Children with word finding difficulties: Continuities and profiles of abilities

Abstract

Word finding difficulties (WFDs) occur in more than a quarter of children who are receiving speech and language therapy. This study provides the first investigation of the continuity in WFDs and investigates whether WFDs are associated with phonological or semantically related abilities. Thirty-eight children with WFDs were seen at age 7;0 and at 9;8. Standardized assessments of word finding, language and literacy were administered. The children’s WFDs and other language abilities showed high levels of stability. Despite their WFDs, many children had standardized scores of phonological awareness, decoding and spelling in the typical range. In contrast, semantic fluency was particularly impaired, and WFDs were a significant predictor of reading comprehension. Cluster analysis indicated that there was one group of children with a profile similar to ‘poor comprehenders’ and a second with depressed language scores indicative of specific language impairment. The relevance of these findings to understanding the causes of WFDs is discussed.

Keywords

Continuity language impairment word finding

An important theoretical and practical question about children who have language difficulties is whether the profile of their abilities changes over time. Stability cannot be assumed. Young children show variability in their language trajectories with some ‘growing out’ of their language difficulties and in others their language continues to be a cause for concern (Bishop & Adams, 1990; Rice, Taylor, & Zubrick, 2008). The nature of the child’s language difficulty may also alter. Conti-Ramsden and Botting (1999) found that nearly half of 200 seven-year-olds changed their clinical profile over a 12-month period.

Given the heterogeneous nature of language difficulties, the value of focusing on a specific aspect of the language system is becoming recognized (Hulme & Snowling, 2009). One aspect of the language system that has been the concern of a number of investigators is word finding difficulties (WFDs). Relative to chronological age peers, children with WFDs retrieve a lower proportion of words from their receptive vocabularies. Thus, their failures to produce words are the result of impaired access to lexical entries in their receptive vocabularies, rather than the absence of entries. The children also are slower and more inaccurate at picture naming (see Messer & Dockrell, 2006 for a review). WFDs have been reported to be present in about a quarter of children who are receiving language therapy (Dockrell, Messer, George, & Wilson, 1998). Their difficulties with the lexicon are likely to impact on progress in educational contexts (Cunningham & Stanovich, 1993, 1997).

As children’s vocabulary size increases their access to the lexical items becomes more efficient (Beck, Perfetti, & McKeown, 1982). On this basis it could be expected that with the development of semantic and phonological representations in the lexicon, WFDs will resolve. However, it is also possible that the children’s difficulties persist. For example, inadequate information in children’s semantic or phonological lexicon may continue to make words vulnerable to retrieval errors (McGregor, Newman, Reilly, & Capone, 2002) or there may be continuing difficulties with the retrieval process itself (Leonard, 1998).

To investigate whether there is continuity in WFDs, children with this difficulty were seen at age 7;0 years and again at age 9;8; data were obtained about WFDs and other aspects of their language abilities. It has been argued that vocabulary abilities become more clearly differentiated from other aspects of language in middle childhood (Tomblin & Zhang, 2006), and show continuity into early adulthood (Beitchman et al., 2008). Therefore, it was anticipated that assessment at this point in development would, primarily, capture lexical processes.

Data were also collected on other aspects of the children’s language abilities to examine continuities in other parts of the language system. The assessments included phonological awareness, word fluency and grammar. The assessments of phonological awareness were selected so that the tasks would minimize the effects of WFDs by requiring only the repetition of words that had been spoken by the experimenter to the child. Fluency, where children have to generate as many words related to the target category as possible, was assessed because poor performance levels are thought to reflect a less sophisticated network of connections between elements in the lexicon (Sieger-Gardner & Schwartz, 2008), and it was expected that children would have impairments on fluency tasks as these involve lexical retrieval. Furthermore, by using the same standardized test it was possible to compare phonological and semantic fluency, two areas of particular theoretical interest where normative data were available. The children’s understanding of grammar was also assessed to provide a more complete picture of the children’s language abilities. Analyses were conducted to investigate whether there were age-related differences in standardized scores of these abilities.

As well as questions about continuity, there are important questions about the causes of this impairment. The accurate and efficient retrieval of vocabulary items is underpinned by both phonological (Snowling, van Wagtendonk, & Stafford, 1988; Truman & Hennessey, 2006) and semantic representations (Funnell, Hughes, & Woodcock, 2006; Hashimoto, McGregor, & Graham, 2007; Levelt, Roelofs, & Meyer, 1999). Related to this are suggestions that the processes contributing to WFDs involve either the phonological system (Best, 2005; Constable, Stackhouse, & Wells, 1997) or the semantic system (McGregor et al., 2002; Sheng & McGregor, 2010), or both (German & Newman, 2007). However, as yet there is no consensus about the endophenotype for WFDs (Messer & Dockrell, 2006). Consequently it was decided to investigate the language and literacy abilities of children with WFDs to identify whether there were impairments to phonological and/or semantic abilities.

Investigations of children’s phonological and literacy performance in relation to picture naming tasks provide a paradigm relevant to this issue (Bishop & Snowling, 2004). Many theories assume that children with dyslexia have impairments to the phonological system and that these are responsible for literacy impairments (e.g. Hulme & Snowling, 2009; Schiff & Lotem, 2011). It would also appear that children with dyslexia have poor word retrieval as shown by slower and more inaccurate picture naming (Snowling et al., 1988; Swan & Goswami, 1997) and slower rapid automatized serial naming (Norton & Wolf, 2012). Snowling et al. (1988) suggested the poor naming abilities of children with dyslexia are caused by less well specified phonological representations which are more difficult to access. Consequently, one can hypothesize that impairments to phonological abilities are responsible for WFDs and if this were the case it would be expected that children with WFDs would have low standardized scores on measures of phonological awareness, decoding and spelling. It also might be expected that children would have lower standardized scores on fluency tasks that involved phonology (words beginning with a particular sound) in comparison to semantic fluency (words belonging to a particular semantic category). To test this hypothesis about phonological abilities and literacy, information about the literacy abilities of the children with WFDs were collected at the second time point, at an age when their reading abilities should have become established.

It also is known that impaired semantic processing impacts on reading comprehension and on tasks such as picture naming (Nation, Marshall, & Snowling, 2001; Tannenbaum, Torgesen, & Wagner, 2006). On the basis of an extensive series of detailed investigations it has been shown that there are an appreciable number of children who can be considered to be ‘poor comprehenders’. These children are characterized as having poor reading comprehension, poor semantic abilities and poor picture naming abilities, but superior phonological and decoding abilities (Nation, Cocksey, Taylor, & Bishop, 2010; Nation et al., 2001). Nation and her colleagues propose that impaired semantic abilities are the cause of poor reading comprehension as well as the cause of slower and more inaccurate naming (e.g. Nation, Clarke, Marshall, & Durand, 2004). Thus, an alternative hypothesis about WFDs is that these difficulties are caused by impairments to the semantic system and that, as a result, children with WFDs will have low standardized scores on measures of semantic fluency and reading comprehension, but will have typical and significantly higher scores of phonological awareness, decoding and spelling. These two contrasting explanations of WFDS were examined by conducting regression analyses to explore the impact of WFDs on single word reading and reading comprehension.

Previous research suggests that children with WFDs have abilities similar to those of ‘poor comprehenders’ with strengths in phonological abilities and decoding (Messer, Dockrell, & Murphy, 2004). However, the previous findings were based on group means which can mask different patterns and profiles. This is a particularly important issue as different cognitive factors may underpin impairments in the children (Best, 2005; German & Newman, 2007). This potential source of individual differences in reading performances has been described in the quadrant model of Bishop and Snowling (2004). They argued that impairments involving phonological processes result in dyslexia, as evidenced by difficulties in single word decoding, while impairments to semantic processes result in poor reading comprehension. The combination of both phonological and semantic impairments was predicted to result in Specific Language Impairment (SLI). Given the previously described associations between WFDS with single word reading and reading comprehension, it is unclear how children identified in the first instance with WFDS will perform. As a consequence, cluster analysis was used to investigate whether different groups of children with WFDs, evaluated according to their literacy, reading comprehension and language abilities, could be identified, and whether the clusters would show different profiles which corresponded to those of children with different patterns of language and literacy difficulties.

Children in the current investigation were identified at the age of 7 and assessed through the use of the standardized Test of Word Finding (TWF; German, 1989), where word retrieval failures are checked with a comprehension test to determine whether or not the failure could be due to an absence of the word from the lexicon. At this point in development children’s phonological abilities are well developed, they possess large vocabularies with rich semantic representations (Funnell et al., 2006) and standardized measures of WFDs are available (German, 1989). The children were then reassessed approximately three years later, a period of sufficient length for change and development to occur.

Thus, our aims were:

To provide a description of the stability of language in a group of children identified as having WFDs using a battery of standardized language measures.

To identify the strengths and weaknesses in the phonological and semantic abilities of the children and examine the extent to which word finding abilities predict performance in single word reading and reading comprehension.

Using cluster analysis, to explore sub-groups of children with WFDs with different profiles of language and literacy abilities.

To reduce possible confounded interpretation of the results, all children were assessed on a measure of non-verbal ability and articulation.

Method

Participants

Language resources and speech and language therapists referred children who they believed had WFDs to the research team. Parental consent and school permission were obtained prior to any preliminary testing. Data about WFDs were collected from 38 children (23 boys and 15 girls) who had a mean age of 7;0 years (SD 4.6 months; range of ages 6;5–7;8) at the beginning of the study. The children were seen again at 9;8 years. Comparisons of the initial scores of children who remained (n = 38) and those who did not remain in the study at 9;8 (n = 9) identified significantly higher scores on two measures at the older age (rhyme awareness, t(46) = −2.46, p = .017, d = −.73; alliteration fluency, t(46) = −2.58, p = .013, d = −.76).

Instruments

The following three tests were used as selection criteria for the project.

Test of Word Finding (TWF) (German, 1989)

This is a standardized measure of WFDs (standardized mean of 100, SD 15). It is designed to assess the word finding skills in middle childhood (6;6–12;11). There are five naming sections which measure accuracy and speed of word finding: (1) picture naming using nouns, (2) sentence completion naming, (3) description naming, (4) picture naming using verbs and (5) picture naming using categories. If a child fails to name an item on a test, the item is represented in a multi-choice comprehension assessment at the end of the session. This assesses whether the failure is due to a word finding difficulty or the absence of the item in a child’s receptive vocabulary. To be included in the study, children had to score at least one standard deviation below the mean, indicating the presence of WFDs relative to lexical comprehension skills. The children in the initial sample had a mean pro-rated standardized score of 71.3 (range 70–81; SD 2.6).

Raven’s Coloured Progressive Matrices (Raven, 1982)

This assesses an individual’s non-verbal cognitive performance based on visual pattern and inference making skills. Children are presented with a pattern that has a piece missing and have to identify one of six figures that will correctly complete the pattern. Children who scored below the 20th percentile were excluded from the study. The children had a mean centile of 59.5 (range 25–95; SD 25.0).

Edinburgh Articulation Test (EAT) (Anthony, Boyle, Ingram, & McIsaac, 1971)

This assesses the phonological articulation of children by asking them to name pictures of common nouns eliciting various consonant articulations. No children needed to be excluded on the basis of this assessment as all these children had standardized scores above 90, this indicates that their WFDs were unlikely to be the result of articulation difficulties.

The following standardized tests were used to obtain a profile of the children’s language abilities at 7;0 and 9;8:

Phonological Assessment Battery (PhAB) (Frederickson, Frith, & Reason, 1997)

This measure was used to assess:

Phonological awareness, which was assessed in two ways. In the test of rhyme awareness, children choose two words that rhyme out of a choice of three (one irrelevant word and two that rhyme) and in a similar alliteration test the chosen words are required to have the same beginning sound.

The fluency tests involve children generating as many lexical items as possible for each of the following categories in two minutes: (1) semantic – food and animals; (2) alliteration – words beginning with m and b; and (3) rhyme – words that sound like whip and more.

The Test for Reception of Grammar (TROG) (Bishop, 1982)

This is a multi-choice test in which a child has to point to the picture related to the words or sentence that is spoken by the experimenter. The test assesses children’s ability to understand different grammatical constructions.

In addition, the following assessments of literacy abilities were administered at 9;8:

BAS II Word Reading Scale (Elliott, Smith, & McCulloch, 1997)

This scale is used to assess the recognition and oral reading of single words. The test items involve single word decoding ability using a sample of words ranging from common ones found in children’s books to those that are less common.

Woodcock Reading Mastery Test Passage Comprehension Test (WRMT) (Woodcock, 1997)

This test measures the child’s ability to read a short passage and identify a key missing word from the passage. Early items are only one sentence long and are accompanied by a line drawing that is related to the text. Later items are longer passages drawn from a variety of reading materials, such as books, newspapers and other documents.

BAS II Spelling Scale (Elliott et al., 1997)

This scale provides a number of phonetically regular and irregular words to assess the child’s ability to produce correct spellings (age range 5–17 years). Each item is first presented in isolation, then within the context of a sentence, and finally in isolation. The child has to respond by writing the word.

Procedure

At Time 1, the selection tests (TWF, EAT and Raven’s) were administered first. If a child met the selection criteria, the naming, phonological and comprehension tasks were administered at a second session. The same order of test presentation occurred at Time 2. All literacy measures were administered at the final assessment point. The three assessments of literacy at 9;8 occurred in a third and final session. Time was spent talking to the child at the beginning of each session to set them at ease. Each session lasted no longer than 40 minutes and the children were tested individually. At the end of each session the child was thanked and returned to class. The study was conducted according to British Psychological Society ethical guidelines and the study had been approved by the appropriate university ethical committee.

Results

Analyses concerning the continuity of WFDs and other language assessments are presented in the first section. The second section contains analyses of the strengths and weaknesses of the children’s language and literacy abilities. The third section contains information about different profiles of abilities in children with WFDs.

Continuity of language across age

Continuity of WFDs

At baseline, the children achieved a mean TWF standard score of 71.3, SD = 2.6, with a wide variation in the pro-rated raw scores (M = 35.3, SD = 9.1). At the follow-up, the children had a mean standard TWF score of 76.8 (SD = 11.3) and a mean pro-rated raw score of 64.6 (SD = 8.3). At the older age, seven of the children (18%) no longer had a standardized score below 85 and therefore their scores could be considered to be in the typical range; however, in only three of these cases were the standardized scores above 86. There was a significant improvement across age in the standardized scores, t(37) = −3.02, p = .005, d = .99, although the mean score was still low. It should be noted that there is the possibility that the significant improvement in TWF scores could be due to regression to the mean, and as a result there is a need for caution when interpreting the age-related improvement.

Continuity of phonological awareness, fluency and receptive grammar

To investigate whether the standardized scores of the other language assessments improved with age, paired t-tests were carried out on each of the six measures. Descriptive information is provided in Table 1. A Bonferroni correction for six comparisons gives a significant level of p < .008. The majority of measures did not show a significant difference between the two ages (rhyme awareness, t(37) = −148, p = .15; TROG, t(37) = .63, p = .53; alliteration fluency, t(37) = −.65, p = .52; semantic fluency, t(37) = −.48, p = .63); the exceptions were a significant improvement in alliteration awareness, t(37) = −2.29, p = .006, d = .75, and rhyme fluency, t(37) = 3.87, p < .001, d = 1.27.

Table 1.

Descriptive statistics about language and literacy abilities.

	Age 7;0			Age 9;8
	M	SD	Stand. score > 84	M	SD	Stand. score > 84
BAS spelling	–	–	–	99.3	19.1	71%
BAS single word reading	–	–	–	95.8	17.0	71%
Phonological awareness rhyme	88.4	12.7	71%	92.2	17.4	63%
Phonological awareness alliteration	85.6	10.5	55%	91.0**	10.4	76%
Reading comprehension	–	–	–	87.2	11.9	68%
TROG	85.5	9.4	50%	84.3	12.5	39%
Fluency rhyme	80.7	10.7	29%	87.1**	8.6	68%
Fluency alliteration	77.7	8.3	24%	78.6	7.8	21%
Fluency semantics	72.6	6.7	5%	73.1	7.3	11%

p < .01; * p < .05.

Profiles of the language and literacy abilities of children with WFDs

Strengths and weaknesses in language and literacy abilities

Table 1 indicates that the children had relative strengths on measures of spelling, single word reading and phonological awareness, with low atypical levels of performance on the fluency tasks. The mean standardized scores for phonological awareness (rhyme and alliteration) were in the typical range at both ages, and the majority of children had scores that were above 84 and therefore were within 1 SD from the mean and in the typical range (from 55% to 76% of children). In addition, at the older age the mean standardized scores for single word reading and for spelling were above 90, and 71% of the children had scores above 84.

In contrast, the mean standardized scores for the fluency assessments involved a lower level of performance. Most of the mean standardized scores were in the low atypical range, with the majority of children having scores below 85; the exception to this pattern was rhyme fluency at the older age, which had a mean score of 87. It is worth noting that the children had a particular weakness with semantic fluency, with mean scores of 73 at both ages. For the other two assessments, the mean standardized scores of the TROG and of reading comprehension were around 85 and thus were just within the average range.

Repeated measures ANOVAs were conducted on the standardized scores to identify whether there were differences at each time point in the performance on the language assessments and at 9;8 on the literacy assessments. For the six language assessments at 7;0, Mauchley’s test indicated that the assumption of sphericity had been violated, χ²(14) = 25.78, p = .03. Therefore, degrees of freedom were corrected using Greenhouse–Geisser estimates of sphericity (ϵ = .82). There was a significant effect of type of assessment, F(4.09, 151.46) = 19.32, p < .001, ηp² = 34. Bonferroni post-hoc tests identified that the standardized scores for both rhyme fluency and semantic fluency were significantly lower than most of the other language assessments. In particular, the standardized scores of the five language assessments were significantly higher than those for semantic fluency (p < .002 in all cases). In addition, the assessments with the three highest scores (rhyme awareness, alliteration awareness and the TROG) had significantly higher standardized scores than those for alliteration fluency (p < .001, except for the comparison of the TROG where p = .032). The only other significant comparison involved the higher standardized scores for rhyme awareness than rhyme fluency. Thus, the children performed poorly on the fluency assessments and particularly poorly on the assessment of semantic fluency.

For the six language assessments at 9;8, Mauchley’s test indicated that the assumption of sphericity had been violated, χ²(14) = 40.5, p < .001. Therefore, degrees of freedom were corrected using Greenhouse–Geisser estimates of sphericity (ϵ = .70). There was a significant effect of type of assessment, F(3.5, 128.98) = 22.52, p < .001, ηp² = 38. As at the younger age, Bonferroni post-hoc tests identified that the standardized scores for both rhyme fluency and semantic fluency were significantly lower than most of the other language assessments. The standardized scores for semantic fluency were significantly lower than those of all the other assessments (p < .001, except for alliteration fluency where p = .003), and the standardized scores for alliteration fluency were significantly lower than rhyme fluency, alliteration awareness and rhyme awareness (p < .001) and marginally significantly lower for the TROG (p = .054). These analyses confirmed that there were significant differences in the children’s language abilities; in particular, semantic fluency was significantly worse than all the other language abilities.

For the three literacy assessments at 9;8, Mauchley’s test indicated that the assumption of sphericity had been violated, χ²(2) = 14.23, p = .001. Therefore degrees of freedom were corrected using Greenhouse–Geisser estimates of sphericity (ϵ = .75). There was a significant effect of type of assessment, F(1.51, 55.79) = 4.10, p = .03, ηp² = .10. Bonferonni post-hoc comparisons revealed that there was a marginally significant difference between spelling and reading comprehension (p = .081), suggesting a weakness in reading comprehension.

Prediction of decoding and reading comprehension from TWF and phonological abilities at 9;8

Multiple hierarchical regression analyses were conducted to investigate whether the TWF and phonological awareness were significant predictors of single word reading or of reading comprehension. Three children with TWF standardized scores above 90 at 9;8 were excluded from the analyses. Mean standardized scores of rhyme and alliteration awareness were used in these analyses. It was anticipated that phonological awareness would be a significant predictor of single word reading and so the scores of TWF were entered at step 1 and those for phonological awareness at step 2. Only phonological awareness was a significant predictor of TWF scores in the regression analysis (for step 1 adjusted R² = .18, R² change = .20, p = .007; at step 2, adjusted R² = .06, R² change = .21, p = .14). For the prediction of reading comprehension we examined whether TWF scores would be a significant predictor after controlling for phonological skills. As such, the scores of phonological awareness were entered at step 1 and those for TWF at step 2. Only the TWF was a significant predictor of reading comprehension (for step 1 adjusted R² = .05, R² change = .08, p = .112; at step 2, adjusted R² = .21, R² change = .18, p = .01).

These findings were confirmed by inspection of the standardized coefficients. These indicated that phonological awareness was a significant predictor of single word reading, but that the TWF was not (for phonological awareness, standardized β = .35, p = .04; for TWF, standardized β = .25, p = .14). By contrast, the TWF was the significant predictor of reading comprehension (for phonological awareness, standardized β = .10, p = .55; for TWF, standardized β = .46, p = .01). Thus, phonological awareness rather than the TWF was a significant independent predictor of single word reading, but the TWF was a significant predictor of reading comprehension.

Sub-groups of children with WFDs

Cluster analysis to identify sub-groups of children with WFDs at 9;8

To identify children with different profiles of abilities, hierarchical cluster analysis was carried out using standardized scores from all the language and literacy measures at 9;8, except the TWF. The children who had TWF standardized scores over 90 were not included in the analysis. The cluster analysis involved a between-group linkage method with squared Euclidean distance.

A limit of three clusters was set for the first analysis because from the Quadrant Model we did not anticipate that any participants would fall in a typically developing cluster. These three clusters corresponded to the possibility of there being children with WFDs who had poor semantic abilities, children with WFDs who had poor decoding abilities as in dyslexia and children with WFDs who had generally poor single word reading, reading comprehension and language abilities as in SLI. The analysis placed 10 children in cluster 1, 23 children in cluster 2 and two children in cluster 3. The two children in cluster 3 had a different profile of abilities from each other, except for both having two of the highest scores of rhyme awareness in the sample. When the limit of the number of clusters was set to four there was only one child in the fourth cluster. Therefore it was decided to evaluate whether the majority of children identified in the cluster analysis for two groups corresponded to any of the profiles identified in the quadrant model (Bishop & Snowling, 2004).

The language and literacy abilities of children in clusters 1 and 2

Table 2 shows that the mean scores on a number of the assessments were higher in cluster 1 than cluster 2. In cluster 1, the mean standardized scores for spelling, decoding and rhyme awareness were above 100, and the only mean scores below 80 were for semantic fluency and the TWF. In contrast, most of the mean standardized scores in cluster 2 were depressed, with the following assessments having mean scores below 80: TROG, alliteration fluency, semantic fluency and the TWF.

Table 2.

Mean standardized scores on language and literacy assessments for children in the two clusters.

	Cluster 1 n = 10		Cluster 2 n = 25
	M	SD	M	SD
BAS spelling	103.8	10.4	86.7	13.8
BAS single word reading	101.6	9.5	86.0	12.5
Reading comprehension	92.8	7.7	84.6	9.9
Rhyme awareness	103.4	12.2	87.8	17.6
Alliteration awareness	96.6	7.3	89.3	10.8
TROG	94.8	11.4	78.2	8.4
Rhyme fluency	90.3	6.4	85.4	9.4
Alliteration fluency	83.2	9.8	76.2	6.1
Semantic fluency	77.2	11.1	71.8	5.0
Pro-rated TWF	78.7	4.8	72.8	7.3

To investigate whether the profiles in each cluster mapped onto the poor comprehender, dyslexia or SLI profile from the Bishop and Snowling (2004) model, separate repeated measures ANOVAs first were conducted on all the language and literacy standardized scores together with the TWF. For clusters 1 and 2, there was a significant effect of assessment type: cluster 1, F(9, 81) = 12.43, p < .001, ηp² = .58. For cluster 2, Mauchley’s test indicated that the assumption of sphericity had been violated χ²(44) = 102.26, p < .001, therefore degrees of freedom were corrected using Greenhouse–Geisser estimates of sphericity (ϵ = .49); cluster 2, F(4.45, 106.77) = 12.26, p < .001, ηp² = .34. As both ANOVAs showed that significant differences were present, post-hoc comparisons were then conducted within each cluster. Reading comprehension was used as the reference variable because it is important in distinguishing between the three relevant profiles in the quadrant model. In poor comprehenders, standardized scores of reading comprehension are reported to be lower than those of single word reading (Nation et al., 2004), whereas in children with SLI reading comprehension scores should be equivalent to those of phonological awareness and single word reading, and in children with dyslexia reading comprehension scores may be higher than those of single word reading and phonological awareness.

For cluster 1, the assumption of sphericity was not violated and so simple contrasts with reading comprehension as the reference variable were used to identify significant differences. The scores for reading comprehension were significantly lower than those of spelling, F(1,9) = 6.41, p = .032, ηp² = .42, and of rhyme awareness, F(1,9) = 4.59, p = .038, ηp² = .40, with the difference being marginally significant for the single word reading scores, F(1,9) = 4.52, p = .063, ηp² = .33. This profile is similar to that of poor comprehenders. In addition, the standardized scores for reading comprehension were significantly higher than for rhyme fluency, F(1,9) = 24.37, p = .001, ηp² = .73, alliteration fluency, F(1,9) = 12.57, p = .006, ηp² = .58, and semantic fluency, F(1,9) = 40.58, p < .001, ηp² = .82.

For cluster 2, the assumption of sphericity was violated. Bonferroni comparisons were used to identify significant differences between the different forms of assessment. Reading comprehension was only found to have significantly higher standardized scores than those for alliteration fluency (p = .047), semantic fluency (p = .002) and the TWF (p < .001). Consequently, there was no evidence that these children had differences between reading comprehension and decoding as seen in children with dyslexia or poor comprehension. Rather, as shown in Table 2, all the means of these standardized scores were depressed, and this is closer to the profile for children with SLI described in the quadrant model, where impairments are evident in both phonological and semantic abilities.

Discussion

Most children identified with WFDs at 7;0 also had this impairment at 9;8; the majority of children at the older age had standardized scores on the TWF below 85 and so WFDs remained a significant language impairment for these children. This contrasts with some reports of variation in children’s language difficulties across age (see Conti-Ramsden & Botting, 1999) and suggests that WFDs involve a coherent and continuing difficulty. Despite the majority of children continuing to have standardized scores below 85 on the TWF, there was a significant age-related improvement in these scores. Therefore, the findings suggest that there was persistence in the children’s WFDs over this age period, with significant, but small improvement in standardized scores. For the majority of children, WFDs were not resolved with increasing age. Consequently, this profile in the standardized scores indicates that the children had word finding abilities that were below what would be expected at the two ages, and that relative to what would be expected there was only modest improvement relative to typical children. An examination of the other language abilities of the children provided a similar picture. From measures of phonological awareness, word fluency and receptive grammar, only two of the six measures showed an increase in standardized scores (alliteration awareness and rhyme fluency). These findings together with those from the TWF suggest that there is continuity in language functioning of children with WFDs between 7 and 10 years.

According to the phonological representation hypothesis (Hulme & Snowling, 2009), difficulties on assessments that involve decoding, spelling and phonological awareness are likely to indicate inadequate phonological representations. In addition, it has been suggested that inadequate phonological representations could make it more difficult to access these representations and result in impaired naming (Snowling et al., 1988). For the current cohort of children with WFDs, there were relative strengths in phonological awareness, single word reading and spelling abilities and this suggests the children had age-appropriate phonological skills which should have supported word finding processes.

In contrast, there was a range of evidence that these children had impaired semantic abilities. As all the fluency assessments were from the same test battery, the standardized scores on the different assessments should be equivalent. However, the children had significantly lower standardized scores of semantic fluency at the older age than any other language assessment, including rhyme fluency. The significantly lower semantic fluency scores suggest that semantic representations were not easily accessible. Another indication of impaired semantic abilities was that the children had marginally significantly lower scores on the measure of reading comprehension than spelling. In addition, regression analyses indicated that the TWF was a significant predictor of reading comprehension, but not of single word decoding.

Thus, several lines of evidence from this investigation suggest that children with WFDs have greater impairments in their semantic abilities than in their phonological abilities. These findings are supported by a study carried out by Ebbels et al. (2012), who reported that a semantic intervention was effective for children with WFDs. In another intervention study, German, Schwanke, and Ravid (2012) reported that a combination of semantic and phonological interventions was more effective than only a semantic intervention in helping vocabulary acquisition in children with WFDs. Given the profiles identified in the cluster analysis, the participants in German et al. (2012) may be more similar to the SLI group, where it is to be expected that both phonology and semantics would benefit from intervention. The ways in which semantic difficulties appear to underpin performance of children with WFDs suggest word finding difficulties in other groups of children may also result from poorer semantic representations. However, there also is the possibility, particularly in children with dyslexia, that impairments to phonological representations are responsible for word finding difficulties.

The children in our sample were receiving language therapy, had scores of non-verbal intelligence in the typical range and had word finding difficulties as assessed by the TWF. Thus, it is likely that a high proportion of the children would be identified as having SLI, and as such our findings have potential relevance to this group. In particular, our findings should alert speech and language therapists to the possibility that WFDs are associated with impaired semantic abilities and, among children with WFDs, there may be those who have relative strengths in decoding and spelling, but weaknesses in reading comprehension.

The findings from the cluster analysis extended the interpretation of the results, and illustrated the importance of identifying children with different profiles of abilities. Analysis of the whole sample suggested that children with WFDs had relative strengths in spelling, single word reading and phonological awareness, but had impaired semantic abilities. The findings from the clusters suggest that this profile was most applicable to the smaller of the two clusters, where most of the children had typical levels of single word reading and spelling and phonological awareness, but had lower reading comprehension abilities. Furthermore, the mean difference in the size of these scores was around 10, and Nation et al. (2004) has used a discrepancy of this size between decoding and reading comprehension to identify poor comprehenders. Thus, the children in cluster 1 had a profile of literacy abilities similar to those of poor comprehenders.

In the other cluster, significant differences were not found between reading comprehension in comparison to literacy and phonological awareness assessments. The relatively flat profile of phonological and semantic abilities in cluster 2 corresponds to the SLI quadrant in Bishop and Snowling’s (2004) model. The identification of these two clusters raises questions for further research with a larger sample to evaluate whether these two profiles have consequences for WFDs.

In sum, the analysis of continuities in WFDs indicated that, despite some improvement, the majority of the children continued to exhibit WFDs at age 9;8. Few of the other language abilities that were assessed showed age-related improvements, indicating continuity in the children’s language abilities. Analysis of the whole sample indicated that the children had phonological and literacy abilities in the typical range, suggesting that inadequate phonological representations were unlikely to be the reason for their WFDs. Cluster analysis suggested that this profile of abilities was confined to a sub-group of children. In both clusters, as in the sample as a whole, the children had particularly poor semantic fluency abilities. This implicates semantic processing or representations as a cause of WFDs.

Footnotes

Funding

This research was partially supported by an ESRC Ph.D. studentship to Nicola Murphy.

References

Anthony

Boyle

Ingram

McIsaac

(1971). The Edinburgh Articulation Test. Hong Kong: Longman Group (FE).

Beck

I. L.

Perfetti

C. A.

McKeown

M. G.

(1982). Effects of long-term vocabulary instruction on lexical access and reading comprehension. Journal of Educational Psychology, 74, 506–521. doi:10.1037/0022-0663.74.4.506

Beitchman

J. H.

Jiang

Koyama

Johnson

C. J.

Escobar

Atkinson

. . .Vida

(2008). Models and determinants of vocabulary growth from kindergarten to adulthood. Journal of Child Psychology and Psychiatry, 49, 626–634. doi:10.1111/j.1469-7610.2008.01878.x

Best

(2005). Evaluation of a new intervention for word finding difficulties in children. International Journal of Language & Communication Disorders, 40, 279–318.

Bishop

D. V. M.

(1982). Test for reception of grammar. Oxon, UK: Thomas Leach.

Bishop

D. V. M.

Adams

(1990). A prospective study of the relationship between specific language impairment, phonological disorders and reading retardation. Journal of Child Psychology and Psychiatry, 31, 1027–1050. doi:10.1111/j.1469-7610.1990.tb00844.x

Bishop

D. V. M.

Snowling

M. J.

(2004). Developmental dyslexia and specific language impairment: Same or different? Psychological Bulletin, 130, 858–886.

Constable

Stackhouse

Wells

(1997). Developmental word-finding difficulties and phonological processing: The case of the missing handcuffs. Applied Psycholinguistics, 18, 507–536. doi:10.1017/S0142716400010961

Conti-Ramsden

Botting

(1999). Classification of children with specific language impairment: Longitudinal considerations. Journal of Speech, Language and Hearing Research, 42, 1195–1204.

10.

Cunningham

A. E.

Stanovich

K. E.

(1993). Children’s literacy environments and early word recognition subskills. Reading and Writing: An Interdisciplinary Journal, 5, 193–204. doi:10.1007/BF01027484

11.

Cunningham

A. E.

Stanovich

K. E.

(1997). Early reading acquisition and its relation to reading experience and ability 10 years later. Developmental Psychology, 33, 934–945. doi:10.1037/0012-1649.33.6.934

12.

Dockrell

J. E.

Messer

George

Wilson

(1998). Children with word-finding difficulties: Prevalence, presentation and naming problems. International Journal of Language & Communication Disorders, 33, 445–454. doi:10.1080/136828298247721

13.

Ebbels

S. H.

Nicoll

Clark

Eachus

Gallagher

Horniman

. . .Turner

(2012). Effectiveness of semantic therapy for children with word finding difficulties in pupils with a persistent language impairment: A randomised control trial. International Journal of Language & Communication Disorders, 47, 35–51.

14.

Elliott

C. D.

Smith

McCulloch

(1997). Technical manual of British Ability Scales II. Slough, UK: NFER-Nelson.

15.

Frederickson

Frith

Reason

(1997). The Phonological Assessment Battery: Standardised edition. Slough, UK: NFER-Nelson.

16.

Funnell

Hughes

Woodcock

(2006). Age of acquisition for naming and knowing: A new hypothesis. Quarterly Journal of Experimental Psychology, 59, 268–295. doi:10.1080/02724980443000674

17.

German

D. J.

(1989). A diagnostic model and a test to assess word-finding skills in children. British Journal of Disorders of Communication, 24, 21–39. doi:10.3109/13682828909011944

18.

German

D. J.

Newman

R. S.

(2007). Oral reading skills of children with oral language (word finding difficulties). Reading Psychology, 28, 397–442. doi:10.1080/02702710701568967

19.

German

D. J.

Schwanke

J. H.

Ravid

(2012). Word finding difficulties: Differentiated vocabulary instruction in the speech and language room. Communication Disorders Quarterly, 33, 146–156. doi:10.1177/1525740111405840

20.

Hashimoto

McGregor

K. K.

Graham

(2007). Conceptual organization at six and eight: Evidence from the semantic priming of object decisions. Journal of Speech, Language, and Hearing Research, 50, 161–176.

21.

Hulme

Snowling

(2009). Developmental cognitive disorders. Oxford, UK: Blackwell.

22.

Leonard

L. B.

(1998). Children with specific language impairment. Boston, MA: MIT Press.

23.

Levelt

W. J. M.

Roelofs

Meyer

A. S.

(1999). A theory of lexical access in speech production. Behavioural and Brain Sciences, 22, 1–38.

24.

McGregor

K. K.

Newman

R. M.

Reilly

Capone

N. C.

(2002). Semantic representation and naming in children with specific language impairment. Journal of Speech, Language and Hearing Research, 45, 998–1014. doi:10.1044/1092-4388(2002/081)

25.

Messer

Dockrell

J. E.

(2006). Children’s word findings difficulties: Descriptions and explanations. Journal of Speech, Hearing and Language Research, 49, 309–324. doi:10.1044/1092-4388(2006/025)

26.

Messer

Dockrell

J. E.

Murphy

(2004). The relation between naming and literacy in children with word finding difficulties. Journal of Educational Psychology, 96, 462–470. doi:10.1037/0022-0663.96.3.462

27.

Nation

Clarke

Marshall

C. M.

Durand

(2004). Hidden language impairments in children: Parallels between poor reading comprehension and specific language impairment? Journal of Speech, Language, and Hearing Research, 47, 199–211. doi:10.1044/1092-4388(2004/017)

28.

Nation

Cocksey

Taylor

J. S.

Bishop

D. V.

(2010). A longitudinal investigation of early reading and language skills in children with poor reading comprehension. Journal of Child Psychology and Psychiatry, 51, 1031–1039.

29.

Nation

Marshall

C. M.

Snowling

M. J.

(2001). Phonological and semantic contributions to children’s picture naming skill: Evidence from children with developmental reading disorders. Language and Cognitive Processes, 16, 241–259. doi:10.1080/01690960042000003

30.

Norton

E. S.

Wolf

(2012). Rapid automatized naming (RAN) and reading fluency: Implications for understanding and treatment of reading disabilities. Annual Review of Psychology, 63, 427–452.

31.

Raven

(1982). The coloured progressive matrices. Dumfries, Scotland: H. K. Lewis.

32.

Rice

M. L.

Taylor

C. L.

Zubrick

S. R.

(2008). Language outcomes of 7-year-old children with or without a history of late language emergence at 24 months. Journal of Speech Language and Hearing Research, 51, 394–407. doi:10.1044/1092-4388(2008/029)

33.

Schiff

Lotem

(2011). Effects of phonological and morphological awareness on children’s word reading development from two socioeconomic backgrounds. First Language, 31, 139–163. doi:10.1177/0142723710393098

34.

Sheng

McGregor

K. K.

(2010). Object and action naming in children with specific language impairment. Journal of Speech, Language and Hearing Research, 53, 1704–1719. doi:10.1044/1092-4388

35.

Seiger-Gardner

Schwartz

R. G.

(2008). Lexical access during word production in children with and without SLI: A cross-modal picture-word interference study. International Journal of Language and Communication Disorders, 43, 528–551.

36.

Snowling

van Wagtendonk

Stafford

(1988). Object naming deficits in developmental dyslexia. Journal of Research in Reading, 11, 67–85. doi:10.1111/j.1467-9817.1988.tb00152.x

37.

Swan

Goswami

(1997). Phonological awareness deficits in developmental dyslexia and the phonological representations hypothesis. Journal of Experimental Child Psychology, 66, 18–41. doi:10.1006/jecp.1997.2375

38.

Tannenbaum

Torgesen

Wagner

(2006). Relationships between word knowledge and reading comprehension in third-grade children. Scientific Studies of Reading, 10, 381–398. doi:10.1207/s1532799xssr1004_3

39.

Tomblin

J. B.

Zhang

(2006). The dimensionality of language ability in school-age children. Journal of Speech, Language and Hearing Research, 49, 1193–1208. doi:10.1044/1092-4388(2006/086)

40.

Truman

Hennessey

N. W.

(2006). The locus of naming difficulties in children with dyslexia: Evidence of inefficient phonological encoding. Language and Cognitive Processes, 21, 361–393.

41.

Woodcock

(1997). Woodcock Reading Mastery Tests—Revised. Circle Pines, MN: American Guidance Service.