Reading therapy in deep dyslexia

Renata Konecny a,*, Sharon Lee Armstrong b, Nadine Martin c

a Ancora Psychiatric Hospital, Ancora, NJ 08037, USAb Psychology Department, La Salle University, Philadelphia, PA 19141, USA

c Department of Communication Sciences, Temple University, Philadelphia, PA 19122, USA

Accepted 6 July 2006

Introdution

Treatments for phonological/deep dyslexia have used sublexical strat-egies to improve grapheme–phoneme conversion (GPC) and sound blend-ing abilities and lexical strategies to improve sight vocabulary that cannotbe read using GPC rules. The re-establishment of sound-blending abilitieshas proved to be challenging in many cases. This study used a combinedsublexical–lexical approach to improve oral reading performance whichwas motivated by two factors. Since the patient (AP) had not shown evi-dence of natural resolution of deep dyslexic symptoms thirteen years postinjury, it was necessary to begin with reestablishment of GPC rules beforeaddressing sound blending abilities and whole word reading. A combinedapproach is also consistent with Hillis and Caramazza’s summationhypothesis (1991, 1995), positing that the process of oral reading combinesinformation from both lexical and sublexical sources.

The primary goal was to develop a fine-grained analysis of specificeffects of oral reading therapy as motivated by the dual model of oral read-ing. The secondary goal was to meet the patient’s goal of learning to readthe Checker’s menu and his favorite programs from the TV Guide.

Methods

Participant

AP was a monolingual, English-speaking, right-handed, 27-year-oldAfrican-American male who sustained extensive brain injury due to anarteriovenous malformation in the deep portion of the left fronto-tempo-ral region at age16. He demonstrated a symptom pattern consistent withnon-fluent aphasia, word-retrieval deficit, and deep dyslexia.

Background testing

Twenty three input and 18 output measures were selected to analyzeAP’s language strengths and weaknesses at the sublexical, lexical, andphrase level as predicted by the dual-route model of reading—subtestsof the PALPA, Philadelphia Comprehension Battery, Boston NamingTest, portions of Boston Diagnostic Aphasia Exam, Peabody Picture

Vocabulary Test, Pyramids and Palm Trees, Phoneme Discrimination,Rhyme Judgment and others.

Treatment

A single subject experimental design was used. First, a sublexical strat-egy was utilized to train grapheme–phoneme conversions and re-establishknowledge of grapheme–phoneme correspondences. This was followed byusing a straightforward phoneme blending approach. A lexical strategyinvolving massive exposure to, and repetition of, words and phrases wasused to improve AP’s reading of his favorite foods and TV titles. AP’streatment consisted of 58 one-hour sessions over nine months. Sets ofitems were trained until the accuracy criterion of missing no more than2 items was reached on two consecutive trials.

Results

AP’s post-treatment performance improved on all treatment tasks. SeeTable 1. He showed perfect performance on the letter-to-sound conversiontask (83–100%, p < .01), and improved performance on the sound-to-letterconversion task (87–97%, p < .05), indicating restoration of grapho-pho-nemic conversion ability.

AP achieved criterion level on the nonsense-syllable blending task(0–83%, p < .001, set one; 0–92%, p < .01, set two) demonstrating re-ac-quired grapho-phonemic skills in reading items in the targeted CVCstructure.

AP acquired sight vocabulary function words (0–75%, p < .001) via lex-ical training. He displayed a similar level of performance (0–88%, p < .001)on food names and titles of TV shows. His improved performance on a setof ‘‘fake’’ menu items (0–56%, p < .01) after completion of lexical trainingon real menu and TV titles indicated that his ability to read phrases wasnot the result of mere rote learning. Significant improvements on thetwo post-tests on the (untrained) Dolch List (33–36%, p < .02 and 36–39%, p < .05), reflect application of reacquired phonological skills ofgrapho-phonemic conversion and blending to words (e.g., am, went, by,my, so, much, just) he was unable to read lexically before treatment.

Discussion

Our results suggest that both lexical and sublexical treatmentapproaches improve single word oral reading in deep dyslexia. AP’s read-

doi:10.1016/j.bandl.2006.06.027

* Corresponding author. Fax: +1 609 567 7272.E-mail address: Renata.Konecny@dhs.state.nj.us (R. Konecny).

www.elsevier.com/locate/b&l

Brain and Language 99 (2006) 8–219

ing performance improved on all trained items and the selected CVCstructure. He also demonstrated generalization to untrained wordsand nonwords of the same CVC structure. Although no generalizationto untrained structures was observed, AP acquired strategies that heused spontaneously during training and applied on re-administrationof the measures. For example, if AP was not able to recognize an itemat sight, he attempted to decode; used his own symbols to cue himselfand dropped them when they were no longer needed. While learning toread selected menu and TV program items, he combined both sight-reading and phonological assembly strategies. Decoding the beginningof a word by sounding out and blending the first phonemes helpedhim ‘‘guess’’ the rest of the word. This suggests that further phonolog-ical training would be beneficial to him. AP was also successful learningselected sight vocabulary consisting of content and function words, sug-gesting that a lexical approach was also an effective treatment strategyfor him.

The most important insight gleaned from working with AP involvesthe length and intensity of treatment, carried out on a small number ofstimuli at a time. While massive exposure and repetition is not a new strat-egy for acquisition of new knowledge and skills, this simple techniqueacquires a different meaning when applied to severely impaired individu-als. For AP, massive repetition meant repeating one structure approxi-mately a thousand times over a period of weeks with the intensity of

two–three hours per week. Extremely important about AP’s course oftreatment is that he initially showed little progress, subsequent improve-ment was limited with extended plateaus, and then suddenly he ‘‘jumped’’to another level, moving on to more complex tasks (e.g., silent blending).Therefore, allowing a patient enough time to show progress seems to be acrucial treatment issue for those with severe and long-standingimpairments.

Acknowledgment

This study was supported by NIDCD Grant DC01924-11 awarded toTemple University (PI: N. Martin).

References

Hillis, A. E., & Caramazza, A. (1991). Mechanism for accessing lexicalrepresentations for output: Evidence for a category-specific semanticdeficit. Brain and Language, 40, 106–144.

Hillis, A. E., & Caramazza, A. (1995). Converging evidence for theinteraction of semantic and sublexical phonological information inaccessing lexical representations for spoken output. Cognitive Neuro-

psychology, 12, 187–227.

Table 1Targeted reading processing skills

Task Pre-test Post-test Chi-square (1) p-Value

Letter to sound 25/30 30/30 7.20 <.01Sound to letter 26/30 29/30 5.33 <.05Sight reading (function words) 0/12 9/12 11.11 <.001Blending-set 1 (nonsense syllables) 0/12 10/12 12.10 <.001Blending-set 2 (nonsense syllables) 4/12 11/12 9.14 <.01Reading phrases (menu, TV titles) 0/16 14/16 16.07 <.001Reading phrases (fake menu) 0/9 5/9 7.20 <.01Reading words

(Dolch list) 36/110 40/110a 6.25 <.02(Dolch list) 40/110 43/110b 5.33 <.05

a Post-test number 1, administered in the middle of Phase II (phonological training).b Post-test number 2, administered at the end of Phase III (lexical training).

Abstracts / Brain and Language 99 (2006) 8–219 35