Free Recall of Word Lists Differing in Phonological Similarity, Word Frequency, and Phonotactic Pattern Frequency in SLI
Total Page:16
File Type:pdf, Size:1020Kb
Free recall of word lists differing in phonological similarity, word frequency, and phonotactic pattern frequency in SLI Jeffry A. Coady Julia L. Evans Elina Mainela-Arnold Keith R. Kluender Boston University San Diego State University Penn State University University of Wisconsin Abstract Memory Deficits in SLI Method Results Phonological encoding refers Phonological Similarity to the reciprocal processes in Children with SLI have difficulty with verbal recall. Participants 1. Group. Children with SLI recalled 100 which phonological strings are SLI TYP fewer words overall. broken down into smaller In recall tasks, children with SLI recall fewer items than their typically developing peers. 16 monolingual English-speaking children with SLI and 16 age-matched control children 80 t units for subsequent tasks like 1 c e r 60 Some theories view this as an underlying cause of SLI , while others view memory participated. All children had highly intelligible articulation and nonverbal IQ scores r o 2. List Length. Accuracy decreased as C recall or repetition, or where t n 2 e c r 40 deficits as resulting from a more pervasive linguistic deficit . greater than 85. The children with SLI included 2 with E-SLI and 14 with ER-SLI, 9 e smaller units are combined P list length increased. This effect was into larger phonological 1Gathercole & Baddeley, 1990; 2van der Lely & Howard, 1993; MacDonald & Christiansen, 2002 females and 7 males aged 8;7 to 11;8. The age-matched control children included 8 20 the same for both groups. strings for articulation. It can be measured by having females and 8 males aged 8;5 to 12;3. 0 dis sim dis sim dis sim dis sim dis sim dis sim 3. Phonological Similarity. All children listeners recall lists of words Children with SLI may use less efficient rehearsal strategies. Length 2 Length 3 Length 4 Length 5 Length 6 Overall differing in phonological recalled fewer phonologically similar Group Age CELF CELF RLS PPVT-III EVT NWR CLPT Word Frequency similarity among list items. In Baddeley & Hitch’s working memory model, auditory words are assumed to be stored 100 words. A significant group × similarity Participants typically recall ELS SLI 10;2 72.4* 68.7 91.3* 81.9* 78.9* 36.3* SLI in the phonological loop, subject to decay unless they are actively maintained, such as TYP fewer phonologically similar 80 interaction revealed that children with words, presumably because (1;0) (11.0) (13.2) (10.3) (6.7) (8.2) (13.7) t through rehearsal. Recall is better for items at the beginning and end of a list—primacy c e 60 r SLI showed a smaller phonological r o traces of similar words C t n e and recency effects, respectively. Primacy effects are attributed to rehearsal of initial c interfere with one another in r 40 similarity effect than controls. TYP 10;2 110.7 --- 107.1 98.3 87.5 55.5 e the phonological store. To items, while recency effects are attributed to output of the storage buffer at time of recall. P (1;2) (12.6) (8.5) (11.9) (5.9) (14.4) 20 examine group differences in Children with SLI exhibit reduced primacy effects, suggesting poorer rehearsal. 4. Word Frequency. All children efficiency of phonological 0 hi lo hi lo hi lo hi lo hi lo hi lo recalled more frequently-occurring encoding, 16 children with SLI Baddeley & Hitch, 1974; Kirchner & Klatzky, 1985; Mainela-Arnold & Evans, 2005 Length 2 Length 3 Length 4 Length 5 Length 6 Overall (mean age 10;2) and age- Stimulus Materials words. This effect was the same for matched controls recalled lists Phonotactic Probability both groups. of CVC words, two to six Reduced capacity or linguistic impairment? Eighty lists of CVC words ranging from two to six items in length were created. Two 100 SLI items in length, differing in TYP lists at each of five lengths varied along three orthogonal dimensions: phonological 80 5. Phonotactic Probability. All children phonological similarity, Another way to examine rehearsal is by having listeners recall lists of phonologically t c similarity, word frequency, and phonotactic probability. e 60 r without replacement. Word r recalled more words with common o C similar vs. dissimilar words. During rehearsal, traces of similar items interfere with one t n lists also differed orthogonally e c r 40 e P phonotactic patterns (p=.07). A along another two dimensions another so they are recalled less well. Children who use less efficient encoding The 1099 English CVC words rated high in familiarity were separated into high and low affecting recall by typical strategies are less influenced by phonological similarity among items to be recalled. word frequency groups by a median split. They were then separated into high and low 20 significant pp × wf interaction adults—word frequency and 0 revealed that this effect was phonotactic probability. Baddeley & Hitch, 1974; Liberman, Shankweiler, Liberman, Fowler & Fischer, 1977 phonotactic probability groups by another median split. Thus, there were four groups of hi lo hi lo hi lo hi lo hi lo hi lo Results reveal that all children words: (1) high word frequency, high phonotactic probability (hi wf, hi pp); (2) hi wf, lo Length 2 Length 3 Length 4 Length 5 Length 6 Overall significant for frequent words, but not recalled fewer words from Serial Position Effects for less frequent words. This was the phonologically similar lists. Children with SLI use less efficient phonological encoding strategies, pp; (3) lo wf, hi pp; and (4) lo wf, hi pp. Words from each of the four groups were 100 same for both groups. While children with SLI combined into 20 lists, four at each length (2 - 6 items). Two lists at each length SLI at least when their memory resources are exceeded. TYP recalled fewer words overall, contained phonologically similar words; two contained phonologically dissimilar words. 80 they were less affected by y 60 6. Serial Position Effects. In longer lists, c a r Children with SLI show the expected phonological similarity effects for shorter word lists, u phonological similarity, Each word appeared in only a single list. c c A all children recalled list initial and list suggesting less efficient but not for longer lists. However, these studies recombined a single set of words, so 40 phonological encoding. All The 320 words were produced by a female hi wf, hi pp hi wf, lo pp 20 final items at higher rates. This effect children recalled more high children heard each word multiple times. Practice effects likely influenced recall. Length 2 Length 3 Length 4 Length 5 Length 6 speaker from the local dialect, with no duration 0 was the same for both groups. frequency words and more Gathercole & Baddeley, 1990; James, van Steenbrugge & Chiveralls, 1994; c.f., Montgomery, 1995; van der Lely & dis sim dis sim 1 2 1 2 3 1 2 3 4 1 2 3 4 5 1 2 3 4 5 6 words with frequently List Position Howard, 1993 differences due to wf or pp. Words were occurring sound patterns. excised, normalized for volume, and give rice mouth peak However, the two groups did Conclusions not differ in their sensitivity to Language knowledge also influences recall. concatenated into lists with a 1-s ISI. Time worse light chain tip these measures. Both groups between lists was 5s for two-item lists, and yell wine loop cheap 1. Children with SLI recalled fewer items overall, but showed the same also showed similar serial In the process of redintegration, decaying, partial phonological traces are reconstructed position curves. These increased with list length. sensitivity as children with NLD to word frequency, phonotactic probability, and from established lexical and/or sublexical knowledge1. Adults use both of these, word ride rain wish pitch results suggest that children serial position. This suggests that they can use language knowledge for with SLI have reduced frequency and phonotactic probability, to facilitate recall2. phone line dog keep capacity and less efficient Procedure redintegration, to facilitate recall. phonological encoding 1Thorn, Gathercole & Frankish, 2005; 2Roodenrys, Hulme, Lethbridge, Hinton & Nimmo, 2002 make wait five chip strategies, but remarkably Children were tested individually in a quiet 2. However, because they are less influenced by phonological similarity among lo wf, hi pp lo wf, lo pp similar memory processes. room. They were told that they’d be hearing items to be recalled, their phonological encoding processes are less efficient, at [Research supported by Research Questions NIDCD DC-05263, DC-04072, list of words, and that their job was to repeat dis sim dis sim least when memory resources are exceeded. and DC-005650.] them back in any order. In a blocked design, 1. Do children with SLI encode phonological information less efficiently? Are they less deed pun zoom jug influenced by phonological similarity among items to be remembered? two-item lists were presented first, with list References hip tin peach chick Baddeley, A.D. & Hitch, G.J. (1974). Working memory. In G. Bower (Ed.), Recent Advances in Learning and Motivation, Vol. VIII. NY: Academic Pr. length increasing, concluding with six-item Gathercole, S.E. & Baddeley, A.D. (1990). Phonological memory deficits in language-disordered children: Is there a causal connection? JML, 29, 336-60. 2. Do children with SLI show the same redintegration processes as children developing James, D., van Steenbrugge, W. & Chiveralls, K. (1994). Underlying deficits in langauge-disordered children with central auditory processing difficulties. App. Psycholing, wool keen knack jerk 15, 311-28. lists. Children’s responses were recorded for Kirchner, D.M. & Klatzky, R.L. (1985). Verbal rehearsal and memory in language-disordered children. JSHR, 28, 556-565.