Paula Clarke, Charles Hulme, Emma Truelove, Maggie Snowling The York READing for Meaning Project Examining the Long Term Effects of Three Interventions to Support Reading Comprehension in Poor Comprehenders      Paula Clarke, Charles Hulme, Emma Truelove, Maggie Snowling

Poor Comprehenders Average word readers but poor at reading comprehension Difficulties with wider oral language skills Vocabulary Narrative Figurative Language Inferencing Verbal reasoning Interventions to date have generated impressive immediate gains in reading comprehension But studies have been small in scale and have focused on text level and metacognitive skills

Project aims Text level training in written language domain Improvements in text comprehension Oral language training in spoken language domain Text level training in written language domain Oral language training in spoken language domain 3

Project aims To compare these approaches to existing classroom practice by monitoring the performance of an untreated waiting control group. To explore the effectiveness of the three approaches in the immediate and long term. To examine mediating and moderating factors that influence response to intervention.

Randomised Controlled Trial Design TC block 1 OL block COM block 1 Control COM block 2 Control TC block 2 OL block TC OL COM Control Maintenance test Mid test Post test Screening Pre test Control block 1 Control block 2 Oct - April 2006 8-9 years July 2007 9 years Dec-Jan 2007 9-10 years Dec 2008 10-11 years July 2009

Participant profile Group 1 Group 2 Group 3 Group 4 Whole sample TOWRE Real words 110.48 108.88 108.30 108.70 109.09 NARA II Reading accuracy 102.93 104.33 101.55 NARA II Reading comprehension 92.95 93.36 92.82 92.99 WIAT II Reading comprehension 94.31 95.43 96.38 97.77 95.96 CELF Listening comprehension 78.75* 80.42 87.36* 83.06 82.40 WASI Vocabulary (t score) 38.31 40.32 39.44 38.26 39.07 WASI Similarities (t score) 47.40 49.05 49.97 49.49 48.97 WASI VIQ 89.41 92.03 91.89 91.00 91.07 Ravens Matrices NVIQ 99.26 100.23 100.66 100.78 *Statistically significant between groups difference (p<0.05)

Programme contents and features Text Comprehension Written Language Context Reading Comprehension Metacognitive Strategies Inferencing from Text Narrative Oral Language Spoken Language Context Listening Comprehension Vocabulary Figurative Language Narrative Combined All eight components Sessions contained both reading and listening comprehension Opportunities for children to encounter new vocabulary/idioms/inferences in both written and spoken language.

Primary outcome - Text comprehension pale = immediate post test bright = maintenance post test Children read (aloud or silently) a range of passages and sentences. (narrative, adverts, non-fiction information etc.) Includes literal, inference and vocabulary dependent question types. Also involves summarisation (finding key themes and ideas), prediction, and question generation.

Secondary outcome - Vocabulary Children orally define a series of tier 2 words (AoA 13 years +, unfamiliar but applicable to a number of contexts) Half of the words are those targeted in the OL and COM interventions and half are non targeted. Definition quality is quantified using a 0,1,2 scoring system, developed from pilot study responses.

Mediation model Time 1 Time 3 Time 4 Chi-Square Test 0.494 df 1 n.s WIAT II Text Comprehension p = .000 p= .000 TC WIAT II Text Comprehension p= .04 N.S. p= .004 OL N.S. p= .000 p= .001 COM p= .000 Vocabulary Chi-Square Test 0.494 df 1 n.s

VIQ as a moderator Standard score points Low VIQ (n=75) High VIQ (n=79) TOWRE Real words 107.83 110.89 NARA II Reading accuracy 102.07 104.19 NARA II Reading comprehension** 91.04 95.33 WIAT II Reading comprehension** 94.16 97.58 CELF Listening comprehension* 80.15 84.87 TOWRE -NARA II discrepancy 16.63 15.73 WASI VIQ*** 83.09 98.65 Ravens Matrices NVIQ Standard score points WIAT II pale = pre test bright = maintenance post test Statistically significant between groups differences *p<0.05,**p<0.01,***p<0.001 VIQ significant predictor of improvement High VIQ make greater gains than Low VIQ No interaction between VIQ and condition

Summary In poor comprehenders, 20-week intervention programmes can produce significant gains in text comprehension and oral language. Importantly these gains are relative to an untreated waiting control group. The gains maintain over time and for the OL condition the difference between the intervention group and controls increases in significance. The gains in text comprehension are partially mediated by vocabulary learning in the OL condition and fully mediated by vocabulary learning in the COM condition. VIQ at pre test moderates response to intervention, however there are no significant differential effects of VIQ across conditions.

Implications Teaching assistants with a relatively small amount of training can deliver high quality effective teaching. The skills that underpin oral language and text comprehension are trainable in children aged 8-11 years. Evidence to suggest a causal relationship between vocabulary and text comprehension. Improvements in text comprehension can be explained, at least in part, by improvements in vocabulary skill.

For more information please contact Thank you For more information please contact pjc118@york.ac.uk http://www.york.ac.uk/res/crl/readme.html