December 201037-975-01
Presented by: Merav BadashDr. Armon Lotem
Psycholinguistic Markers for Specific Language Impairment (SLI)
Reading: Conti-Ramsden, G., Botting, N., Faragher, B. (2001)Journal of Child Psychology and Psychiatry, 6, 741-748.
Introduction
The study aims to compare and contrast psycholinguistic markers for SLI and examine whether such markers can differentiate between individuals with/without language disorder. Current literature examines the relationship between language impairment and specific verbal memory deficits as well aslimitations in processing capacity. Some of the skills examined are:
1. Phonological short-term memory tasks such as nonword repetition tests have recently been examined in children with SLI (Stothard et al., 1998; Gathercole et al., 1990). Such tasks are considered to be good indicators of underlying language difficulty even when the impairment appears to have resolved.
2. Syntactic tense has also been thought as a possible marker for identifying SLI (Marchman et al., 1999; Rice et al., 1996, 1998).
The clinical motivation for identifying markers can lead therapists to try alternative approaches and help identify markers most suited to intensive language provision in education.
The study investigates five main issues:
1.The independent contribution of markers to the identification of SLI.
2. The effect on identification accuracy of combining markers.
3. Identification of markers with relation to current language status which relates to children whose difficulties appear to be resolved at age 11 but can still be marked by these variables.
4. The relationship between children with a history of SLI and normally developed children.
5. The relationship between potential markers and nonverbal intelligence.
Method
Participants
160 Children with SLI. 11 years old with a documented history of SLI at age 7. 43 girls (27%); 22 children (14%) were exposed to languages other than English at home, average age 10;9.
100 Normally developing children.From three primary schools completed the marker tasks, rural and urban settings, 51 girls with a mean age of 10; 9.
a. No normative data on non word repetition, PTT, TPSwas available for this age group.
b. Sensitivity and specificity analysis requires available data for BOTH groups.
Procedure
Individual assessment was done in schools based on families' consent. The present study suggests a combinatory analysis in which more than one type of marker is examined in parallel; a combined power that may examine different types of memory, and processing abilities.
a. Marker tasks
*Past tense task (PTT, Marchman et al. 1999) – assessment of correct grammatical usage of verbs in past tense form.
*Third person singular task (TPS, Simkin & Conti-Ramsden, 2001) - assessment of correct grammatical usage of verbs in third person singular form.
*CELF-R Recalling Sentences subtest (RecS, Semel, Wiig &Secord, 1994) – exact repetition of sentences which became longer and more complex. Responses are scored in relation to the number of errors made in each sentence.
*Children'sTest of Nonword Repetition (CN Rep, Gathercole & Baddeley, 1990), assessment of verbal/phonological short term memory. The researcher says a nonword covering his lips to avoid any visual strategy. The child must repeat the words exactly.
b. Tests of non verbal cognition–children with SLI were also tested on Block Design and Picture Completion using the Wechsler Intelligence Scale for Children (WISC-III, 1992).
c. Language measures for comparison of markers across current language ability – children with SLI completed language tests on important areas of language development (vocabulary expression, syntactic comprehension, and word associations)
Results
*None of the marker tasks correlated highly with performance IQ in the SLI group. This point is important since any marker for SLI should not be a proxy for cognitive levels.
*Identification of children with SLI using marker tasks–each marker was analyzed for sensitivity and specificity. This analysis requires that threshold scores be used as cutoff points for predicting group membership.Children scoring at or below cutoff were defined as "impaired" and children scoring above the cutoff were classified as "nonimpaired".
Sensitivity = number of impaired children scoring below cutoff point/total number of impaired children; probability that an impaired child will be correctly identified by the test.
Specificity = number of non impaired children scoring above cutoff point/ total number of non impaired children; probability that a non impaired child will be correctly identified by the test.
The three cutoff points of clinical interest were: below the 16th centile (1 SD from population mean), below the 10th centile (1.25 SD from population mean), below the 2.5th centile (2 SD from population mean). Clinically, the 16th centile cutoff (-1 SD) proved to be the most useful cutoff point for correct SLI/incorrect SLI identification. According to Records &Tomblin (1994), this is the point where most clinicians make the decision that a child has SLI or not, see Table 1.
Table 1: Sensitivity, Specificity, and Accuracy of Tasks Using Various Thresholds
SensitivitySpecificity Overall accuracy
TPS
16th 63% (98/155) 90% (90/100) 74% (188/255)
10th 52% (81/155) 93% (93/100) 68% (174/255)
2.5th21% (33/155) 100% (100/100) 52% (133/255)
PTT
16th 74% (114/154) 89% (89/100) 80% (203/254)
10th 71% (109/154) 93% (93/100) 80% (202/254)
2.5th33% (51/154) 100% (100/100) 59% (151/254)
CNRep
16th 78% (124/159) 87% (87/100) 82% (211/259)
10th 74% (117/159) 92% (92/100) 81% (209/259)
2.5th42% (67/159) 98% (98/100) 64% (165/259)
RecS
16th 90% (144/160) 85% (85/100) 88% (229/260)
10th 86% (137/160) 92% (92/100) 88% (229/260)
2.5th 54% (87/160) 99% (99/100) 72% (186/260
The table shows several results: a. Markers differed in their accuracy of predicting children who did and did not have a history of SLI; b. TPS was the least useful test for the identification of children with SLI; c. the 16th centile cutoff point is the most favorable threshold for correct group identification useful; and d. RecS is the most accurate marker for SLI.
The effects of using different diagnostic cutoff points for each marker were evaluated using ROC curves (receiver operating characteristic, Dunn, 2000). The idea was to identify the "best" cutoff point for any of the tests. Results show that the recalling sentence task was the most predictive of the markers used (with an area of .9227 under ROC curve).
*Markers of children with SLI in relation to current language status
Coding of each child for how many tests fell below the 16th centile for age. The distribution was as follows: 17 (11%) had no tests below; 25 (16%) had one test below; 39 (24%) had two tests; a further 39 (24%) had three tests; and 40 children (25%) had all four tests below.
Sentence repetition and non word repetition are the only tasks that identified children with mild/resolved difficulties at a 50% level of accuracy or greater, see Fig. 3
*Relationship between markers – Recalling sentences is the only task to have consistently high correlations with all other tests (+PTT .62, +TPS .57, +CNRep .55)
*Combination of markers–The sensitivity and specificity of pairs of marker were evaluated by using two types of combinations:whether the child scored low on either task and whether the child scored low on both taskssee Table 2.
Table 2: Sensitivity, Specificity, and Accuracy of Combined Tasks
Sensitivity Specificity Overall accuracy
Either test combinations
TPS or PTT 82% (123/151) 82% (82/100) 82% (205/251)
TPS or CNRep86% (133/154) 81% (81/100) 84% (214/254)
TPS or RecS 93% (144/155) 78% (78/100) 87% (222/255)
PTT or CNRep 92% (140/153) 80% (80/100) 87% (220/253)
PTT or RecS 93% (143/154) 78% (78/100) 87% (221/259)
CNRep or RecS 96% (152/159) 78% (78/100) 89% (230/259)
Both test combinations
TPS and PTT 54% (82/151) 97% (97/100) 71% (179/251)
TPS and CNRep 55% (85/154) 96% (96/100) 71% (181/254)
TPS and RecS 60% (93/155) 97% (97/100) 75% (190/255)
PTT and CNRep 61% (93/153) 96% (96/100) 75% (189/253)
PTT and RecS 71% (110/154) 96% (96/100) 81% (206/254)
CNRep and RecS 73% (116/159) 94% (94/100) 81% (210/259)
The most impressive values are seen in "either test combinations" of CNRep or RecS
Discussion
The tasks involving short-term memory, i.e., sentence repetition and nonword repetition are the best candidate psycholinguistic markers for SLI. Such repetition tasks measure phonological short-term memory with deficits associated with limitations in the capacity of phonological storing (Gathercole et al. 1990).
*A useful criterion for a marker is that it should be largely independent of IQ as one is interested in identifying language deficits and not learning abilities (Bishop et al. 1996) → the study showed that none of the markers correlated highly with Performance IQ. The highest correlation obtained was .30 for the PTT.
*Nonword repetition deficits were identifiable even in children with a history of SLI whose language difficulties have resolved at the time of the assessment.
*The study showed significant correlation between repetition tasks. The two tasks share some underlying mechanism involving limitations in short term memory.
*Overall advantage of sentence repetition over nonword repetition in terms of sensitivity, specificity and accuracy as marker for SLI.
*Sentence repetition tasks involve more prior language knowledge and not only single word processing skills.
Limitations
It is possible that at the developmental stage of 11 years old sentence repetition has an advantage over nonword repetition tasks in terms of marking SLI.Perhaps, different short term memory tasks mark language impairments more accurately at different developmental stages.
Tense markers were better at identifying children who currently had severe language impairments and less successful in identifying children with a documented SLI who seemed to be resolved at age 11 (as opposed to repetition tasks which identified children with resolved difficulties.Further research should address the impact of repetition deficits.
Kamhi and Catts(1986) claim that in terms of nonword repetition, children with severe reading difficulties are also impaired. The fact that nonword repetition deficits appear to be a characteristic of literacy difficulties as well as SLI may point to a common limitation of phonological short term memory in these children (Snowling et al. 2000). On the other hand, children with more global learning difficulties (e.g., Down syndrome, Jarrold et al. 2000) suggest that difficulties with nonword repetition is more closely related to any language impairment and that the specific nature of SLI still remains to be understood fully.