop-
ogy, 15(3), 421–442. mental lag or deficit? A selective meta-analysis of
the literature. Learning Disability Quarterly,
Chiappe, P., Hasher, L., & Siegel, L. S. (2000). 21(2), 123–148.
Working memory, inhibitory control, and reading Passolunghi, M. C., Cornoldi, C., & De Liberto, S.
disability. Memory and Cognition, 28(1), 8–17. (1999). Working memory and intrusions of irrel-
evant information in a group of specific poor
Cooney, J. B., & Swanson, H. L. (1987). Memory problem solvers. Memory and Cognition, 27(5),
and learning disabilities: An overview. In H. L. 779–790.
Swanson (Ed.), Advances in learning and behav- Ransby, M., & Swanson, H. L. (2001). Reading
ioral disabilities: Memory and learning disabilities comprehension skills of young adults with child-
(Suppl. 2, pp. 1–40). Greenwich, CT: JAI Press. hood diagnosis of dyslexia. Unpublished manu-
script, University of California, Riverside.
Daneman, M., & Carpenter, P. A. (1980). Individ- Siegel, L. S. (1993). Phonological processing deficits
ual differences in working memory and reading. as a basis for reading disabilities. Developmental
Journal of Verbal Learning and Verbal Behavior, Review, 13(3), 246–257.
19(4), 450–466. Siegel, L. S. (1994). Working memory and reading:
A life-span perspective. International Journal of
Daneman, M., & Merikle, P. M. (1996). Working Behavioral Development, 17(1), 109–124.
memory and language comprehension: A meta- Siegel, L. S., & Ryan, E. B. (1989). The develop-
analysis. Psychonomic Bulletin and Review, 3(4), ment of working memory in normally achieving
442–433. and subtypes of learning disabled children. Child
Development, 60(4), 973–980.
De Beni, R., Palladino, P., Pazzaglia, F., & Cornol- Smith, E. E., & Jonides, J. (1999). Storage and ex-
di, C. (1998). Increases in intrusion errors and ecutive processes in the frontal lobes. Science,
working memory deficit of poor comprehenders. 283(5408), 1657–1661.
Quarterly Journal of Experimental Psychology: Stanovich, K. E., & Siegel, L. (1994). Phenotypic
Human Experimental Psychology, 51(2), performances profile of children with reading dis-
305–320. abilities: A regression-based test of the phonolog-
ical-core variable-difference model. Journal of
de Jong, P. (1998). Working memory deficits of Education Psychology, 86(1), 24–53.
reading disabled children. Journal of Experimen- Swanson, H. L. (1984). Effects of cognitive effort
tal Child Psychology, 70(2), 75–95. and word distinctiveness on learning disabled
readers’ recall. Journal of Educational Psycholo-
Dempster, F. (1985). Short-term memory develop- gy, 76(5), 894–908.
ment in childhood and adolescence. In C. Brainerd Swanson, H. L. (1986). Do semantic memory defi-
& M. Presseley (Eds.), Basic processes in memory ciencies underlie disabled readers encoding
(pp. 209–248). New York: Springer-Verlag. processes? Journal of Experimental Child Psy-
chology, 41(3), 461–488.
Engle, R. W., Cantor, J., & Carullo, J. J. (1992). In- Swanson, H. L. (1989). Verbal coding deficits in
dividual differences in working memory and learning disabled readers: A multiple stage model.
comprehension: A test of four hypotheses. Jour- Educational Psychology Review, 1(3), 235–277.
nal of Experimental Psychology: Learning, Mem- Swanson, H. L. (1991). Learning disabilities, dis-
ory and Cognition, 18(5), 972–992. tinctive encoding and hemispheric resources: An
information processing perspective. In J. E.
Engle, R. W., Kane, M. J., & Tuholski, S. (1999). Obrzut & G. W. Hynd (Eds.), Neurological foun-
Individual differences in working memory capac- dations of learning disabilities: A handbook of is-
ity and what they tell us about controlled atten- sues, methods, and practice (pp. 241–280). San
tion, general fluid intelligence, and functions of Diego, CA: Academic Press.
the prefrontal cortex. In A. Miyake & P. Shah Swanson, H. L. (1992). Generality and modifica-
(Eds.), Models of working memory: Mechanisms tion of working memory among skilled and less
of active maintenance and executive control (pp. skilled readers. Journal of Educational Psycholo-
102–134). Cambridge, UK: Cambridge Universi- gy, 84(4), 473–488.
ty Press. Swanson, H. L. (1993a). Executive processing in
learning disabled readers. Intelligence, 17(2),
Engle, R. W., Tuholski, S. W., Laughlin, J. E., & 117–149.
Conway, A. R. (1999). Working memory, short- Swanson, H. L. (1993b). Individual differences in
term memory, and general fluid intelligence: A la- working memory: A model testing and subgroup
tent-variable approach. Journal of Experimental analysis of learning-disabled and skilled readers.
Psychology: General, 128(3), 309–331. Intelligence, 17(3), 285–332.

Friedman, A., & Polson, M. V. (1981). Hemi-
spheres as independent resource systems: Limit-
ed-capacity processing and cerebral specializa-
tion. Journal of Experimental Psychology:
Human Perception and Performance, 7(5),
1031–1058.

Miyake, A., Friedman, N. P., Emerson, M. J., Witz-
ki, A. H., & Howerter, A. (2000). The unity and
diversity of executive functions and their contri-
butions to complex “frontal lobe” tasks: A latent

198 CAUSES AND BEHAVIORAL MANIFESTATIONS

Swanson, H. L. (1993c). Working memory in learn- Swanson, H. L., & Berninger, V. W. (1996). Individ-
ing disability subgroups. Journal of Experimental ual differences in children writing: A function of
Child Psychology, 56(1), 87–114. working memory or reading or both processes?
Reading and Writing: An Interdisciplinary Jour-
Swanson, H. L. (1994). Short-term memory and nal, 8(4), 357–383.
working memory: Do both contribute to our un-
derstanding of academic achievement in children Swanson, H. L., & Cochran, K. (1991). Learning
and adults with learning disabilities? Journal of disabilities, distinctive encoding, and hemispheric
Learning Disabilities, 27(1), 34–50. resources. Brain and Language, 40(2), 202–230.

Swanson, H. L. (1995). Swanson Cognitive Process- Swanson, H. L., Cooney, J. B., & O’Shaughnessy, T.
ing Test (S-CPT): A dynamic assessment measure (1998). Memory and learning disabilities. In B. Y.
(p. 122). Austin, TX: Pro-Ed. Wong (Ed.), Understanding learning disabilities
(2nd ed.) San Diego, CA: Academic Press.
Swanson, H. L. (1996). Individual and age-related
differences in children’s working memory. Memo- Swanson, H. L., Mink, J., & Bocian, K. M. (1999).
ry and Cognition, 24(1), 70–82. Cognitive processing deficits in poor readers with
symptoms of reading disabilities and ADHD:
Swanson, H. L. (1999a). Reading comprehension More alike than different? Journal of Education-
and working memory in skilled readers: Is the al Psychology, 91(2), 321–333.
phonological loop more important than the exec-
utive system? Journal of Experimental Child Psy- Swanson, H. L., & Mullen, R. (1983). Hemisphere
chology, 72(1), 1–31. specialization in learning disabled readers’ recall
as a function of age and level of processing. Jour-
Swanson, H. L. (1999b). What develops in working nal of Experimental Child Psychology, 35(3),
memory? A life span perspective. Developmental 457–477.
Psychology, 35(4), 986–1000.
Swanson, H. L., & Sachse-Lee, C. (2001a). Learn-
Swanson, H. L. (2000). Are working memory ing disabled readers’ working memory: What
deficits in readers with learning disabilities hard does or does not develop? Unpublished manu-
to change? Journal of Learning Disabilities, script, University of California, Riverside.
33(6), 551–566.
Swanson, H. L., & Sachse-Lee, C. (2001b). Mathe-
Swanson, H. L., & Alexander, J. (1997). Cognitive matical problem solving and working memory in
processes as predictors of word recognition and children with learning disabilities: Both executive
reading comprehension in learning disabled and and phonological processes are important. Jour-
skilled readers: Revisiting the specificity hypothe- nal of Experimental Child Psychology,79(3),
sis. Journal of Educational Psychology, 89(1), 294–321.
128–158.
Swanson, H. L., & Siegel, L. (2001a). Elaborating
Swanson, H. L., & Ashbaker, M. (2000). Working on working memory and learning disabilities: A
memory, Short-term memory, articulation speed, reply to commentators. Issues in Education: Con-
word recognition, and reading comprehension in tributions from Educational Psychology, 7(1),
learning disabled readers: Executive and/or artic- 107–129.
ulatory system? Intelligence, 28(1), 1–30.
Swanson, H. L., & Siegel, L. (2001b). Learning dis-
Swanson, H. L., Ashbaker, M., & Lee, C. (1996). abilities as a working memory deficit . Issues in
Working-memory in learning disabled readers as Education: Contributions from Educational Psy-
a function of processing demands. Journal of chology, 7(1), 1–48.
Child Experimental Psychology, 61(3), 242–275.
Wilson, K., & Swanson, H. L. (2001). Are mathe-
Swanson, H. L., & Berninger, V. W. (1995). The matics disabilities due to a domain-general or do-
role of working memory in skilled and less skilled main-specific working memory deficit? Journal
readers’ word comprehension. Intelligence, 21, of Learning Disabilities, 34(3), 237–248.
83–108.

12

Learning Disabilities in Arithmetic:
Problem-Solving Differences
and Cognitive Deficits

David C. Geary

The complexity of the field of mathematics The use of these models and methods to
makes the study of any associated learning guide the study of children with LD has re-
disability daunting. In theory, a mathemati- vealed a consistent pattern of cognitive
cal learning disability can result from strengths and weaknesses. These studies
deficits in the ability to represent or process suggest that most children with LD are nor-
information used in one or all of the many mal (i.e., performance is similar to academi-
areas of mathematics (e.g., arithmetic and cally normal peers) or only slightly delayed
geometry), or in one or a set of individual in the development of number concepts
domains (e.g., theorems vs. graphing) within (Geary, Hamson, & Hoard, 2000; Gross-
each of these areas (Russell & Ginsburg, Tsur, Manor, & Shalev, 1996). At the same
1984). One approach that can be used to fo- time, several studies have shown that many
cus the search for any such learning disabili- children with LD do not understand certain
ty (LD) is to apply the models and methods counting concepts (Geary, Bow-Thomas, &
used to study mathematical development in Yao, 1992; Geary, Hoard, & Hamson,
academically normal children to the study of 1999), and many studies have revealed that
children with poor achievement in mathe- these children have a variety of deficits in
matics (e.g., Geary & Brown, 1991). Unfor- simple arithmetic (Ackerman & Dykman,
tunately, for most mathematical domains, 1995; Barrouillet, Fayol, & Lathuli(re,
such as geometry and algebra, not enough is 1997; Bull & Johnston, 1997; Garnett, &
known about the normal development of Fleischner, 1983; Geary, Brown, & Sama-
the associated competencies to provide a ranayake, 1991; Geary, Widaman, Little, &
systematic framework for the study of LD. Cormier, 1987; Jordan & Hanich, 2000;
Theoretical models and experimental meth- Jordan, Levine, & Huttenlocher, 1995; Jor-
ods are, however, sufficiently well developed dan & Montani, 1997; Ostad, 1997,
in the areas of number, counting, and simple 1998a; Räsänen & Ahonen, 1995; Rourke,
arithmetic to provide such a framework 1993; Svenson & Broquist, 1975). The
(Briars & Siegler, 1984; Geary, 1994; Gel- deficits in the basic arithmetical competen-
man & Meck, 1983; Siegler, 1996; Siegler & cies of children with LD (hereafter, arith-
Shrager, 1984). metical disability, or AD) have been found

199

200 CAUSES AND BEHAVIORAL MANIFESTATIONS

in studies conducted in the United States and mathematical topics, whereas children
(e.g., Garnett & Fleischner, 1983), Israel with AD often have severe deficits in some
(Shalev, Manor, & Gross-Tsur, 1993), and of these areas and average or better compe-
several European nations (Ostad, 2000; tencies in others. The result of averaging
Svenson & Broquist, 1975). The difficulties across these topics is a level of performance
children with AD have solving simple arith- (e.g., at the 20th percentile) that overesti-
metic problems are the focus of the next sec- mates competencies in some areas and un-
tion. The second section provides an derestimates them in others.
overview of research on the cognitive and
potential neural mechanisms contributing to PREVALENCE AND ETIOLOGY
these deficits.
Large-scale epidemiological studies of the
Arithmetical Learning Disability prevalence of AD have not been conducted,
although several smaller-scale studies that
The first part presents background informa- included more than 300 children from a
tion on the diagnosis, prevalence, and etiol- well-defined population have (e.g., all
ogy of AD; the second provides an overview fourth-graders in an urban school district).
of our research program in AD. Measures, designed from neuropsychologi-
cal studies of number and arithmetic deficits
Background following brain injury, are more sensitive to
AD than are standard achievement tests
DIAGNOSIS have been used in these studies. They have
been conducted in the United States (Badi-
Unfortunately, measures that are specifically an, 1983), Europe (Kosc, 1974), and Israel
designed to diagnose AD are not available. (Gross-Tsur et al., 1996; Shalev et al.,
As a result, most researchers rely on stan- 2001). The findings across studies suggest
dardized achievement tests, often in combi- that 5 to 7% of school-age children exhibit
nation with measures of intelligence (IQ). A some form of AD. Ostad (1998a) described
score lower than the 20th or 25th percentile several related studies of elementary-school
on a mathematics achievement test com- children conducted in Norway during the
bined with a low-average or higher IQ score 1950s. These studies revealed that 8% of
are typical criteria for diagnosing AD (e.g., the children likely had some form of AD.
Geary, Hamson, & Hoard, 2000; Gross- Thus, the best estimate, at this time, is that
Tsur et al., 1996). There are, however, two between 5 and 8% of children have some
difficulties with these criteria. form of AD.

First, if applied in only a single academic Some children with AD exhibit comor-
year, the criteria often lead to a number of bid attention-deficit/hyperactivity disorder
false positives, that is identifying children as (ADHD) or reading disability (RD). The
AD who in fact have no cognitive deficits most comprehensive of these studies indi-
and typically show improved achievement cated that 26% of the children with AD had
scores in later grades (Geary, 1990; Geary et symptoms of ADHD, and 17% had RD
al., 1991): We have found that most chil- (Gross-Tsur et al., 1996). Badian (1983), in
dren who meet these criteria across two suc- contrast, found that nearly half of the chil-
cessive grades do appear to have some form dren with AD also showed comorbid read-
of cognitive deficit and AD. Second, the cut- ing difficulties, and Ostad (1998a) found
off of the 25th percentile on a mathematics that just over half of the children with AD
achievement test does not fit with the esti- had a comorbid spelling disability (SD). At
mation, described below, that between 5 this time, it appears that children with AD
and 8% of children have some form of AD. constitute at least two different subgroups,
The discrepancy results from the nature of those with only difficulties in arithmetic and
standardized achievement tests and the of- those with comorbid learning disabilities in
ten rather specific deficits of children with other areas. The latter most typically in-
AD. By design, standardized achievement volve language-related deficits, that is RD
tests sample a broad range of arithmetical and (or) SD (for related discussion, see
Geary, 1993; Geary & Hoard, 2001).

Learning Disabilities in Arithmetic 201

As with other forms of LD, twin and fa- sion equations. Here, statistical models rep-
milial studies suggest both genetic and envi- resenting the approaches potentially used
ronmental contributions to both forms of while problem solving, such as counting or
AD. In a twin study, Light and DeFries memory retrieval, are fit to RT patterns. As
(1995) provided evidence that the same an example, if children counted both ad-
genes may contribute to AD and RD and dends in the problem, starting from one,
thus their comorbidity in many children. then RTs should increase linearly with the
Shalev and her colleagues (2001) studied fa- sum of the problem, and the value of the
milial patterns of AD, excluding children raw regression slope should be consistent
with comorbid ADHD or RD. The results with estimates of the speed with which chil-
showed that family members (e.g., parents dren count implicitly (for general discus-
and siblings) of children with AD are 10 sion, see Geary, Widaman, & Little, 1986;
times more likely to be diagnosed with AD Widaman, Geary, Cormier, & Little, 1989).
than are members of the general population.
In the first study in which we used these
Research Program techniques, second-, fourth-, and sixth-grade
children with AD were compared to their
As noted, performance on standardized academically normal peers (Geary et al.,
achievement tests does not provide informa- 1987). The RT patterns suggested that chil-
tion on the strengths and weaknesses of indi- dren with AD differed from other children in
vidual children within the broad domain of terms of the form and frequency of counting
mathematics, only relative performance av- strategies used to solve simple addition prob-
eraged across all the assessed mathematical lems. For a problem such as 9 + 5, academi-
subareas. The only means to better under- cally normal second-grade children typically
stand learning in mathematics, as well as stated “nine” and then counted, “ten, eleven,
learning disabilities, is to focus research ef- twelve, thirteen, fourteen” (termed the
forts on circumscribed mathematical do- counting-on procedure; Fuson, 1982). Chil-
mains and to use methods that enable a fine- dren with AD tended to count, starting from
grained assessment of performance in each one (the counting-all procedure). Cross-
domain. To this end, our initial efforts were sectional comparisons suggested that most
focused on simple addition and were guided academically normal children gradually
by Ashcraft’s (e.g., Ashcraft & Battaglia, switched from counting to direct retrieval of
1978) information-processing studies of cog- the answer, whereas most children with AD
nitive arithmetic and later by Siegler’s (1996; did not make this transition (Geary et al.,
Siegler & Shrager, 1984) strategy choice 1987). Rather, they still counted to solve ad-
model of cognitive development. dition problems, although many of these
children appeared to use the more efficient
INFORMATION PROCESSING counting-on procedure in later grades. The
same pattern has recently been reported in a
Beginning with Svenson and Broquist’s study that contrasted the subtraction compe-
(1975) study more than 25 years ago and tencies of children with AD with those of
continuing today, the information-process- other children (Ostad, 2000).
ing approach has guided much of the cogni-
tive research on children with AD. In our STRATEGY CHOICE
first study of children with AD (Geary et al.,
1987), we employed the reaction time (RT) Soon after beginning data collection for this
techniques developed by Groen and Park- first study (i.e., Geary et al., 1987), I read
man (1972) and later elaborated by Siegler and Shrager’s (1984) strategy-choice
Ashcraft and his colleagues (for a review, model of arithmetical and later more gener-
see Ashcraft, 1995). Here, simple arithmetic al cognitive development (Siegler, 1996).
problems, such as 3 + 2 = 4 or 9 + 5 = 14, The approach was based on a combination
are presented on a computer monitor. The of the RT methods used by cognitive psy-
child indicates by button push whether the chologists, such as Ashcraft (1995), and di-
presented answer is correct. The resulting rect observation of problem solving used by
RTs are then analyzed by means of regres- educational researchers, such as Carpenter
and Moser (1984). The goal was not only to

202 CAUSES AND BEHAVIORAL MANIFESTATIONS

describe the types of strategies children used problem is solved through counting, the
to solve simple arithmetic problems but also strength of the association between 5 + 3
to determine the mechanisms that governed and the generated answer (typically 8) in-
whether a child would use one strategy creases. Eventually children automatically
(e.g., counting) or another (e.g., retrieval) to retrieve 8, or whatever answer has been
solve each particular problem. A related most frequently generated, when presented
goal was (and still is) to understand devel- with 5 + 3. So, if an answer is not readily re-
opmental change in the mechanisms govern- trieved, due to a low associative strength be-
ing strategy choices. Although the model tween the problem and potential answers,
has been elaborated over the years, the basic children resort to some form of backup
mechanisms are the same (Siegler, 1996). strategy to complete problem solving; Table
12.1 provides a description of retrieval and
In all domains that have been studied, in- the primary backup strategies used to solve
cluding arithmetic, children use a mix of simple addition problems (Geary, 1994).
strategies during problem solving. In solving
arithmetic problems, children will some- Our first study that followed Siegler and
times retrieve the answer to solve one prob- Shrager’s (1984) method replicated and ex-
lem and count to solve the next problem. tended their basic findings (e.g., the relation
Memory retrieval is assumed to be based on between RTs and strategy choices) by
an associative relationship between the pre- demonstrating that individual differences in
sented problem and all potential answers to strategy choices were related to individual
the problem. These associations appear to performance differences on standard arith-
develop as children use other types of strate- metical achievement and ability tests (Geary
gies during problem solving. Counting on to & Burlingham-Dubree, 1989). Strong per-
solving 5 + 3, for instance, appears to result formance on the strategy-choice task (e.g.,
in the formation of a long-term memory as- fast and accurate strategy execution) was
sociation between this problem and the an- predictive of above-average performance on
swer generated by the count. Each time the the achievement and ability measures (see

TABLE 12.1. Strategies Used to Solve Simple Addition Problems

Strategy Description Example

Finger counting: A number of fingers representing To solve 2 + 3, two fingers are lifted
Counting all the augend and addend are lifted on one hand and three on the other.
and then counted starting from 1. All uplifted fingers are then counted.

Finger counting: A number of fingers representing To solve 2 + 3, two fingers are lifted
Counting on the augend and addend are lifted on one hand and three on the other.
and then counted starting from the The count starts with “three” and
larger number. proceeds “four, five.”

Verbal counting: As above, but counting is done To solve 2 + 3, the child counts
Counting all without the use of fingers. (explicitly or implicitly), “one, two,
three, four, five.”

Verbal counting: As above, but counting is done To solve 2 + 3, the child states “three”
Counting on without the use of fingers. and then counts “four, five.”

Retrieval Direct retrieval of a basic fact The child states an answer quickly
from long-term memory. and without signs of counting;
typically stating “just knew it.”

Decomposition Retrieval of a partial sum and To solve 2 + 3, the child first retrieves
counting on. the answer to 2 + 2 and then counts
up to “five”

Note. See Geary (1994) for further discussion and illustration.

Learning Disabilities in Arithmetic 203

also Siegler, 1988). The results also indicat- (Geary, 1990). The children in the improved
ed that the methods and theoretical model group had below-average mathematics
proposed by Siegler and Shrager would like- achievement scores at the end of kinder-
ly provide an excellent framework for guid- garten but average or better scores at the
ing the study of children with AD. The ap- end of first grade. Children in the no-change
proach was followed in an initial study of group had below-average scores at both as-
first-grade children with AD (Geary, 1990), sessments. There were no differences com-
many of whom were reassessed in second paring children in the improved group to
grade (Geary et al., 1991), and another children in an academically normal control
study of fourth-graders that included chil- group in terms of strategy choices, error
dren with AD as well as gifted children rates, or RTs. This pattern was subsequently
(Geary & Brown, 1991). replicated (Geary, Hamson, & Hoard,
2000), which bolstered the conclusion that
All these studies involved the use of a vari- the initial low mathematics achievement of
ant of Siegler and Shrager’s (1984) strategy children in the improved group was not
assessment task for simple addition, which likely to be due to any form of cognitive
provides information on problem-solving deficit or AD. Their initial poor perfor-
strategies, accuracy of strategy use, and ac- mance may have been due to inattention
companying RTs. Here, simple addition during test taking or poor early math in-
problems, such as 5 + 6, are presented one at struction. In any case, the improved group
a time on a computer monitor. The child is is not considered further.
instructed to solve the problem using what-
ever means is easiest for him or her. With the An unexpected finding was that children
completion of problem solving, the child im- with AD (i.e., the no-change group) did not
mediately speaks the answer into a voice- differ from their academically normal peers
activated relay which triggers an internal in terms of the mix of strategies used to
timing device in the computer (for recording solve simple addition problems, as shown in
RTs). The child’s problem solving, such as Table 12.2. Differences were, however,
whether fingers are used, is monitored and found in percentage of retrieval and count-
recorded by the experimenter. The child is ing errors and in the use of the counting-on
then asked to describe how he or she got the procedure, all favoring the academically
answer. High-levels of agreement between normal group. Error and RT patterns for
experimenter observation and child reports problems on which an answer was retrieved
(typically > 90% of trials), along with a con- also differed comparing the academically
sistency between these reports and associat- normal and AD groups. For children with
ed RTs patterns, attest to the utility of the AD, the distribution of RTs was unusual.
method (e.g., Geary, 1990; Siegler, 1987). The pattern was not similar to that found in
younger, academically normal children and
In the first study based on this approach, seemed to reflect a highly variable speed of
first-grade children with AD were divided fact retrieval. The interpretation of this pat-
into two groups: improved and no change

TABLE 12.2. Addition Strategy Characteristics Comparing Academically Normal Children and
Children with AD

Trials on which _______E_r_r_o_r_s_(_%__)_____ ___C__o_u_n_t_in_g__o_n__(%__)___
__s_t_r_a_te_g_y__u_s_ed__(_%__)__

Strategy Normal AD Normal AD Normal AD

Counting 56 21 50 100 69
fingers 60 64 7 31 100 86
35 26 5 22 ——
Verbal
counting

Retrieval

Note. Data based on Geary (1990).

204 CAUSES AND BEHAVIORAL MANIFESTATIONS

tern was that it suggested “an anomalous supporting fact retrieval (see also Garnett &
long-term memory representation of addi- Fleischner, 1983). Subsequent studies using
tion facts” (Geary, 1990, p. 379). Further the same model and methods have been
analyses revealed greater variability in the conducted in the United States by Jordan
speed with which children with AD execut- and her colleagues (Jordan & Hanich,
ed other numerical processes, such as num- 2000; Jordan et al., 1995; Jordan & Mon-
ber articulation, in comparison to their aca- tani, 1997) and by Ostad and others in Eu-
demically normal peers. rope (e.g., Barrouillet et al., 1997; Ostad,
1997, 1998b, 2000). These studies have
A year later, many of these children were confirmed the differences in counting-strate-
reassessed on the strategy-choice task and gy use and retrieval deficit and extended the
were administered a numerical digit span domain of study to subtraction, multiplica-
task (Geary et al., 1991). In keeping with tion, and word problems, among others
models of arithmetical development (Ash- (e.g., Hanich, Jordan, Kaplan, & Dick,
craft, 1982), the academically normal chil- 2001; Ostad, 1998b). Subsequent research
dren showed an across-grade shift from re- has also led to the discovery of at least two
liance on verbal counting (56% to 44% different forms of AD (Jordan & Montani,
across years) to retrieval (39% to 51% 1997)—that is, AD with no comorbid forms
across years). The academically normal chil- of LD and AD with comorbid RD or other
dren also showed faster retrieval times and forms of language-related disorder (e.g.,
fewer retrieval errors (6% to 2%), compar- SD). Subsequent research has further
ing grade 1 to grade 2 performance. As with demonstrated that the differences compar-
the previously described cross-sectional ing children with AD to other children can-
study (Geary et al., 1987), the children with not be attributed to differences in IQ (Geary
AD showed no developmental change in the et al., 1999; Geary, Hamson, & Hoard,
mix of problem-solving strategies (e.g., 26% 2000; McLean & Hitch, 1999).
to 25% retrieval across years) or in retrieval
accuracy (e.g., 18% to 16% retrieval errors). Cognitive Mechanisms and Deficits

The children with AD did, however, show The aforementioned studies led to attempts
improvement in how effectively they used to discern the nature of the cognitive deficits
counting to solve addition problems. In underlying some children’s developmental
grade 2, they almost always used the count- delay in the use of counting procedures and
ing-on procedure when using a counting their difficulties in representing and/or re-
strategy to problem solve and showed a trieving basic facts from long-term memory.
marked reduction in the proportion of Cognitive studies combined with research
counting errors (e.g., from 49% to 10% for on arithmetical difficulties associated with
finger counting). Analyses of RTs indicated brain injury (i.e., dyscalculia) and with be-
that the academically normal children havioral genetic studies of individual differ-
showed faster counting comparing grade 2 ences in mathematical abilities provided
to grade 1, but the children with AD clues as to possible sources of the problem-
showed no change in counting speed. solving characteristics of children with AD.
Again, the distribution of retrieval RTs of The integration of these literatures resulted
children with AD differed from that of their in a taxonomy of three general subtypes of
academically normal peers. An important mathematical disability (MD), procedural,
discovery was that the pattern of retrieval semantic memory, and visuospatial (Geary,
RTs of children with AD was similar to that 1993). Table 12.3 shows the defining char-
found with children who had suffered from acteristics of these subtypes.
an early (before 8 years of age) lesion to the
left hemisphere or subcortical regions The development delay in the use of
(Ashcraft, Yamashita, & Aram, 1992). counting procedures while solving arith-
metic problems is subsumed under the more
This pattern of developmental change general procedural subtype of MD. Deficits
suggested that the children with AD were in the retrieval of basic arithmetic facts is
developmentally delayed in terms of their the defining feature of the semantic memory
ability to use counting to solve arithmetic
problems and were fundamentally different
from normal children in the mechanisms

Learning Disabilities in Arithmetic 205

TABLE 12.3. Subtypes of Learning Disabilities in Mathematics

Procedural subtype

Cognitive and performance features
A. Relatively frequent use of developmentally immature procedures (i.e., the use of procedures that
are more commonly used by younger, academically normal children)
B. Frequent errors in the execution of procedures
C. Poor understanding of the concepts underlying procedural use
D. Difficulties sequencing the multiple steps in complex procedures

Neuropsychological features
Unclear, although some data suggest an association with left-hemispheric dysfunction and in some
cases (especially for feature D above) a prefrontal dysfunction

Genetic features
Unclear

Developmental features
Appears, in many cases, to represent a developmental delay (i.e., performance is similar to that of
younger, academically normal children, and often improves across age and grade)

Relation to RD
Unclear

Semantic memory subtype

Cognitive and performance features
A. Difficulties retrieving mathematical facts, such as answers to simple arithmetic problems
B. What facts are retrieved, there is a high error rate
C. For arithmetic, retrieval errors are often associates of numbers in the problem (e.g., retrieving 4
to 2 + 3 = ?; 4 is the counting-string associate that follows 2, 3)
D. RTs for correct retrieval are unsystematic

Neuropsychological features
A. Appears to be associated with left-hemispheric dysfunction, possibly the posterior regions for one
form of retrieval deficit and the prefrontal regions for another
B. Possible subcortical involvement, such as the basal ganglia

Genetic features
Appears to be a heritable deficit

Developmental features
Appears to represent a developmental difference (i.e., cognitive and performance features differ from
that of younger, academically normal children, and do not change substantively across age or grade)

Relation to RD
Appears to occur with phonetic forms of RD

Visuospatial subtype

Cognitive and performance features
A. Difficulties in spatially representing numerical and other forms of mathematical information and
relationships
B. Frequent misinterpretation or misunderstanding of spatially represented information

Neuropsychological features
Appears to be associated with right-hemispheric dysfunction, in particular, posterior regions of the
right hemisphere, although the parietal cortex of the left hemisphere may be implicated as well

Genetic features
Unclear, although the cognitive and performance features are common with certain genetic disorders
(e.g., Turner’s syndrome)

Developmental features
Unclear

Relation to RD
Does not appear to be related

Note. Adapted from Geary (1993, 2000).

206 CAUSES AND BEHAVIORAL MANIFESTATIONS

subtype. Still, semantic memory deficits (Hitch & McAuley, 1991; McLean &
should affect other mathematical competen- Hitch, 1999; Siegel & Ryan, 1989: Swan-
cies that are based on the retrieval of facts, son, 1993). There are several ways in which
such as recalling prime numbers. In any a working-memory deficit could affect the
case, the respective sections that follow de- procedural competencies of children with
scribe research on the arithmetical problem AD.
solving of children with AD in terms of the
three forms of MD subtype, and as related As an example, children with AD (and
to the cognitive and neural systems that younger, academically normal children) ap-
may underlie their problem-solving charac- pear to use finger counting as a working-
teristics (e.g., the retrieval deficit). memory aid, in that fingers appear to help
these children to keep track of the counting
Procedural Deficits process (Geary, 1990). In particular, repre-
senting the problem addends on fingers and
Much of the research on children with AD then using fingers to note the counting se-
has focused on their use of counting proce- quence should greatly reduce the working-
dures to solve simple arithmetic problems. memory demands of the counting process.
As described, when they solve such prob- Working memory may also contribute to
lems, children with AD often commit more the tendency of children with AD to under-
errors than do their academically normal count or overcount during the problem-
peers, and they often use problem-solving solving process. Such miscounting can occur
procedures, such as counting all, that are if the child loses track of where he or she is
more commonly used by younger children in the counting process, that is, how many
(Geary, 1990; Jordan et al., 1995; Jordan & fingers he or she has counted and how many
Montani, 1997). The errors result when remain to be counted. Working memory is
these children miscount, typically under- also implicated in the difficulties that chil-
counting or overcounting by 1 (Geary, dren with AD have during the solving of
1990). As a group, children with AD also more complex arithmetic problems. The
rely on finger counting, as contrasted with procedural errors for the children with AD
verbal counting, more frequently and use assessed by Russell and Ginsburg (1984)
this strategy for more years than do academ- appeared to result from difficulties monitor-
ically normal children. A few studies have as- ing and coordinating the sequence of prob-
sessed the procedural competencies of chil- lem-solving steps, which, in turn, suggest
dren with AD during the solving of multistep compromised executive functions.
arithmetic problems, such as 45 × 12 or 126
+ 537. Russell and Ginsburg (1984) found At a more basic level, a working-memory
that fourth-grade children with AD commit- deficit could result from difficulties with
ted more errors than did their IQ-matched representing information in the basic pho-
academically normal peers when solving netic/articulatory or visuospatial working
such problems. These errors involved (1) the memory systems, or from a deficit in ac-
misalignment of numbers while writing companying executive processes, such as at-
down partial answers or (2) errors while car- tentional or inhibitory control (see McLean
rying or borrowing from one column to the & Hitch, 1999). In theory, difficulties with
next. The following sections discuss these representing and manipulating information
procedural characteristics of children with in the phonetic buffer could disrupt the rep-
AD in terms of working memory, conceptual resentation of number words and their ar-
knowledge, and neural correlates. ticulation during the counting process. At-
tentional and other executive difficulties
WORKING MEMORY could result in problems keeping track of
the counting process and sequencing the
Although the relationship between working multiple steps involved in executing com-
memory and difficulties in executing arith- plex procedures (Geary, 1993).
metical procedures is not yet fully under-
stood, it is clear that children with AD have CONCEPTUAL KNOWLEDGE
some form of working-memory deficit
In addition to working memory, a poor un-
derstanding of the concepts underlying a

Learning Disabilities in Arithmetic 207

procedure can also contribute to a develop- the potential neural systems contributing to
mental delay in the adoption of more so- the procedural deficits of children with AD
phisticated procedures and reduce the abili- (Geary, 1993; Geary & Hoard, 2001). As is
ty to detect procedural errors. found with children with AD, individuals
with acquired or developmental dyscalculia
For instance, delayed use of the counting- are generally able to count arrays of objects,
on procedure and frequent counting errors recite the correct sequence of number words
of children with AD appear to be related, in during the act of counting (e.g., counting
part, to immature counting knowledge. In from 1 to 20), and understand many basic
our first study in the area of counting, we counting concepts (e.g., cardinality;
found that first-grade children with AD and Hittmair-Delazer, Sailer, & Benke, 1995;
RD understood most of the essential fea- Seron et al., 1991; Temple, 1989). Individu-
tures of counting, such as cardinality—that als with dyscalculia caused by damage to
is, that the last stated number word repre- the right hemisphere sometimes show diffi-
sents the total number of items in the count- culties with the procedural component of
ed set (Geary et al., 1992; for discussion of counting, specifically, difficulties with sys-
counting, see Briars & Siegler, 1984; Gel- tematically pointing to successive objects as
man & Gallistel, 1978). However, these they are enumerated (Seron et al., 1991).
children consistently made errors on tasks However, the relation between this feature
that assessed other features of counting, in of dyscalculia and the procedural deficits of
particular order irrelevance. Many of the children with AD is not clear.
children with AD believed that a correct but
nonsequential counting of items (e.g., skip- Difficulties solving complex arithmetic
ping items and then coming back to count problems are also common with acquired
them) resulted in an incorrect count. The and developmental dyscalculia (Semenza,
pattern suggests that although most chil- Miceli, & Girelli, 1997; Temple, 1991). As
dren with AD know the standard counting an example, in an extensive assessment of
sequence and understand some counting the counting, number, and arithmetic com-
concepts, they nonetheless appear to view petencies of a 17-year-old—M.M.—with se-
counting as a rote, mechanical activity. vere congenital damage to the right frontal
and parietal cortices, Semenza and his col-
Children, including many children with leagues reported deficits similar to those re-
AD, who do not understand the order-irrel- ported by Russell and Ginsburg (1984) for
evance concept use the counting-on proce- children with AD. Basic number and count-
dure during problem solving much less fre- ing skills were intact, as was the ability to
quently than do other children (Geary et al., retrieve basic facts (such as 8 for 5 + 3) from
1992; Geary, Hamson, & Hoard, 2000). It memory. However, M.M. had difficulty
is possible that the switch from use of the solving complex division and multiplication
counting-all procedure to the counting-on problems, such as 32 × 67. Of particular
procedure requires an understanding that difficulty was tracking the sequence of par-
counting does not have to start from one tial products. Once the first step was com-
and proceed in the standard sequential or- pleted (2 × 7), difficulties placing the partial
der. The immature counting knowledge of product (4) in the correct position and car-
children with AD may also contribute to rying to the next column were evident.
their frequent counting errors, in particular Thus, the primary deficit of M.M. appeared
a failure to detect and thus self-correct these to involve difficulties sequencing the order
errors. In other words, conceptual knowl- of operations and monitoring the problem-
edge not only guides procedural use, it may solving process, as is often found with dam-
also provide a frame for evaluating the ac- age to the frontal cortex (Luria, 1980);
curacy with which procedures are executed. Temple (1991) reported a similar pattern of
procedural difficulties for an individual
NEURAL CORRELATES with neurodevelopmental abnormalities in
the right frontal cortex. It remains to be
Given the similarity between the deficits as- seen if a compromised right frontal cortex
sociated with AD and those associated with contributes to aspects of the procedural
acquired dyscalculia, neuropsychological deficits of children with AD.
studies of dyscalculia provide insights as to

208 CAUSES AND BEHAVIORAL MANIFESTATIONS

Semantic Memory Deficits arithmetic problems by means of counting
should eventually result in associations
As described earlier, many children with AD forming between problems and generated
do not show the shift from procedural- answers (Siegler, 1996; Siegler & Shrager,
based problem solving to memory-based 1984). Because counting typically engages
problem solving that is commonly found in the phonetic and semantic (e.g., understand-
academically normal children (Geary et al., ing the quantity associated with number
1987; Ostad, 1997). The pattern suggests words) memory systems, any disruption in
that children with AD have difficulties stor- the ability to represent or retrieve informa-
ing or accessing arithmetic facts in or from tion from these systems should, in theory,
long-term memory. Indeed, disrupted mem- result in difficulties in forming problem/an-
ory-based processes are consistently found swer associations during counting (Geary,
with comparisons of children with AD and 1993; Geary, Bow-Thomas, Fan, & Siegler,
other children (Barrouillet et al., 1997; Bull 1993). Although not definitive with respect
& Johnston, 1997; Garnett & Fleischner, to this hypothesis, the work of Dehaene and
1983; Geary, 1993; Geary & Brown, 1991; his colleagues suggests that the retrieval of
Geary et al., 1987; Jordan & Montani, arithmetic facts is indeed supported by a
1997; Ostad, 1997). Disruptions in the abil- system of neural structures that appear to
ity to retrieve basic facts from long-term support phonetic and semantic representa-
memory might, in fact, be considered a tions and are engaged during incrementing
defining feature of AD (Geary, 1993). Most processes, such as counting. These areas in-
of these individuals can, however, retrieve clude the left basal ganglia and the left pari-
some facts, and disruptions in the ability to eto–occipito–temporal areas (Dehaene &
retrieve facts associated with one operation Cohen, 1995, 1997). Damage to either the
(e.g., multiplication) are sometimes found subcortical or cortical structures in this net-
with intact retrieval of facts associated with work is associated with difficulties accessing
another operation (e.g., subtraction), at previously known arithmetic facts (Dehaene
least when retrieval deficits are associated & Cohen, 1991, 1997). However, it is not
with overt brain injury (Pesenti, Seron, & currently known if the retrieval deficits of
Van Der Linden, 1994). children with AD are the result of damage
to or neurodevelopmental abnormalities in
As described in Table 12.3, when they re- the regions identified by Dehaene and Co-
trieve arithmetic facts from long-term mem- hen (1995, 1997).
ory, children with AD commit many more
errors than do their academically normal More recent studies of children with AD
peers and show error and RT patterns that suggest a second form of retrieval deficit,
often differ from the patterns found with specifically, disruptions in the retrieval
younger, academically normal children process due to difficulties in inhibiting the re-
(Geary, 1993; Geary, Hamson, & Hoard, trieval of irrelevant associations. This form
2000). The RT patterns are similar to the of retrieval deficit was first discovered by
patterns found with children who have suf- Barrouillet and colleagues (1997), based on
fered from an early (before age 8 years) le- the memory model of Conway and Engle
sion to the left hemisphere or associated (1994), and was recently confirmed in our
subcortical regions (Ashcraft et al., 1992), laboratory (Geary, Hamson, & Hoard,
as noted earlier. Although this pattern does 2000; see also Koontz & Berch, 1996). In the
not indicate that children with AD have suf- Geary and colleagues (2000) study, one of
fered from some form of overt brain injury, the arithmetic tasks required children to use
it does suggest that the memory-based only retrieval—the children were instructed
deficits of many of these children may re- not to use counting strategies—to solve sim-
flect the same mechanisms underlying the ple addition problems (see also Jordan &
retrieval deficits associated with dyscalculia Montani, 1997). Children with AD, as well
(Geary, 1993; Rourke, 1993). as children with RD, committed more re-
trieval errors than did their academically
However, the cognitive and neural mech- normal peers, even after controlling for IQ.
anisms underlying these deficits are not The most common of these errors was a
completely understood. On the basis of counting-string associate of one of the ad-
Siegler’s strategy-choice model, solving

Learning Disabilities in Arithmetic 209

dends. For instance, common retrieval errors or knowledge of geometric theorems (e.g.,
for the problem 6 + 2 were 7 and 3, the num- Dehaene, Spelke, Pinel, Stanescu, &
bers following 6 and 2, respectively, in the Tsivkin, 1999; Geary, 1993, 1996). Many
counting sequence. Hanich and colleagues children with the procedural and/or seman-
(2001) found a similar pattern, although the tic memory forms of AD, at least as related
proportion of retrieval errors that were to simple arithmetic, do not appear to differ
counting-string associates was lower than from other children in basic visuospatial
that found by Geary and colleagues. competencies (Geary, Hamson, & Hoard,
2000; Morris et al., 1998). There is, howev-
The pattern in these more recent studies er, evidence that some children with AD
(e.g., Geary, Hamson, & Hoard, 2000) and who show broader performance deficits in
that of Barrouillet and colleagues (1997) is mathematics may have a deficit in visuospa-
in keeping with Conway and Engle’s (1994) tial competencies.
position that individual differences in work-
ing memory and retrieval efficiency are re- McLean and Hitch (1999) found that chil-
lated, in part, to the ability to inhibit irrele- dren with AD showed a performance deficit
vant associations. In this model, the on a spatial working-memory task, although
presentation of a to-be-solved problem re- it is not clear if the difference resulted from
sults in the activation of relevant informa- an actual spatial deficit or from a deficit in
tion in working memory, including problem executive functions (e.g., the ability to main-
features—such as the addends in a simple tain attention on the spatial task). In any
addition problem—and information associ- case, Hanich and her colleagues (2001)
ated with these features. Problem solving is found that children with AD differed from
efficient when irrelevant associations are in- their peers on an estimation task and in the
hibited and prevented from entering work- ability to solve complex word problems. Al-
ing memory. Inefficient inhibition results in though performance on both of these tasks is
activation of irrelevant information, which supported by spatial abilities (Dehaene et al.,
functionally lowers working-memory ca- 1999; Geary, 1996; Geary, Saults, Liu, &
pacity. In this view, children with AD make Hoard, 2000), it is not clear whether the re-
retrieval errors, in part because they cannot sults of Hanich and colleagues were due to a
inhibit irrelevant associations from entering spatial deficit in children with AD.
working memory. Once in working memo-
ry, these associations either suppress or Conclusion
compete with the correct association for ex-
pression. Whatever the cognitive mecha- The theoretical models and experimental
nism, these results suggest that the retrieval methods used to study the development of
deficits of some children with AD may number, counting, and arithmetical compe-
spring from either delayed development of tencies in academically normal children
those areas of the prefrontal cortex that have provided a much needed framework
support inhibitory mechanisms, or neurode- for guiding the study of children with AD.
velopmental abnormalities in these regions We now understand the problem-solving
(Bull, Johnston, & Roy, 1999; Welsh & functions and deficits of children with AD,
Pennington, 1988). The results also suggest at least as related to the solving of simple
that inhibitory mechanisms should be con- arithmetic problems (e.g., 4 + 7) and simple
sidered potential contributors to the comor- word problems (Geary, Hamson, & Hoard,
bidity of AD and ADHD in some children. 2000; Hanich et al., 2001; Ostad, 2000).
Most of these children use problem-solving
Visuospatial Deficits procedures that are more commonly used
by younger, academically normal children,
The relation between visuospatial compe- and tend to commit more procedural errors.
tencies and AD has not been fully explored. Over the course of the elementary-school
In theory, visuospatial deficits should affect years, the procedural competencies of many
performance in some mathematical do- children with AD tend to improve, and thus
mains, such as certain areas of geometry their early deficits seem to represent a devel-
and the solving of complex word problems, opmental delay and not a fundamental cog-
but not other domains, such as fact retrieval

210 CAUSES AND BEHAVIORAL MANIFESTATIONS

nitive deficit. At the same time, many chil- Ashcraft, M. H., & Battaglia, J. (1978). Cognitive
dren with AD have difficulties retrieving ba- arithmetic: Evidence for retrieval and decision
sic arithmetic facts from long-term memory, processes in mental addition. Journal of Experi-
a deficit that often does not improve and mental Psychology: Human Learning and Memo-
thus may represent a developmental differ- ry, 4, 527–538.
ence. Some insights have also been gained
regarding the cognitive and neural mecha- Ashcraft, M. H., Yamashita, T. S., & Aram, D. M.
nisms contributing to the procedural and re- (1992). Mathematics performance in left and
trieval characteristics of children with AD, right brain-lesioned children. Brain and Cogni-
including compromised working memory tion, 19, 208–252.
and executive functions.
Badian, N. A. (1983). Dyscalculia and nonverbal
Much remains to be accomplished, how- disorders of learning. In H. R. Myklebust (Ed.),
ever. In comparison to simple arithmetic, Progress in learning disabilities (Vol. 5, pp.
relatively little research has been conducted 235–264). New York: Stratton.
on the ability of children with AD to solve
more complex arithmetic problems (but see Barrouillet, P., Fayol, M., & Lathulière, E. (1997).
Russell & Ginsburg, 1984), and even less Selecting between competitors in multiplication
has been conducted in other mathematical tasks: An explanation of the errors produced by
domains. Even in the area of simple arith- adolescents with learning disabilities. Interna-
metic, the cognitive and neural mechanisms tional Journal of Behavioral Development, 21,
that contribute to the problem-solving char- 253–275.
acteristics of children with AD are not fully
understood and are thus in need of further Briars, D., & Siegler, R. S. (1984). A featural analy-
study. Other areas that are in need of atten- sis of preschoolers’ counting knowledge. Devel-
tion include the development of diagnostic opmental Psychology, 20, 607–618.
instruments for AD; cognitive and behav-
ioral genetic research on the comorbidity of Bull, R., & Johnston, R. S. (1997). Children’s arith-
AD and other forms of LD and ADHD; metical difficulties: Contributions from process-
and, of course, the development of remedial ing speed, item identification, and short-term
techniques. If progress over the past 10 memory. Journal of Experimental Child Psychol-
years is any indication, we should see signif- ogy, 65, 1–24.
icant advances in many of these areas over
the next 10 years. Bull, R., Johnston, R. S., & Roy, J. A. (1999). Ex-
ploring the roles of the visual–spatial sketch pad
Acknowledgments and central executive in children’s arithmetical
skills: Views from cognition and developmental
I thank Cathy DeSoto and Mary Hoard for com- neuropsychology. Developmental Neuropsychol-
ments on an earlier draft. Preparation of the chapter ogy 15, 421–442.
was supported by grant R01 HD38283 from the
National Institute of Child Health and Human De- Carpenter, T. P., & Moser, J. M. (1984). The acqui-
velopment. sition of addition and subtraction concepts in
grades one through three. Journal for Research in
References Mathematics Education, 15, 179–202.

Ackerman, P. T., & Dykman, R. A. (1995). Read- Conway, A. R. A., & Engle, R. W. (1994). Working
ing-disabled students with and without comorbid memory and retrieval: A resource-dependent in-
arithmetic disability. Developmental Neuropsy- hibition model. Journal of Experimental Psychol-
chology, 11, 351–371. ogy: General, 123, 354–373.

Ashcraft, M. H. (1982). The development of mental Dehaene, S., & Cohen, L. (1991). Two mental cal-
arithmetic: A chronometric approach. Develop- culation systems: A case study of severe acalculia
mental Review, 2, 213–236. with preserved approximation. Neuropsycholo-
gia, 29, 1045–1074.
Ashcraft, M. H. (1995). Cognitive psychology and
simple arithmetic: A review and summary of new Dehaene, S., & Cohen, L. (1995). Towards an
directions. Mathematical Cognition, 1, 3–34. anatomical and functional model of number pro-
cessing. Mathematical Cognition, 1, 83–120.

Dehaene, S., & Cohen, L. (1997). Cerebral path-
ways for calculation: Double dissociation be-
tween rote verbal and quantitative knowledge of
arithmetic. Cortex, 33, 219–250.

Dehaene, S., Spelke, E., Pinel, P., Stanescu, R., &
Tsivkin, S. (1999). Sources of mathematical
thinking: Behavioral and brain-imaging evidence.
Science, 284, 970–974.

Fuson, K. C. (1982). An analysis of the counting-on
solution procedure in addition. In T. P. Carpenter,
J. M. Moser, & T. A. Romberg (Eds.), Addition
and subtraction: A cognitive perspective (pp.
67–81). Hillsdale, NJ: Erlbaum.

Garnett, K., & Fleischner, J. E. (1983). Automatiza-
tion and basic fact performance of normal and

Learning Disabilities in Arithmetic 211

learning disabled children. Learning Disabilities Evidence for a single memory network. Memory
Quarterly, 6, 223–230. and Cognition, 14, 478–487.
Geary, D. C. (1990). A componential analysis of an Geary, D. C., Widaman, K. F., Little, T. D., &
early learning deficit in mathematics. Journal of Cormier, P. (1987). Cognitive addition: Compari-
Experimental Child Psychology, 49, 363–383. son of learning disabled and academically normal
Geary, D. C. (1993). Mathematical disabilities: elementary school children. Cognitive Develop-
Cognitive, neuropsychological, and genetic com- ment, 2, 249–269.
ponents. Psychological Bulletin, 114, 345–362. Gelman, R., & Gallistel, C. R. (1978). The child’s
Geary, D. C. (1994). Children’s mathematical devel- understanding of number. Cambridge, MA: Har-
opment: Research and practical applications. vard University Press.
Washington, DC: American Psychological Asso- Gelman, R., & Meck, E. (1983). Preschooler’s
ciation. counting: Principles before skill. Cognition, 13,
Geary, D. C. (1996). Sexual selection and sex differ- 343–359.
ences in mathematical abilities. Behavioral and Groen, G. J., & Parkman, J. M. (1972). A chrono-
Brain Sciences, 19, 229–284. metric analysis of simple addition. Psychological
Geary, D. C. (2000). Mathematical disorders: An Review, 79, 329–343.
overview for educators. Perspectives, 26, 6–9. Gross-Tsur, V., Manor, O., & Shalev, R. S. (1996).
Geary, D. C., Bow-Thomas, C. C., Fan, L., & Developmental dyscalculia: Prevalence and de-
Siegler, R. S. (1993). Even before formal instruc- mographic features. Developmental Medicine
tion, Chinese children outperform American chil- and Child Neurology, 38, 25–33.
dren in mental addition. Cognitive Development, Hanich, L. B., Jordan, N. C., Kaplan, D., & Dick, J.
8, 517–529. (2001). Performance across different areas of
Geary, D. C., Bow-Thomas, C. C., & Yao, Y. mathematical cognition in children with learning
(1992). Counting knowledge and skill in cogni- difficulties. Journal of Educational Psychology,
tive addition: A comparison of normal and math- 93, 615–626.
ematically disabled children. Journal of Experi- Hitch, G. J., & McAuley, E. (1991). Working mem-
mental Child Psychology, 54, 372–391. ory in children with specific arithmetical learning
Geary, D. C., & Brown, S. C (1991). Cognitive ad- disabilities. British Journal of Psychology, 82,
dition: Strategy choice and speed-of-processing 375–386.
differences in gifted, normal, and mathematically Hittmair-Delazer, M., Sailer, U., & Benke, T.
disabled children. Developmental Psychology, (1995). Impaired arithmetic facts but intact con-
27, 398–406. ceptual knowledge—A single-case study of dys-
Geary, D. C., Brown, S. C., & Samaranayake, V. A. calculia. Cortex, 31, 139–147.
(1991). Cognitive addition: A short longitudinal Jordan, N. C., & Hanich, L. B. (2000). Mathemati-
study of strategy choice and speed-of-processing cal thinking in second-grade children with differ-
differences in normal and mathematically dis- ent forms of LD. Journal of Learning Disabilities,
abled children. Developmental Psychology, 27, 33, 567–578.
787–797. Jordan, N. C., Levine, S. C., & Huttenlocher, J.
Geary, D. C., & Burlingham-Dubree, M. (1989). (1995). Calculation abilities in young children
External validation of the strategy choice model with different patterns of cognitive functioning.
for addition. Journal of Experimental Child Psy- Journal of Learning Disabilities, 28, 53–64.
chology, 47, 175–192. Jordan, N. C., & Montani, T. O. (1997). Cognitive
Geary, D. C., Hamson, C. O., & Hoard, M. K. arithmetic and problem solving: A comparison of
(2000). Numerical and arithmetical cognition: A children with specific and general mathematics
longitudinal study of process and concept deficits difficulties. Journal of Learning Disabilities, 30,
in children with learning disability. Journal of 624–634.
Experimental Child Psychology, 77, 236–263. Koontz, K. L., & Berch, D. B. (1996). Identifying
Geary, D. C., & Hoard, M. K. (2001). Numerical simple numerical stimuli: Processing inefficiencies
and arithmetical deficits in learning-disabled chil- exhibited by arithmetic learning disabled chil-
dren: Relation to dyscalculia and dyslexia. Apha- dren. Mathematical Cognition, 2, 1–23.
siology, 15, 635–647. Kosc, L. (1974). Developmental dyscalculia. Jour-
Geary, D. C., Hoard, M. K., & Hamson, C. O. nal of Learning Disabilities, 7, 164–177.
(1999). Numerical and arithmetical cognition: Light, J. G., & DeFries, J. C. (1995). Comorbidity
Patterns of functions and deficits in children at of reading and mathematics disabilities: Genetic
risk for a mathematical disability. Journal of Ex- and environmental etiologies. Journal of Learn-
perimental Child Psychology, 74, 213–239. ing Disabilities, 28, 96–106.
Geary, D. C., Saults, S. J., Liu, F., & Hoard, M. K. Luria, A. R. (1980). Higher cortical functions in
(2000). Sex differences in spatial cognition, com- man (2nd ed.). New York: Basic Books.
putational fluency, and arithmetical reasoning. McLean, J. F., & Hitch, G. J. (1999). Working
Journal of Experimental Child Psychology, 77, memory impairments in children with specific
337–353. arithmetic learning difficulties. Journal of Experi-
Geary, D. C., Widaman, K. F., & Little, T. D. mental Child Psychology, 74, 240–260.
(1986). Cognitive addition and multiplication: Morris, R. D., Stuebing, K. K., Fletcher, J. M.,

212 CAUSES AND BEHAVIORAL MANIFESTATIONS

Shaywitz, S. E., Lyon, G. R., Shankweiler, D. P., Developmental Medicine and Child Neurology,
Katz, L., Francis, D. J., & Shaywitz, B. A. (1998). 35, 593–601.
Subtypes of reading disability: Variability around Shalev, R. S., Manor, O., Kerem, B., Ayali, M.,
a phonological core. Journal of Educational Psy- Badichi, N., Friedlander, Y., & Gross-Tsur, V.
chology, 90, 347–373. (2001). Developmental dyscalculia is a familial
Ostad, S. A. (1997). Developmental differences in learning disability. Journal of Learning Disabili-
addition strategies: A comparison of mathemati- ties, 34, 59–65.
cally disabled and mathematically normal chil- Siegel, L. S., & Ryan, E. B. (1989). The develop-
dren. British Journal of Educational Psychology, ment of working memory in normally achieving
67, 345–357. and subtypes of learning disabled children. Child
Ostad, S. A. (1998a). Comorbidity between mathe- Development, 60, 973–980.
matics and spelling difficulties. Log Phon Vovol, Siegler, R. S. (1987). The perils of averaging data
23, 145–154. over strategies: An example from children’s addi-
Ostad, S. A. (1998b). Developmental differences in tion. Journal of Experimental Psychology: Gen-
solving simple arithmetic word problems and eral, 116, 250–264.
simple number-fact problems: A comparison of Siegler, R. S. (1988). Individual differences in strate-
mathematically normal and mathematically dis- gy choices: Good students, not-so-good students,
abled children. Mathematical Cognition, 4, 1–19. and perfectionists. Child Development, 59,
Ostad, S. A. (2000). Cognitive subtraction in a de- 833–851.
velopmental perspective: Accuracy, speed-of-pro- Siegler, R. S. (1996). Emerging minds: The process
cessing and strategy-use differences in normal of change in children’s thinking. New York: Ox-
and mathematically disabled children. Focus on ford University Press.
Learning Problems in Mathematics, 22, 18–31. Siegler, R. S., & Shrager, J. (1984). Strategy choice
Pesenti, M., Seron, X., & Van Der Linden, M. in addition and subtraction: How do children
(1994). Selective impairment as evidence for men- know what to do? In C. Sophian (Ed.), Origins of
tal organisation of arithmetical facts: BB, a case of cognitive skills (pp. 229–293). Hillsdale, NJ: Erl-
preserved subtraction? Cortex, 30, 661–671. baum.
Räsänen, P., & Ahonen, T. (1995). Arithmetic dis- Svenson, O., & Broquist, S. (1975). Strategies for
abilities with and without reading difficulties: A solving simple addition problems: A comparison
comparison of arithmetic errors. Developmental of normal and subnormal children. Scandinavian
Neuropsychology, 11, 275–295. Journal of Psychology, 16, 143–151.
Rourke, B. P. (1993). Arithmetic disabilities, specif- Swanson, H. L. (1993). Working memory in learn-
ic and otherwise: A neuropsychological perspec- ing disability subgroups. Journal of Experimental
tive. Journal of Learning Disabilities, 26, Child Psychology, 56, 87–114.
214–226. Temple, C. M. (1989). Digit dyslexia: A category-
Russell, R. L., & Ginsburg, H. P. (1984). Cognitive specific disorder in developmental dyscalculia.
analysis of children’s mathematical difficulties. Cognitive Neuropsychology, 6, 93–116.
Cognition and Instruction, 1, 217–244. Temple, C. M. (1991). Procedural dyscalculia and
Semenza, C., Miceli, L., & Girelli, L. (1997). A number fact dyscalculia: Double dissociation in
deficit for arithmetical procedures: Lack of developmental dyscalculia. Cognitive Neuropsy-
knowledge or lack of monitoring? Cortex, 33, chology, 8, 155–176.
483–498. Welsh, M. C., & Pennington, B. F. (1988). Assess-
Seron, X., Deloche, G., Ferrand, I., Cornet, J.-A., ing frontal lobe functioning in children: Views
Frederix, M., & Hirsbrunner, T. (1991). Dot from developmental psychology. Developmental
counting by brain damaged subjects. Brain and Neuropsychology, 4, 199–230.
Cognition, 17, 116–137. Widaman, K. F., Geary, D. C., Cormier, P., & Little,
Shalev, R. S., Manor, O., & Gross-Tsur, V. (1993). T. D. (1989). A componential model for mental
The acquisition of arithmetic in normal children: addition. Journal of Experimental Psychology:
Assessment by a cognitive model of dyscalculia. Learning, Memory, and Cognition, 15, 898–919.

13

Language Processes:
Keys to Reading Disability

Virginia A. Mann

Introduction and Statement and 2 years behind that which is predicted
of the Problem on the basis of their age, IQ, and social
standing.
What makes a poor reader a poor reader?
Some 4 to 10% of children encounter severe Recent reviews have seriously discredited
difficulty in learning to read, and it is the the use of a discrepancy between IQ and
objective of this chapter to review one of the reading ability as a definitive characteristic
most prevalent causes of their problems. that sets children with dyslexia apart from
Since the early 1970s, an ever-growing body other “garden-variety” poor readers (see
of evidence has linked developmental read- Fletcher, Francis, Rourke, Shaywitz, & Shay-
ing problems to inadequacies in one or more witz, 1992; Francis et al., 1996; Shankweiler
areas of spoken language. et al., 1995). Rather than forming into two
separate groups, dyslexic and garden-variety
The focus of this review is reading difficul- poor readers seem to form a continuous dis-
ty as it is manifest in the elementary grades. It tribution (see Stanovich, 1988). Both types
includes but is not limited to developmental of readers with disabilities have problems
dyslexia. Developmental dyslexia is a syn- with the language skills that are of primary
drome defined by discrepancy between a interest in this chapter. Most of the differ-
child’s intelligence level and his or her level of ences between the groups seem to involve
reading ability. Individuals with dyslexia “real-world knowledge” and “strategic abil-
read significantly below the level that would ities”: The children with dyslexia possess su-
be expected based on their IQ scores alone. perior skills in these “nonlinguistic” areas;
Although learning to read is a complex learn- hence the discrepancy between their IQ and
ing task that correlates about 0.6 with IQ their reading ability, whereas the garden-
(Rutter, 1978), there nevertheless exist chil- variety poor readers may show inadequacies
dren who possess a seemingly adequate IQ in these skills as well as in their language
(typically 90 or higher) but nonetheless en- skills (see Stanovich, 1988, for a discussion).
counter reading problems. Such children are
said to have a specific reading difficulty, as Over the course of this chapter, my intent
their actual reading ability lags between 1 is to introduce and justify a language-based
approach to reading disability. This ap-

213

214 CAUSES AND BEHAVIORAL MANIFESTATIONS

proach owes largely to the work of I. Y. writing system and Japanese Kanji represent
Liberman and her husband and colleague, language in terms of units of smaller units
Alvin Liberman. Their synergistic insights of meaning called morphemes. Syllabaries
and their erudition with respect to educa- such as Hebrew and Japanese Kana repre-
tion, psychology, linguistics, and the speech sent language at the syllables. Alphabets
sciences were a necessary catalyst to the re- such as Spanish, German, French, Italian,
alization that many instances of reading and English represent units that are called
problems are rooted in spoken language. phonemes (e.g., consonants and vowels).

A Theoretical Perspective: Each type of writing system places certain
English Transcribes Phonemes demands on a reader (for discussion, see
and Morphemes Hung & Tzeng, 1981; Watt, 1989). As first
noted by the Libermans and their colleagues
Two insights underscore the role of lan- (see, e.g., Liberman, Liberman, Mattingly,
guage problems in poor reading. One is the- & Shankweiler, 1980; Mattingly, 1972) one
oretical and historical; the other is research of the most important demands involves
driven. The theoretical and historical in- language awareness. A reader of a writing
sight concerns the relation between the system needs to be aware of the unit the
units of written languages and spoken lan- writing system is representing. Otherwise it
guage, how the one is designed to map onto will be difficult to understand how written
the other. The research-based insight con- words relate to the spoken language.
cerns the active role of language skills in
skilled reading. We often think of reading Alphabets are a case in point. Because al-
as a “visual” skill, yet visual perception is phabets represent phonemes, someone wish-
only the tip of the reading iceberg. For ing to learn to read an alphabet needs to be
readers to successfully decode the words sensitive to the fact that spoken language can
and uncover the sentences and paragraphs be broken down into phonemes. In contrast,
on the page, seeing is not enough. Readers a reader of a syllabary need only be aware of
must successfully map the units of their syllables. This sensitivity to the phonemes
written language onto the units of their within words is referred to as phoneme
spoken language. Spoken language comes awareness and is an important trait of suc-
first in both the history of the individual cessful beginning readers and a definitive
and the history of our species; it is both problem for many young children and poor
universal and natural. Writing and reading readers, in particular, as we shall see in
are parasitic upon speaking and listening the section “Reading Problems, Phoneme
and operate by virtue of some of the very Awareness, and Morpheme Awareness.”
same skills that allow us to be speakers and
hearers of our language. The English Writing System

How Writing Systems Represent The English alphabet represents a special
Spoken Language case because it is not a pure alphabet as
much as an alphabet with logographic over-
A writing system, or “orthography,” writes tones. It does not provide the consistent
language by representing certain units of a one-to-one mapping of letter to phonemes
spoken language. All writing systems repre- that one finds in Spanish or German, for
sent units of a spoken language, but there example. Rather, it provides a “deeper,”
are differences that turn on the type of lin- more abstract level of representation that
guistic units that are being represented. For goes beyond phonemes and sounds to the
example, ideographies such as American In- morphemes and meanings of words. As
dian petroglyphs or the universal set of road such it has been, referred to as a mor-
signs that a driver encounters in a daily phophonological transcription because it
commute represent language at the level of combines the transcription of morphemes
“ideas.” Logographies such as the Chinese and phonemes As noted by Chomsky
(1964), English alphabetic transcription
corresponds not so much to the consonants
and vowels that speakers and hearers think

Language Processes 215

they pronounce and perceive as much as it 3,000 characters for a newspaper and over
to the way theoretical linguistics assumes 10,000 for scholarly works. Phoneme tran-
that words are abstractly represented in the scription is also highly productive. By em-
ideal speaker/hearer’s mental dictionary, or bodying a highly “rule-governed” relation-
“lexicon.” Words in the mental lexicon are ship between written and spoken words, it
represented in terms of morphemes (e.g., allows the reader to read not only highly fa-
root words, prefixes, and suffixes) as well miliar words but also less familiar ones such
as in terms of phonemes. When speakers of as “skiff,” and even nonsense words such as
English produce or perceive language, they “ifts” or “polypluckable.” A reader of a lo-
convert the morphophonological represen- gography may have difficulty pronouncing
tations of the words in their lexicon to less an unfamiliar word even when he or she has
abstract, “phonetic” representations by us- memorized thousands of distinct characters.
ing an ordered series of phonological rules
that alter, insert, or delete phonemes. These The benefit that accrues from the tran-
same rules can help them to pronounce scription of morphemes involves clues to
words spelled according to their mor- word meaning and function. By transcribing
phophonological representations. a “deep,” relatively abstract level of phono-
logical structure where units of meaning are
If a writing system correctly represents represented, the English writing system
words in a deep manner it will sometimes fail helps convey cues to meaning as well as to
to represent the more superficial phonetic sound. Recognizing “joy” in “enjoy,”
representations with which we are most fa- “heal” in “health,” “atom” in “atomic,” or
miliar. Witness, for example, the spellings of “relate” in “relation” can facilitate the re-
words in pairs such as “atom” versus “atom- covery of that word’s meaning. Recognizing
ic,” “heal” versus “health,” and “relate” suffixes such as –ly, -ing, and -ness can pro-
versus “relation.” In each of these word vide cues to a word’s function in a sentence.
pairs, the similar spelling of the base and de- The transcription of morphemes can also
rived form captures the relatedness of their offer a common denominator to people who
meanings. But the similar spelling comes at speak with different accents. Rather than
the cost of having a letter or letter sequence specifying each surface phoneme, it leaves it
represent different phonemes in different to the speaker to apply his or her accent to
words (e.g., the two pronunciations of “o,” the morphophonological representation
“ea,” and “t”). Preservation of common available on the page. A final advantage of
roots and common word ancestries is one of morphophonological spelling is that it can
the reasons the English vowel system is so disambiguate homonyms. Indeed, one of the
complicated and one of the reasons English most common justifications of the use of a
uses nearly 120 spelling patterns for 40 logography in Chinese concerns the number
phonemes. It is also the source of homo- of homophones in the Chinese language.
phones such as to-two-too and their-there-
they’re. The different spellings of these Ultimately, the utility of a given orthogra-
homonyms reflects their different meanings phy rests on the nature of the spoken lan-
(e.g., morphology); the common pronuncia- guage it transcribes. A logography is appro-
tion is what makes their usage so hard. The priate for Chinese because it allows people
important point to be remembered is that the to read the same text even though they can-
English alphabet represents phonemes and not understand each other’s speech. For
morphemes and there is a certain trade-off Japanese, the Kana syllabaries are quite well
between then two types of representation. suited to the 100 or so syllables in the
Japanese language. English, however, has a
Why have a writing system that tran- less profound dialectical variation than Chi-
scribes both phonemes and morphemes? nese, and the English language employs
Each unit has certain advantages. Economy more than 1,000 syllables. Hence, an alpha-
of characters is a clear benefit of transcrib- bet is appropriate, and historical change
ing phonemes. By transcribing phonemes, and language infusions have made that al-
the alphabet can get away with 26 letters phabet morphophonemic. It would be less
and 120 spelling patterns where a logo- efficient and even a disservice to present the
graphic orthography requires 2,000 to English writing system otherwise.

216 CAUSES AND BEHAVIORAL MANIFESTATIONS

A Research-Based Perspective: Language are recognized as units within words and
Processes Support Skilled Reading may play a part in the reading of English.
However, it is not clear whether the recogni-
We have reviewed the way in which the tion of morphemic units mediates word
English alphabet transcribes spoken English recognition for all words as opposed to
language as a form of deductive evidence morphologically complex ones.
about the importance of spoken language
skills to reading. Now let us turn to a com- Language Skills and the Reading of Sentences
plementary source of empirical evidence, and Paragraphs
namely, investigations of the process of
skilled reading. Studies of adult readers From the point of word perception onward,
show a clear involvement of certain spoken the involvement of speech processes in read-
language processes in the skilled reading of ing is quite clear. First, there is considerable
words, sentences, and paragraphs. They evidence that temporary working or “short-
confirm that reading is really quite “para- term” memory for written material involves
sitic” upon spoken language processes. a speech code. This speech code is some-
times referred to as silent speech, or phonet-
Language Skills and Word Recognition ic representation, and it is used whether the
to-be-remembered material is spoken or
Whether words must be recoded into some written or whether it is isolated letters,
type of “silent speech” has preoccupied psy- printed nonsense syllables, or printed
chological studies of skilled reading. It espe- words. Both the nature of the errors that
cially preoccupied studies of the “lexical ac- subjects make in recalling such material and
cess” processes that make it possible for us the experimental manipulations that help or
to decode and recognize the words of our hurt their memory performance have shown
vocabulary (for recent reviews of how read- us that a phonetic representation is being
ers gain lexical access, see Berent & Perfetti, used. That is, subjects are temporarily re-
1995; Frost, 1998; van Orden, 1987). Un- membering a sequence of written words in
der some circumstances, silent speech does terms of the consonants and vowels within
not appear necessary for word recognition; each word, rather than the visual shape of
some words may be directly perceived as vi- the letters, the shape of the words, and so
sual units, instead of being decoded into a on (cf., for example, Baddeley, 1978; Con-
string of phonemes. But there is clear evi- rad, 1964, 1972; Levy, 1977). It is further
dence implicating at least some “speech the case that subjects appear to rely on pho-
code” involvement in word perception, netic representation when they are required
making many psychologists favor a “dual to comprehend sentences written in either
access” or “parallel race horse” model in alphabetic (Kleiman, 1975; Levy, 1977;
which both phonetic and visual access occur Slowiaczek & Clifton, 1980) or logographic
in parallel. Others believe that a “speech orthographies (Tzeng, Hung, & Wang,
code” or “phonetic” route may be most 1977). Understanding a sentence often re-
heavily used in the case of less frequent quires the reader to hold several words in
words and unfamiliar ones (Seidenberg, memory until the structure of the sentence is
1985), and still others regard the speech apparent. This is one reason we may ob-
code as playing an early, dominant role in serve such significantly high correlations be-
all lexical access (Frost, 1998; Rayner, tween reading and listening comprehension
Sereno, Lesch, & Pollatsek, 1995; van Or- across a variety of languages and orthogra-
den, 1987). phies, including English (Daneman & Car-
penter, 1980; Jackson & McClelland,
As for the recognition of morphemic units 1979).
within spoken words, various authors have
investigated the role of morpheme-size units Thus, regardless of the way in which the
in fluent reading. Use of such methodologies reader recognizes each word on the page,
as letter cancellation tasks (Drenowski & the processes involved in reading sentences
Healy, 1980) and lexical decision (Taft, and paragraphs place certain obvious de-
1984; Taft & Forster, 1975) has given some mands on temporary memory, and tempo-
support to the view that stems and affixes rary memory for language appears to make

Language Processes 217

use of phonetic representation in short-term compare and manipulate the consonants
memory. In the section “Reading Problems and vowels that comprise spoken words.
and Language Processing” we will see that Taking the “t” off “cat,” realizing that
problems with phonetic representation are “clay” and “cream” start with the same
often found among poor beginning readers, sound, realizing that “shoe” and “toe” have
in the form of “short-term memory prob- the same number of sounds—are some oth-
lems.” er examples of activities that require
phoneme awareness.
For morphemes as well as for speech cod-
ing, the study of sentences and paragraphs One slight problem with the term
yield a clearer picture. For example, readers’ “phoneme awareness” is that it is often
eye movements tend to be guided by the used interchangeably with several other
morphemic structure of words within a text terms: “phonemic awareness,” “phonologi-
(Lima, 1987; Rayner & McConkie, 1976). cal awareness,” “metalinguistic awareness”
It is also the case that whether or not words and “linguistic awareness,” to name a few.
are automatically decomposed into their By using the term “phoneme awareness” (or
morphemic constituents prior to lexical ac- “phonemic awareness”) we confine the is-
cess, their morphemic composition is essen- sue to sensitivity about phonemes. “Phono-
tial to sentence comprehension. Adult logical awareness” could also include sensi-
skilled readers are able to use the mor- tivity to rhyme, syllables, and morphemes
phemic structure of words to complete a and the phonological rules that operate on
sentence completion task, and they do so them; “linguistic awareness” and “metalin-
more extensively than do poor readers (Ma- guistic” awareness would further include
hony, 1994; Tyler & Nagy, 1990), a finding sensitivity to syntax (i.e., grammar), seman-
elaborated on in the next section. tics (i.e., meaning), and their rules. To date,
awareness of phonemes has been most often
Reading Problems, Phoneme Awareness, studied and deficient phoneme awareness is
and Morpheme Awareness a major factor in reading disorders.

If reading is a language skill, why doesn’t EVIDENCE FROM THE ANALYSIS OF
every speaker of English automatically be-
comes a successful beginning reader of Eng- READING ERRORS
lish? The answer to this question is that
“knowing” spoken English is a necessary The errors a person makes can be informa-
but not a sufficient requirement for skilled tive about the difficulties that produce those
reading. Would-be readers must go one step errors, and oral reading errors can offer an
further than merely being a speaker/hearer important source of evidence about the
of their language—they must be able to con- cause of reading problems. A consideration
sciously analyze and manipulate the units of these errors has shown that a lack of
that their writing system represents. phoneme awareness is responsible for mak-
ing beginning reading difficult for all young
Phoneme Awareness children (Shankweiler & Liberman, 1972),
including those with dyslexia (Fischer,
Phoneme awareness, as first discussed by Liberman, & Shankweiler, 1977). Such er-
Mattingly (1972), and later developed in rors do not tend to involve visual confu-
several other places (A. M. Liberman, 1999; sions or letter or sequence reversals to any
I. Y. Liberman, 1982; Liberman et al., appreciable degree. What they instead re-
1980), is not something we use in the nor- flect is a problem with integrating the
mal activities of speaking and hearing. We phonological information that letter se-
use it in certain “secondary language activi- quences convey. Hence, children often tend
ties” such as appreciating verse (i.e., alliter- to be correct as to the pronunciation of the
ation), making jokes (i.e., “Where do you first letter in a word but to have more and
leave your dog? In a barking lot . . .”), and more difficulty with subsequent letters, and
talking in secret languages (i.e., Pig Latin). a particular problem with vowels as op-
Such activities require that we consciously posed to consonants. For more detailed pre-
sentation of these findings and their impli-
cations, the reader is referred to work by

218 CAUSES AND BEHAVIORAL MANIFESTATIONS

Shankweiler and Liberman (1972) and Fis- ness about phonemes and current problems
cher and colleagues (1977), and also Russell in learning to read (to name but a few, see,
(1982), which suggests that deficient e.g., Bradley & Bryant, 1985; Fox &
phoneme awareness may also account for Routh, 1976; Muter, Hulme, Snowling, &
the reading difficulties of adult dyslexics. Taylor, 1997; Yopp, 1988; see also Adams,
1990; Brady & Shankweiler, 1991; Perfetti,
EVIDENCE FROM TASKS THAT MEASURE 1985; Wagner & Torgesen, 1987, for re-
views). Studies of kindergarten children
AWARENESS DIRECTLY provide evidence that problems with
phoneme segmentation can presage and
Most studies of phoneme awareness have predict reading problems (see, e.g., Blach-
concerned tasks that measure awareness di- man, 1984; Mann, 1993; Wagner et al.,
rectly. These tasks require children to play 1997; see also Hulme & Joshi, 1998, for an
language “games” that manipulate the expanded set of references). Two examples
phonemes within a word in one way or an- come to mind. Eighty-five percent of a pop-
other: counting them, deleting them, choos- ulation of kindergarten children who went
ing words which contain the same on to become good readers in the first
phoneme, and so on. The use of these tasks grade correctly counted the number of syl-
has revealed that phoneme awareness devel- lables in spoken words, whereas only 17%
ops later than phonetic perception and the of the future poor readers could do so
use of phonetic representation, and it re- (Mann & Liberman, 1984). Sixty-six per-
mains a chronic problem for those individu- cent of first-grade variance in reading
als who are poor readers. scores could be accounted for by a kinder-
garten battery of tests that assessed
The use of such tasks in the study of be- phoneme awareness (Stanovich, Cunning-
ginning readers began in the 1970s, when ham, & Cramer, 1984).
Liberman and her colleagues asked whether
a sample of 4-, 5-, and 6-year-olds could FACTORS THAT UNDERLIE DEFICIENT
learn to play syllable counting games and
phoneme counting games. In each game PHONEME AWARENESS
the idea was to “tap” the number of sylla-
bles/phonemes in a spoken word and the Research is showing that the relation be-
child was given examples to illustrate the tween phoneme awareness and reading is a
concept before being asked to play the game complex two-way street. On the one hand,
with a set of words (Liberman, Shankweiler, exposure to the alphabet and the alphabet-
Fischer, & Carter, 1974). It was discovered ic principle has a clear effect on the devel-
that none of the nursery school children opment of phoneme awareness. For exam-
could tap the number of phonemes in a spo- ple, illiterate adults are unable to
ken word, although half of them managed manipulate the phonetic structure of spo-
to tap the number of syllables. Only 17% of ken words (Morais, Cary, Alegria, & Ber-
the kindergarteners could tap phonemes, al- telson, 1979; Read, Zhang, Nie, & Ding,
though, again, about half of them could tap 1986). It is also the case that the onset of
syllables. At 6, 90% of the children could phoneme awareness in the early grades fol-
tap syllables, and 70% were able to tap lows exposure to alphabetic instruction:
phonemes. From such findings it is clear Children in Germany, who start to learn to
that the awareness of phonemes and sylla- read in first-grade, begin to develop
bles develops considerably between the ages phoneme awareness in the first grade where
of 4 and 6. It is also clear that awareness of American children who begin learning to
phonemes is slower to develop than aware- read in kindergarten show an earlier onset
ness of syllables. Finally, both types of (Mann & Wimmer, 2002). It would seem
awareness markedly improve at just the age that awareness of phonemes is enhanced by
when children are learning to read (Liber- methods of reading instruction that direct
man et al., 1974). the child’s attention to the phonetic struc-
ture of words, and it may even depend on
Numerous experiments involving widely such instruction.
diverse subjects, school systems, and mea-
surement devices have shown a strong pos- However, experience alone cannot be the
itive correlation between a lack of aware-

Language Processes 219

only factor behind some children’s failure to ferences appeared to be largely a function
achieve phoneme awareness, which is aptly of age, letter knowledge, and especially vo-
shown by Bradley and Bryant’s (1978) find- cabulary knowledge.
ing that among a group of 6-year-old skilled
readers and 10-year-old readers with dis- Morpheme Awareness
abilities who were matched for reading abil-
ity, the readers with disabilities performed In word pairs such as reduce-reduction,
significantly worse on a phonological atom-atomic and personal-personality, the
awareness task, even though they would be presence of the derivational suffix (i.e. -ion,
expected to have had more reading instruc- -ic, and –ity) changes the base or stem
tion than the younger children. Here it word’s stress and pronunciation even
could be argued that some constitutional though its spelling remains constant. Thus
factor limited the ability of readers with dis- the simple one-to-one relationship between
abilities to profit from instruction, and thus graphemes and phonemes is destroyed, and
limited their attainment of phonological so- the solution is to take morpheme-size units
phistication. Indeed, Pennington, Van Or- into account. Some authors have noted that
den, Kirson, and Haith (1991) have offered young readers must take advantage of mor-
some new and interesting evidence that defi- pheme-sized units if they are to succeed in
cient phoneme awareness is the primary the reading of multisyllabic words (Carlisle,
trait of individuals who are “familial” 1995; Carlisle & Nomanbhoy, 1993;
dyslexics. Fowler & Liberman, 1995). Only if they are
aware of morpheme-size units may it be-
What might that constitutional factor be? comes possible for them to realize that at-
Elbro (1996), Fowler (1991), and Walley taching certain suffixes to a word can alter
(1993) have all in one way or another sug- the word’s pronunciation in regular, pre-
gested that problems with phoneme aware- dictable ways.
ness might reflect a deficiency in children’s
internal phonological representations. In The literature to date does show some in-
Foy and Mann (2001), we recently attempt- triguing relationships between morpheme
ed to investigate this possibility by examin- awareness and reading. Morphological
ing putative measures of phonological word formation can be generalized into two
strength (perception, production, naming) types: inflectional and derivational, both of
in relation to phonological awareness and which are transcribed in the English orthog-
reading development. The results did not raphy. Mastery of inflections (e.g., past tense
validate strength of phonological represen- and plural markers) is usually accomplished
tation as a unitary construct underlying relatively early in life in a fixed manner and
phonological awareness. Instead they re- has been subtly linked to reading progress
vealed a more selective pattern of associa- during the first and second grades (Carlisle,
tions between spoken-language tasks and 1995; Carlisle & Nomanbhoy, 1993). Mas-
rhyme and phoneme awareness as separable tery of derivational morphology (e.g., suffix-
aspects of phonological awareness. Speech es that do not automatically apply to each
perception (to be discussed in the section member of a syntactic category) seems to in-
“Reading Problems and Deficient Language volve a longer, more open-ended course
Processing”) was closely associated with the (Tyler & Nagy, 1990) that coincides with the
awareness of rhyme, even when age, vocab- ages at which children are learning to be-
ulary, and letter knowledge were con- come proficient readers. There are growing
trolled. Children with a less developed indications that the production of derived
sense of rhyme also had a less mature pat- forms is related to the ability to read English
tern of articulation, independent of age, vo- in the first and second grades (Carlisle, 1995,
cabulary, and letter knowledge. Phoneme 2000; Carlisle & Nomanbhoy, 1993) and
awareness, per se, associated with phono- middle elementary grades (Fowler & Liber-
logical perception and production and chil- man, 1995; Singson, Mahony, & Mann,
dren with low phoneme awareness skills 2000), as well as in the junior and high
showed a different pattern of speech per- school years (Mahony, 1994)
ception and articulation errors than did
children with strong abilities, but these dif- One problem with the interpretation of
studies that link morpheme awareness to

220 CAUSES AND BEHAVIORAL MANIFESTATIONS

reading is that many tests of morpheme question of morphological involvement by
awareness may presume phonological using a morphological measure that is ex-
awareness. Performance on phoneme- and clusively focused on derivational morpholo-
syllable-segmentation tasks is well docu- gy and by using recognition as opposed to
mented to be related to reading ability, and production. We, too, have found that mor-
there have been successful attempts to show pheme awareness significantly correlated
that poor sensitivity to derivational mor- with phoneme awareness. Our analysis fur-
phology relates to reading simply because it ther controlled for the effects of vocabulary
also requires phonological awareness. knowledge whereas Shankweiler and col-
Carlisle and Nomanbhoy (1993; see also leagues (1995) studied controlled for age
Carlisle, 1995), for example, studied first- and IQ. Yet our results were similar. When
graders using separate measures of reading, we controlled for performance on the
phoneme awareness, and morpheme aware- phonological awareness and vocabulary
ness. Their results indicated that the two tasks, we found that morphological aware-
skills were related to each other and to read- ness accounted for 4% of decoding vari-
ing ability, but that once the contribution of ance; Carlisle (1995) found 4%, and
phoneme awareness was considered, mor- Shankweiler and colleagues found 5%. If
pheme awareness made little additional what we had found was a mere statistical
contribution to the children’s variance in artifact, it would be hard to explain how at
reading ability. Using various phoneme and least three arrived at the same 4–5% contri-
morpheme production tasks, Fowler and bution of morphological awareness to de-
Liberman (1995) investigated the levels of coding ability. Rather, it would be more par-
phonological and morphological awareness simonious to conclude that morphological
in less-skilled readers ages 7½ to 9½ years awareness has a small but reliable contribu-
to see whether a low level in each skill cor- tion to decoding ability. When we extend
related with poor reading ability. Their con- our scope of study to children in the later el-
clusions pointed more to a phonologically ementary grades, we may find even higher,
based deficit in poor readers, rather than a more reliable levels of contribution (see
morphological one. They reasoned that Carlisle, 2000; Singson et al., 2000) due to
morphological production tasks often draw the fact that reading ability turns on the
on a mixture of skills (ranging from ortho- ability to decode and comprehend multisyl-
graphic knowledge to phonological sensitiv- labic words.
ity and receptive vocabulary) that are well
documented to be underdeveloped in poor Reading Problems and Deficient
readers. In their view, the poor readers’ low Language Processing
performance on morphological tasks is
most likely a consequence of their deficiency Without spoken English, there would be
in these other skills. nothing for the English orthography to
transcribe; the well-known difficulties of
In yet another relevant study which ad- readers who are deaf attest to the impor-
ministered separate phoneme- and mor- tance of spoken language skills for success-
pheme-based tests to children with reading ful reading. But children who are deaf are
disabilities between 7½ and 9½ years of not the only ones for whom deficient lan-
age, Shankweiler and colleagues (1995) guage abilities are a cause of reading prob-
conclude that phonological deficits and de- lems. As we shall see shortly, many of the
ficient production of morphologically relat- children who hear and are poor readers also
ed forms stem from a common weakness in suffer from spoken language problems, and
the phonological components of language although their problems are considerably
and not from separable problems with more subtle than those of children who are
phonology and morphology. Taken togeth- deaf, they are no less critical. The following
er, these results suggest that any finding that paragraphs summarize various forms of evi-
morphological awareness is related to read- dence about the types of language process-
ing must be analyzed with the caveat that ing skills that have been linked to reading
phonological awareness is an underlying disability. This evidence can be organized in
factor.

We have made another approach to the

Language Processes 221

terms of four levels of language processing: Mann (2001) have confirmed a relationship
speech perception, vocabulary skills, lin- between speech perception and early read-
guistic short-term memory, and syntax and ing skill, but their data on preschool chil-
semantics. dren reveal that speech perception is more
closely tied to the awareness of rhyme than
Speech Perception to the awareness of phonemes, per se (see
section “Reading Problems, Phoneme
The possibility that some aspect of speech Awareness, and Morpheme Awareness”).
perception might be a special problem for
poor readers is illustrated and supported by Vocabulary Skills
Brady, Shankweiler, and Mann (1983). In
considering a group of beginning readers There are quite a few indications that read-
who did not differ from each other in age, ing ability is related to certain vocabulary
IQ, or audiometry scores but strongly dif- skills, depending on how reading ability is
fered in reading ability, they asked the chil- measured and on what type of vocabulary
dren to identify spoken words or environ- skill is at issue. Reading ability can be mea-
mental sounds under a normal listening sured in terms of the ability to read individ-
condition and under a “noisy” condition. ual words (e.g., in terms of “decoding”) or
When the performances of the good and in terms of the ability to understand the
poor readers were compared, the good and meaning of sentences and paragraphs (e.g.,
poor readers were equally able to identify in terms of “comprehension”). In the case
the environmental sounds, whatever the lis- of beginning readers, decoding and compre-
tening condition. They were also equivalent hension tests are correlated quite highly, im-
on the unmasked words. However, the poor plying that children who differ on one type
readers made almost 33% more errors than of test will usually differ on the other as
did the good readers when they were asked well. Still, there are cases in which the two
to identify the spoken words in the “noisy” types of tests identify different groups of
condition. good and poor readers that may lead re-
searchers to different conclusions about the
Other research has confirmed that chil- cause of poor reading (see Stanovich, 1988,
dren who are poor readers may have prob- for discussion). Vocabulary skills are a case
lems perceiving speech. They require a in point; future research may uncover other
longer segment of a gated word to perceive cases as well.
it correctly (Metsala, 1997). At least some
poor readers do not perceive synthetic Vocabulary skills can be tested with two
speech as categorically as good readers do basic types of test. “Recognition vocabu-
(Manis et al., 1997; Mody, Studdert- lary” tests such as the Peabody Picture Vo-
Kennedy, & Brady, 1997). Following Walley cabulary Test require the child to point to a
(1993), Metsala (1997) suggested that the picture that illustrates a word. Recognition
perceptual problem associated with poor vocabulary has sometimes been related to
reading and the concomitant difficulty with early reading ability (see Stanovich, Cun-
phoneme awareness may have a common ningham, & Feeman, 1984), although it is
source. Both may follow from the fact that not always a significant predictor (see
the mental representation of phonemes may Mann, 1984; Wolf & Goodglass, 1986).
gradually develop over childhood, as the The utility of this test may depend on how
growth of spoken vocabulary causes lexical “reading ability” is measured, as the rela-
representations to become more segmental. tionship seems stronger for tests of reading
Similarly, Manis and his colleagues (1997) comprehension, such as the Reading Survey
have proposed that if children cannot per- of the Metropolitan (see Stanovich, Nathan,
ceive clear distinctions between phonemes it & Zolman, 1988), than for tests of word
will be hard for them to have representa- recognition such as the Word Identification
tions that can be easily accessed. Problems and Word Attack tests of the Woodcock (see
with accessing phonological representations Mann & Liberman, 1984).
will in turn lead to difficulty segmenting
and manipulating phonemes and in learning “Naming” or “productive vocabulary”
grapheme–phoneme relationships. Foy and tests such as the Boston Naming Test re-
quire the child to produce the word that a

222 CAUSES AND BEHAVIORAL MANIFESTATIONS

picture illustrates. Productive vocabulary tunno, 1990; see also Stanovich et al., 1988).
give clearer indications of a link between Thus it is likely that something other than a
reading ability and vocabulary skill and lack of educational experience is preventing
there is evidence that this link exists these children from naming the letter names
whether reading skill is measured in terms as fast as other children can.
of decoding or in terms of comprehension.
Performance on the Boston Naming Test, Should problems with letter naming to be
for example, predicted both the word recog- viewed in the context of language and
nition and the reading comprehension abili- phonological representation, or do they owe
ty of kindergarten children far more accu- to something outside the realm of language?
rately than did performance on the Peabody One pertinent piece of evidence about the
Picture Vocabulary Test (Wolf & Good- naming problems of poor readers comes
glass, 1986). In one study of the naming from a study by Katz (1986), who found
speed among individuals with dyslexia, the that children who perform poorly on a de-
performance of 17-year-olds was closest to coding test are particularly prone to difficul-
that of 8-year-old normal children for col- ties in producing low frequency and poly-
ors, digits, letters, and pictures of common syllabic names, and suggested that for such
objects (Fawcett & Nicolson, 1994). Tests words, these children may possess less
of continuous naming (sometimes called “phonologically complete” lexical represen-
rapid automatized naming), which require tations than do good readers. On the basis
children to name a series of repeating ob- of his research, he further suggests that be-
jects, letters, or colors, have also shown that cause poor readers often have access to as-
children who are poor readers take longer pects of the correct phonological represen-
to name the series than do good readers tation of a word, even though they are
(see, e.g., Blachman, 1984; Denckla & unable to produce that word correctly, their
Rudel, 1976; Wolf, 1984). Performance on problem may be attributable to phonologi-
these tests bears an interesting relationship cal deficiencies in the structure of the lexi-
to the success of certain remediation mea- con rather than to the process of lexical ac-
sures (Torgesen & Davis, 1996). cess per se. McBride-Chang (1996) similarly
argues from her extensive studies of phono-
A causal link between naming problems logical skills among elementary-school-age
and reading problems is indicated by the dis- children that naming speed owes its effect
covery that performance on naming tests can on reading to its relation with speech per-
predict future reading ability (for a review, ception and phonological processing.
see Bowers & Wolf, 1993). Wolf (1984) has
noted that whereas continuous naming tests However, Wolf and her colleagues (Wolf
using objects and colors are predictive of ear- & Bowers, 1999; Wolf, Bowers, & Biddle,
ly problems with word recognition, prob- 2000) have recently countered such propos-
lems with rapid letter recognition and re- als with a view that naming problems stem
trieval can play a more prolonged role in the from a broader source than deficient phono-
reading of severely impaired readers, com- logical representation. In their view, naming
promising both decoding and reading com- speed deficiencies may reflect a more perva-
prehension. Using a test of letter-naming sive rate-of-processing problem that affects
ability in our longitudinal studies of kinder- varied aspects of reading and other modali-
garten children (e.g., Mann, 1984; Mann & ties as well. Their “double-deficit” account
Ditunno, 1990) my colleagues and I have proposes that separate deficits can underlie
consistently found that kindergarteners who problems in the phonological system and
take longer to name a randomized array of problems in rapid serial naming. It is the
the capital letters are significantly more like- children who suffer from both deficits that
ly to perform poorly on word decoding tests become the poorest readers.
and comprehension tests that are adminis-
tered in first grade. It further seems that pre- Phonetic Working Memory
sent letter naming predicts future reading
ability more consistently than present read- One of the more fruitful lines of research in
ing ability predicts future letter naming abil- the field of reading disability stems from the
ity (for relevant evidence, see Mann & Di- observation that poor readers perform less
well than do good readers on a variety of

Language Processes 223

working memory or “short-term” memory One might ask, at this point, whether
tests. It has often been noted that poor read- poor readers are avoiding phonetic repre-
ers tend to perform less well on the digit sentation altogether or merely using it less
span test and are deficient in the ability to well. We have obtained little evidence that
recall strings of letters, nonsense syllables, poor readers employ a visual form of mem-
or words in order, whether the stimuli are ory instead of a phonetic one (Mann, 1984).
presented by ear or by eye. Poor readers Evidence that poor readers are attempting
even fail to recall the words of spoken sen- to use phonetic representation has been
tences as accurately as good readers do found in the types of errors that they make
(for reviews, see Brady & Shankweiler, as they attempt to recall or recognize spo-
1991; Jorm, 1979; Mann, Liberman, & ken words in a short-term memory task
Shankweiler, 1980; McBride-Chang, 1996; (Brady et al., 1983; Brady, Mann, &
Shankweiler, Liberman, Mark, Fowler, & Schmidt, 1987). These errors reveal that
Fischer, 1979, for references to these ef- poor readers make use of many of the same
fects). Evidence that these differences are features of phonetic structure as good read-
not merely consequences of differences in ers do. They make the same sort of phoneti-
reading ability has come from a longitudinal cally principled errors—they merely make
study which showed that problems with re- more of them.
calling a sequence of words can precede the
attainment of reading ability and may actu- Syntax and Semantics
ally serve to presage future reading prob-
lems (Mann & Liberman, 1984). The observation that poor readers cannot
repeat spoken sentences as accurately as
In searching for an explanation of this pat- good readers has led to some obvious ques-
tern of results, researchers were inspired by tions about higher-level language skills and
research indicating that linguistic materials their involvement in reading problems. To
such as letters, words, and so on, are held in date, quite a few studies have examined the
short-term memory through use of phonetic syntactic abilities of poor readers. There is
representation. Liberman, Shankweiler, and an accumulating body of evidence that poor
their colleagues (Shankweiler et al., 1979) readers do not comprehend sentences as
were the first to suggest that the linguistic well as good readers do (see Mann, Cowin,
short-term memory difficulties of poor read- & Schoenheimer, 1989, for a review). It has
ers might reflect a problem with using this been shown that good and poor readers dif-
type of representation. Several experiments fer in the ability to both repeat and compre-
have supported this hypothesis. These exper- hend spoken sentences that contain relative
iments show that when recalling letter clauses such as “The dog jumped over the
strings (Shankweiler et al., 1979), word cat that chased the monkey” (Mann,
strings (Mann et al., 1980; Mann & Liber- Shankweiler, & Smith, 1986). They also
man, 1984), and sentences (Mann et al., perform less well on instructions from the
1980), poor readers are much less sensitive Token Test, such as “Touch the small red
than good readers to a manipulation of the square and the large blue triangle” (Smith,
phonetic structure of the materials (i.e., the Mann, & Shankweiler, 1987). They also are
density of words that rhyme). Indeed, good less able to distinguish the meaning of spo-
readers can be made to appear as error-prone ken sentences such as “He showed her bird
as poor readers when they are asked to recall the seed” from “He showed her the bird-
a string of words in which all of the words seed,” which use the stress pattern of the
rhyme (such as “bat, cat, rat, hat, mat”), sentence (its “prosody”) and the position of
whereas poor readers perform at the same the article “the” to mark the boundary be-
level whether or not the words rhyme. This tween the indirect object and the direct ob-
observation has led to the postulation that ject.
poor readers—and children who are likely to
become poor readers—are for some reason In searching to explain these and other
less able to use phonetic structure as a means sentence comprehension problems that have
of holding material in short-term memory been observed among the poor readers we
(Mann et al., 1980; Mann & Liberman, have studied, my colleagues and I have been
1984; Shankweiler et al., 1979). struck by the fact that a short-term memory

224 CAUSES AND BEHAVIORAL MANIFESTATIONS

problem could lead to problems with com- further research. Such deficits as do exist
prehending sentences whose processing are relatively subtle, with poor readers
somehow stresses short-term memory. merely performing like somewhat younger
When we examined the results of the afore- children than do the good readers.
mentioned studies, we found little evidence
that the poor readers were having trouble As for the question of semantic impair-
with the grammatical structures being used ments among poor readers, here, there is no
in the sentences that caused them problems. reason to presume that any real deviance
In fact, the structures were often ones which exists. If anything, poor readers place
young children master within the first few greater reliance on semantic context and se-
years of life and ones which the poor read- mantic representation than do good readers,
ers could understand if the sentence was perhaps in compensation for their other lan-
short enough (see Mann et al., 1989, for a guage difficulties (see Stanovitch, 1982, for
discussion). Instead, we found much evi- a review; see also Simpson, Lorsbach, &
dence that the comprehension problem was Whitehouse, 1983).
predominantly due to the phonetic working
memory problem discussed in the previous Conclusion and Applications
section. It seems as if poor readers are just
as sensitive to syntactic structure as good The literature on the relation between lan-
readers; they fail to understand sentences guage processing skills and reading prob-
because they cannot hold an adequate rep- lems indicates that poor readers—and
resentation of the sentence in short-term children who are likely to become poor
memory (see Gottardo, Stanovich, & Siegel, readers—tend to have problems with
1996; Mann et al., 1985, 1989; Smith, phoneme awareness, morpheme awareness,
Macaruso, Shankweiler, & Crain, 1989). and three aspects of language processing
skill: (1) speech perception under difficult
Another task that at first glance seems to listening conditions; (2) vocabulary, espe-
indicate a syntactic impairment on the part cially when measured in terms of naming
of poor readers is the sentence completion ability; and (3) using phonetic representa-
task used by Mahony (1994) and Singson tion in linguistic short-term memory. As
and colleagues (2000). That task requires Stanovich has noted, there is a logical inter-
children to choose among derivational relation between the behavioral difficulties
forms that complete a sentence, as in “He of poor readers, for they all involve
was blinded by the __________: bright, “phonological processes” that concern the
brighten, brightly, brightness.” Poorer read- sound pattern of language. Hence we may
ers, both children and adults, perform less speculate that the cause of many instances
well on this test than do normal readers, of reading disability involves a certain dys-
whether they are reading the sentences or function within the “phonological” sys-
hearing them aloud, and whether the word tem—a “phonological core deficit” (see
that fills in the blank is a real word or an Stanovich, 1988, and Stanovich & Beck,
appropriately derived word such as “frood- 2000, for further discussion).
ness.” Here the poor readers’ difficulty in
distinguishing among nouns, verbs, and ad- The research surveyed by this chapter
jectives appears to be due to a problem with may be of helpful interest to those who are
derivational morphology and not a problem concerned with the remediation of reading
with syntax. problems. As we come closer and closer to
identifying the linguistic problems associat-
At present, then, although it is clear that ed with specific reading difficulty—and
poor readers do have sentence comprehen- their causes—we should also come closer to
sion problems, there is little reason to think being able to point the way toward more ef-
that their difficulties reflect a problem with fective procedures for their remediation. As
the syntax of language. Problems with for remediation of deficient morpheme
working memory and problems with mor- awareness and deficiencies in such process-
phology seem a more likely source of the ing skills as speech perception, naming, and
difficulty. But the issue of whether or not phonetic representation, the jury is still out.
poor readers are deficient in syntactic skills It is certainly reasonable to think that ex-
is far from resolved and will have to await

Language Processes 225

plicit practice and training can benefit some, Baddeley, A. D. (1978). The trouble with Levels: A
if not all, of these skills, but clearly more re- reexamination of Craik and Lockhardt’s frame-
search is needed. In the meantime it would work for memory research. Psychological Re-
seem best to use clear enunciation, repeti- view, 85, 139–152.
tion, and drilling to compensate for the bot-
tleneck created by spoken language prob- Berent, I., & Perfetti, C. A. (1995). A rose is a
lems. REEZ: The two-cycles model of phonology as-
sembly in reading English. Psychological Review,
Certainly the brightest prospects for reme- 102, 146–184.
diation are offered by evidence that various
types of training can facilitate phoneme Blachman, B. (1984). Relationship of rapid naming
awareness. The best favor we can do for all and language analysis skills to kindergarten and
children is to promote their phoneme aware- first-grade reading achievement. Journal of Edu-
ness so that we may let them in on the secrets cational Psychology, 76, 610–622.
of the alphabetic principle as early as possi-
ble. Some interesting and practical advice on Bowers, P. G., & Wolf, M. (1993). Theoretical links
how to facilitate phoneme awareness is cur- among naming, speed, precise timing mecha-
rently available from the work of such re- nisms, and orthographic skills in dyslexia. Read-
searchers as Liberman, Blachman, Torgesen, ing and Writing, 5, 69–85.
and Bradley and their colleagues. A variety
of word games, nursery rhymes, and other Bradley, L., & Bryant, P. E. (1978). Difficulties in
“prereading” activities exist that can help auditory organization as a possible cause of read-
nurture a child’s awareness of the way in ing backwards. Nature, 271, 746–747.
which words break down into phonemes.
Such activities will undoubtedly pave the Bradley, L., & Bryant, P. (1985). Rhyme and reason
way for “phonics”-oriented methods of in- in reading and spelling. Ann Arbor: University of
struction so obviously favored by current re- Michigan Press.
search and so obviously in keeping with this
chapter’s focus on the importance of Brady, S., Mann, V., & Schmidt, R. (1987). Errors
phoneme awareness in early reading. in short-term memory for good and poor readers.
Memory and Cognition, 15, 444–453.
Cunningham’s (1990) research on
phoneme awareness training offers the Brady, S., & Shankweiler, D. (1991). Phonological
caveat that we should pay attention not only processes in literacy. Hillsdale, NJ: Erlbaum.
to the activities that promote phoneme
awareness but also to how we are integrating Brady, S. Shankweiler, D., & Mann, V. (1983).
these into instruction. Children need to ap- Speech perception and memory coding in relation
preciate the value application and utility of to reading ability. Memory and Cognition, 35,
phoneme awareness to reading, and activi- 345–367.
ties that encourage phoneme awareness are
the most beneficial when the children are Carlisle, J. F. (1995). Morphological awareness and
shown how these activities are beneficial to early reading achievement. In L. Feldman (Ed.),
reading. Other caveats come from Torgesen Morphological aspects of language processing
and Davis (1996), who have united screen- (pp. 189–209). Hillsdale, NJ: Erlbaum.
ing and training to show how one predicts
the success of the other. In their research, Carlisle, J. F. (2000). Awareness of morphological
children who reveal a three-way deficiency in structure and meaning: Impact on reading. Read-
phoneme awareness, letter naming, and ing and Writing, 12, 169–190.
phonemic representation may require much
longer, more explicit training than the typical Carlisle, J. F., & Nomanbhoy, D. M. (1993).
8–12-week course seen in the literature. Phonological and morphological awareness in
first graders. Applied Psycholinguistics, 14,
References 177–195.

Adams, M. J. (1990). Beginning to read: thinking Chomsky, N. (1964). Comments for project literacy
and learning about print. Cambridge, MA: MIT meeting. In M. Lester (Ed.), Reading in applied
Press. transformational grammar. New York: Holt
Rinehart & Winston.

Conrad, R. (1964). Acoustic confusions in immedi-
ate memory. British Journal of Psychology, 55,
75–84.

Conrad, R. (1972). Speech and reading. In J. F. Ka-
vanaugh & I. G. Mattingly (Eds.), Language by
ear and by eye: The relationships between speech
and reading (pp. 205–240). Cambridge, MA:
MIT Press.

Cunningham, A. E. (1990). Explicit versus implicit
instruction in phoneme awareness. Journal of Ex-
perimental Child Psychology, 50, 429–444.

Daneman, M., & Carpenter, P. A. (1980). Individ-
ual differences in working memory and reading.
Journal of Verbal Learning and Verbal Behavior,
19, 450–466.

Denckla, M. B., & Rudel, R. G. (1976). Naming of
object drawings by dyslexic and other learning-
disabled children. Brain and Language, 3, 1–15.

226 CAUSES AND BEHAVIORAL MANIFESTATIONS

Drenowski, A., & Healy, A. F. (1980). Missing –ing Katz, R. B. (1986). Phonological deficiencies in chil-
in reading: Letter detection errors in word end- dren with reading disability: Evidence from an
ings. Journal of Verbal Learning and Verbal Be- object naming task. Cognition, 22, 225–257.
havior, 19, 247–262.
Kleiman, G. (1975). Speech recoding in reading.
Elbro, C. (1996). Early linguistic abilities and read- Journal of Verbal Learning and Verbal Behavior,
ing development: A review and a hypothesis. 14, 323–339.
Reading and Writing, 8, 453–485.
Levy, B. A. (1977). Reading: Speech and meaning
Fawcett, A. J., & Nicolson, R. I. (1994). Naming processes. Journal of Verbal Learning and Verbal
speed in children with dyslexia. Journal of Learn- Behavior, 16, 623–638.
ing Disabilities, 27, 641.646.
Liberman, A. M. (1999). The reading researcher
Fischer, F. W., Liberman, I. Y., & Shankweiler, D. and the reading teacher need the right theory of
(1977). Reading reversals and developmental speech. Scientific Studies of Reading, 3, 95–111.
dyslexia: A further study. Cognition, 14,
496–510. Liberman, I. Y. (1982). A Language-oriented view
of reading and its disabilities. In H. Mykelburst
Fletcher, J. M., Francis, D. J., Rourke, B. P., Shay- (Ed.), Progress in learning disabilities (Vol. 5, pp.
witz, S. E., & Shaywitz, B. A. (1992). The validi- 81–101). New York: Grune & Stratton.
ty of discrepancy-based definitions of reading dis-
abilities. Journal of Learning Disabilities, 25, Liberman, I. Y., Liberman, A. M., Mattingly, I. G., &
555–561. Shankweiler, D. (1980). Orthography and the be-
ginning reader. In J. Kavanaugh & R. Venezky
Fowler, A. E. (1991). How early phonological de- (Eds.), Orthography, reading and dyslexia (pp.
velopment might set the stage for phoneme 137–154). Baltimore: University Park Press.
awareness. In S. Brady & D. Shankweiler (Eds.),
Phonological processes in literacy: A tribute to Liberman, I. Y., Shankweiler, D., Fischer, F. W., &
Isabelle Y. Liberman (pp. 97–117). Hillsdale, NJ: Carter, B. (1974). Explicit syllable and phoneme
Erlbaum. segmentation in the young child. Journal of Ex-
perimental Child Psychology, 18, 201–212.
Fowler, A., & Liberman, I. (1995). The role of
phonology and orthography in morphological Lima, S. D. (1987). Morphological analysis in sen-
awareness. In L. Feldman (Ed.), Morphological tence reading. Journal of Memory and Language,
aspects of language processing (pp. 157–188). 26, 84–99.
Hillsdale, NJ: Erlbaum.
Mahony, D. L. (1994). Using sensitivity to word
Fox, B., & Routh, D. K. (1976). Phonemic analysis structure to explain variance in high school and
and synthesis as word-attack skills. Journal of college level reading ability. Reading and Writing,
Educational Psychology, 69, 70–74. 6, 19–44.

Foy, J. G., & Mann, V. (2001). Does strength of Manis, F. R., McBride-Chang, C., Seidenberg, M.
phonological representations predict phonologi- S., Keating, P., Doi, L. M., Munson, B., &
cal awareness? Applied Psycholinguistics, 22, Petersen, A. (1997). Are speech perception
301–325. deficits associated with developmental dyslexia?
Journal of Experimental Child Psychology, 66,
Francis, D. J., Shaywitz, S. E., Stuebing, K. K., 211–235.
Shaywitz, B. A., & Fletcher, J. M. (1996). Devel-
opmental lag versus deficit accounts of reading Mann, V. A. (1984). Longitudinal prediction and
disability: A longitudinal, individual growth prevention of early reading difficulty. Annals of
curves analysis. Journal of Educational Psycholo- Dyslexia, 34, 117–136.
gy, 88, 3–17.
Mann, V. A. (1993). Phoneme awareness and future
Frost, R. (1998). Toward a strong phonological the- reading ability. Journal of Learning Disabilities,
ory of visual word recognition: True issues and 26, 259–269.
false trails. Psychological Bulletin, 123, 71–99.
Mann, V. A., Cowin, E., & Schoenheimer, J. (1989).
Gottardo, A., Stanovich, K., & Siegel, L. (1996). Phonological processing, language comprehen-
The relationships between phonological sensitivi- sion and reading ability. Journal of Learning Dis-
ty, syntactic processing, and verbal working abilities, 22, 76–89.
memory in the reading performance of third-
grade children. Journal of Experimental Child Mann, V. A., & Ditunno, P. (1990). Phonological
Psychology, 63, 563–582. deficiencies: effective predictors of future reading
problems. In G. Pavlides (Ed.), Dyslexia: Neu-
Hung, D. L., & Tzeng, O. J. L. (1981). Ortho- ropsychological and learning perspectives (pp.
graphic variations and visual information pro- 105–131). New York: Wiley.
cessing. Psychological Bulletin, 90, 377–414.
Mann, V. A., & Liberman, I. Y. (1984). Phonologi-
Hulme, C, & Joshi, R. M. (1998). Reading and cal awareness and verbal short-term memory:
spelling: Development and disorders. Mahwah, Can they presage early reading success? Journal
NJ: Erlbaum. of Learning Disabilities, 17, 592–598.

Jackson, M., & McClelland, J. L. (1979). Process- Mann, V. A., Liberman, I. Y., & Shankweiler, D.
ing determinants of reading speed. Journal of Ex- (1980). Children’s memory for sentences and
perimental Psychology: General, 108, 151–181. word strings in relation to reading ability. Memo-
ry and Cognition, 8, 329–335.
Jorm, A. F. (1979). The cognitive and neurological
basis of developmental dyslexia: A theoretical Mann, V. A., Shankweiler, D., & Smith, S. T.
framework and review. Cognition, 7, 19–33. (1985). The association between comprehension
of spoken sentences and early reading ability:

Language Processes 227

The role of phonetic representation. Journal of Shankweiler, D., Liberman, I. Y., Mark, L. S.,
Child Language, 11, 627–643. Fowler, C. A., & Fischer, F. W. (1979). The
Mann, V. & Wimmer, H. (2002). Phoneme aware- speech code and learning to read. Journal of Ex-
ness and pathways to literacy: A comparison of perimental Psychology: Human Perception and
German and American children. Reading and Performance, 5, 531–545.
Writing, 15, 653–682.
Mattingly, I. G. (1972). Reading, the linguistic Shankweiler, D., Crain, S., Katz, L., Fowler, A. E.,
process, and linguistic awareness. In J. F. Ka- Liberman, A. E., Brady, S. A., Thornton, R.,
vanaugh & I. G. Mattingly (Eds.), Language by Lundquist, E., Dreyer, L., Fletcher, J. M., Stue-
ear and by eye: The relationship between speech bing, K. K., Shaywitz, S. E., & Shaywitz, B. A.
and reading (pp. 133–148). Cambridge, MA: (1995). Cognitive profiles of reading-disabled
MIT Press. children: Comparisons of language skills in
McBride-Chang, C. (1996). Models of speech per- phonology, morphology and syntax. Psychologi-
ception and phonological processing in reading. cal Science, 6, 149–156.
Child Development, 67, 1836–1856.
Metsala, J. (1997). Spoken word recognition in Simpson, G. B., Lorsbach, T. C., & Whitehouse, D.
reading disabled children. Journal of Educational (1983). Encoding and contextual components of
Psychology, 89(1), 159–173. word recognition in good and poor readers. Jour-
Mody, M., Studdert-Kennedy, M., & Brady, S. nal of Experimental Child Psychology, 35,
(1997). speech perception deficits in poor read- 161–171.
ers: Auditory processing or phonological coding?
Journal of Experimental Child Psychology, 64, Singson, M., Mahony, D., & Mann, V. (2000). The
199–231. relation between reading ability and morphologi-
Morais, J., Cary, L., Alegria, J., & Bertelson, P. cal skills: Evidence from derivational suffixes.
(1979). Does awareness of speech as a sequence Reading and Writing, 12, 219–252.
of phonemes arise spontaneously? Cognition, 7,
323–331. Slowiaczek, M. L., & Clifton, C. (1980). Subvocal-
Muter, V., Hulme, C., Snowling, M., & Taylor, S. ization and reading for meaning. Journal of Ver-
(1997). Segmentation, not rhyming, predicts ear- bal Learning and Verbal Behavior, 19, 573–582.
ly progress in learning to read. Journal of Experi-
mental Child Psychology, 65, 370–396. Smith, S. T., Macaruso, P., Shankweiler, D., & Crain,
Pennington, B. F., Van Orden, G., Kirson, D., & S. (1989). Syntactic comprehension in young poor
Haith, M. (1991). What is the causal relation be- readers. Applied Psycholinguistics, 10, 429–454.
tween verbal STM problems and dyslexia? In S.
A. Brady & D. P. Shankweiler (Eds.), Phonologi- Smith, S. T., Mann, V. A., & Shankweiler, D.
cal processing skills in literacy (pp. 173–186). (1986). Spoken sentence comprehension by good
Hillsdale, NJ: Erlbaum. and poor readers: A study with the token test.
Perfetti, C. A. (1985). Reading skill. Hillsdale, NJ: Cortex, 22, 627–632.
Erlbaum.
Rayner, K., & McConkie, G. W. (1976). What Stanovich, K. (1982). Individual differences in the
guides a readers’ eye movements? Vision Re- cognitive processes of reading: II. Text-level
search, 16, 829–837. processes. Journal of Learning Disabilities, 15,
Rayner, K., Sereno, S. C., Lesch, M. F., & Pollatsek, 549–554.
A. (1995). Phonological codes are automatically
activated during reading: Evidence from an eye Stanovich, K. (1988). Explaining the differences be-
movement priming paradigm. Psychological Sci- tween the dyslexic and the garden-variety poor
ence, 6, 26–32. reader: The phonological-core variable difference
Read, C, Zhang, Y., Nie, H., & Ding, B. (1986). model. Journal of Learning Disabilities, 21,
The ability to manipulate speech sounds depends 590–604.
on knowing alphabetic writing. Cognition, 24,
31–44. Stanovich, K. E., & Beck, I. (2000). Progress in un-
Russell, G. (1982). Impairment of phonetic reading derstanding reading. New York: Guilford Press.
in dyslexia and its persistence beyond child-
hood—Research note. Journal of Child Psycholo- Stanovich, K. E., Cunningham, A. E., & Cramer, B.
gy and Child Psychiatry, 23, 459–475. B. (1984). Assessing phonological awareness in
Rutter, M. (1978). Prevalence and types of dyslexia. kindergarten children: Issues of task comparabili-
In A. L. Benton & D. Pearl (Eds.), Dyslexia: An ty. Journal of Experimental Child Psychology,
appraisal of current knowledge (pp. 3–28) New 38, 175–190.
York: Oxford University Press.
Shankweiler, D., & Liberman, I. Y. (1972). Misread- Stanovich, K. E., Cunningham, A. E., & Feeman, D.
ing: A search for the causes. In J. F. Kavanaugh & J. (1984). Intelligence, cognitive skills and early
I. G. Mattingly (Eds.), Language by ear and by reading progress. Reading Research Quarterly,
eye: The relationships between speech and reading 19, 278–303.
(pp. 293–318). Cambridge, MA: MIT Press.
Stanovich, K. E., Nathan, R. G., & Zolman, J. E.
(1988). The developmental lag hypothesis in
reading: Longitudinal and matched reading-level
comparisons. Child Development, 59, 71–86.

Taft, M., & Forster, K. I. (1975). Lexical storage
and retrieval of prefixed words. Journal of Verbal
Learning and Verbal Behavior, 14, 638–647.

Taft, M. (1984). Evidence for an abstract lexical
representation of word structure. Memory and
Cognition, 12, 264–269.

Torgesen, J. K., & Davis, C. (1996). Individual dif-
ference variables that predict response to training