PARENTAL LITERACY PREDICTS CHILDREN’S LITERACY: A LONGITUDINAL FAMILY-RISK STUDY

This family-risk (FR) study examined whether the literacy skills of parents with dyslexia are predictive of the literacy skills of their offspring. We report data from 31 child-parent dyads where both had dyslexia (FR-D) and 68 dyads where the child did not have dyslexia (FR-ND). Findings supported the differences in liability of FR children with and without dyslexia: the parents of the FR-D children had more severe difficulties in pseudoword-reading and spelling accuracy, in rapid word recognition, and in text-reading fluency than the parents of the FR-ND children. Finally, parental skills were found to be significant predictors of children’s Grade 3 reading and spelling. Parental skills predicted children’s reading and spelling accuracy even after controlling for children’s preschool skills. Our findings suggest that the literacy skills of a parent with dyslexia might be valuable in assessing early on their child’s liability to dyslexia.

consequently, replication is warranted. The two studies involving Dutch samples suggested that parental skills can predict children's liability for dyslexia in FR samples: the affected parents of children with dyslexia showed poorer performance than the affected parents of children without dyslexia in word-and pseudoword reading (van Bergen et al., 2011) and in word-reading and rapid automatized naming (RAN; van Bergen et al., in press). The present study compares a Finnish sample of FR children with and without dyslexia and extends previous studies by including a prediction of children's continuous reading and spelling skills from parental skills.
Dyslexia is a heterogeneous disorder at the levels of both phenotype (see e.g. Leinonen, Müller, Leppänen, Aro, Ahonen, & Lyytinen, 2001) and genotype (see e.g. Raskind et al., 2000) and our purpose therefore is to include a more fine-grained analysis rather than simple prediction of whether a child has dyslexia or not. Assessment of parental skills can provide important predictive information on their children's skills if children tend to develop similar kinds of difficulties.
Finally, we also examine the influences of environmental and motivational factors on reading and spelling development. It is of interest to examine whether the home environment factor or children's interest in reading would differentiate the FR children with and without dyslexia. Previously, it has been shown that variation in the learning environments at home, shared reading in particular, (e.g., Senechal & LeFevre, 2002;Torppa et al., 2007a), and in children's motivation (e.g., Lepola, Poskiparta, Laakkonen, & Niemi, 2005;Onatsu-Arvilommi, Nurmi, Aunola 2002;Wigfield;Eccles, & Rodrigues, 1998) are linked to language and literacy skills.
In the present study we aim to answer the following questions: 1) Do children's skills, parental skills, children's interest in reading, or home literacy environment, discriminate between the groups of FR children with and without dyslexia?
2) Do parental skills predict FR children's Grade 3 reading and spelling skills?
3) Do parental skills predict FR children's Grade 3 reading and spelling skills after accounting for children's skills prior to school entry?

Participants
Participants were 99 families, each containing a parent with dyslexia (FR group).
Children were followed from birth as part of the Jyväskylä Longitudinal Study of Dyslexia (JLD, e.g. Lyytinen et al., 2008). Parents with dyslexia reported a history of difficulties in learning to read. Their dyslexia status was confirmed through extensive individual assessment comprising reading, spelling, and phonological and orthographic processing (see Leinonen, et al., 2001). In addition, all FR-group parents reported a close family member(s) with reading and spelling problems.

Parental assessment
The FR-group parents with dyslexia were assessed on tests of reading and spelling before the child's birth. When the children were between 3 and 6 years old, we invited the parents for reassessments to measure phoneme deletion, RAN, verbal short-term memory, block design, and vocabulary. Because we were not able to reassess all parents with dyslexia, the sample size for these measures is somewhat lower than for the reading and spelling measures (i.e., n = 79 vs. n = 99).
Text-reading accuracy and fluency. Parents were asked to read aloud two passages (218 and 128 words, respectively) as fluently and accurately as possible. A measure of reading accuracy was calculated from the total number of reading errors occurring across both texts, with reading fluency as the average reading time for the two texts (s). Cronbach's alpha for reading fluency was .97.
Pseudoword-reading accuracy. Parents were asked to read a list of 30 pseudowords comprising two to four syllables. There was no time pressure. The number of correct responses was used as the measure of pseudoword-reading accuracy.
Rapid (pseudo)word recognition. The task included 40 words and 12 pseudowords as target stimuli, which were presented for 80 ms on a monitor. Immediately after each stimulus, a backward-masking stimulus was presented. Participants were asked to read aloud both the target and the masking stimulus. The next stimulus was presented after a response or after a fivesecond period of silence -whichever occurred first. Each masking stimulus had as many letters (4, 6 or 8) and syllables (2 to 4) as the target stimulus. Half of the target-masking pairs consisted of words and the other half consisted of pseudowords. The number of correctly read target words/pseudowords was the measure of rapid word recognition, making the maximum score 52.
Spelling Accuracy. Parents were asked to spell to dictation 10 pseudowords (delivered via headphones) containing 7 to 11 letters and of 2 to 4 syllables. There was no time limit to complete the task and, if necessary, participants could listen to each stimulus twice. The amount of spelling errors in pseudowords was derived as the measure.
Phoneme deletion. Parents were asked to pronounce a given word without the second phoneme. The task included 16 words (e.g., kaupunki 'city' became kupunki) of 4 to 10 letters with 2 to 4 syllables. Deletion of the second phoneme yielded a pseudoword. Stimuli were presented via headphones. A new stimulus was presented after a response or after a 20secondperiod of silence. The number of correct responses was calculated.
Rapid Naming (RAN). On each of three tasks, participants were asked to name, as rapidly as possible, a matrix of 50 items comprising objects, digits, or a mixture of digits, objects, and letters. In the parental assessment, stimuli were presented on a computer screen. Total naming time (in seconds) was used as the score. A mean composite score of the three standardized RAN scores was calculated for the prediction analyses. Cronbach's alpha for the RAN composite was .86.
Verbal short-term memory. The Digit Span subtest of WAIS-III (Wechsler, 1991) was administered with the requirement to repeat strings of digits, increasing in length, in both the forward and reverse directions.. Two sets of items, one for forward, the other for reverse, were used. Scaled scores were derived from the manual.
Vocabulary. In the Vocabulary subtest of WAIS-III (Wechsler, 1991), participants were required to define 35 words in his/her own words. Scaled scores were derived from the manual.
Block Design. This subtest of WAIS-III (Wechsler, 1991) included nine items which involved putting sets of blocks together to match patterns displayed in a booklet. Scaled scores were derived from the manual.

Home literacy environment and children's reading interest
When the children were 4, 5, and 6 years old, parents were asked to complete the Reading Models Questionnaire, which included questions concerning various features of the home literacy environment and children's interests. See a more detailed description in Torppa et al. (2007a).

Shared reading.
To produce a composite score of shared reading, we obtained parental report of both frequency and time spent on children's reading activities in the home. Two items assessed the frequency: How often 1) mother reads with the child, and 2) father reads with the child. Two items covered the amount of time spent with print materials: 1) the typical duration of a reading episode (i.e., the child reads with an adult), 2) the total time per day the child spends reading a book with an adult. Shared reading composites were derived by calculating the mean of these four item scores.
Access to written language. The composite measure included six questions: three on library visits and one on membership of a book club, one on subscription to children's magazines, and one on the number of books at home.
Children's reading interest. Parents estimated their child's interest in various activities. Altogether, 11 different activities were listed. Of these, two concerned reading: 1) picture books and children's magazines, and 2) listening to storytelling. To estimate how interested a child was in reading compared to reported interest in general, we created an 'interest-preference score' by dividing the mean of the items concerning reading activities by the mean of the items concerning all listed activities.

Children's skill assessment
Children's letter knowledge was assessed at age 5 years, phonological awareness and vocabulary at age 5.5 years, and RAN objects and digit span at age 6.5 years. Reading and spelling were assessed in Grade 3 spring (mean age: 9.7 years). All assessments were conducted individually by trained testers.
Letter knowledge. Children named 23 capital letters presented one at a time. Testing always began by presenting the first letter of the child's first name. Subsequent letters were presented in the order in which letters are typically taught in Finnish schools. The number of correctly named letters was the letter knowledge measure.
Expressive vocabulary. The Boston Naming Test (Kaplan, Goodglass, & Weintraub, 1983) contains 60 picture items which the child is asked to name. Testing is continued until six consecutive errors are incurred. The score is the number of items that are spontaneously correct plus the number of items correctly identified following provision of a semantic stimulus cue.
Verbal short-term memory. Memory was assessed with forward and backward digit span (see Gathercole & Adams, 1994). The score was the number of correct repetitions.
Phonological awareness. Four measures of phonological awareness were assessed. The mean of the standardized scores formed the composite score. Cronbach's alpha for the composite score was .76. (1) Segment identification. The child was presented on a computer screen three pictures of objects with simultaneous pronunciation of the object's name The child was asked to identify the picture containing a specified sub-word level segment. The size of the segment to be identified varied from one to four phonemes (1-2 syllables). Segments came from the beginning, middle, or end of the word. There were 20 items.
(2) Blending. The task was to blend segments into a word (e.g., per-ho-nen (butterfly)). The segments consisted of words, syllables, or phonemes. The segments were separated by 750 ms and were presented via headphones. One item consisted of a compound word, eight items required synthesis of syllables, and three items required the synthesis of syllables and phonemes. There were 16 items, and one point was awarded for each correct answer. (3) Initial phoneme identification. The child was shown four pictures of objects. Simultaneously, the object names were presented. Subsequently, a phoneme was given and the task was to select the picture of the object that starts with the phoneme. For examle: "Which word begins with /m/?". There were nine items, and one point was awarded for each correct answer. (4) Initial phoneme production. The experimenter showed a picture to the child and asked what she/he saw in the picture. Next, the child was asked to listen to the tester pronouncing the word and to articulate the initial sound. The number of correct initial sounds or letter answers out of eight items formed the score. For details of the phonological tasks, see Puolakanaho, Poikkeus, Ahonen, Tolvanen, & Lyytinen (2003).

RAN of objects.
Children were asked to name as rapidly as possible, a matrix of 50 items made up from five different pictures of objects: a car, a house, a fish, a pencil and a ball. All the Finnish names for these objects are two-syllabic high-frequency words. Total naming time (in seconds) was used as the score.
Pseudoword/word reading. Children read 20 (10 x three syllable, 10 x four syllable) pseudowords and 20 four syllable words presented successively on a computer screen. The number of correctly read words and pseudowords was used as a measure of pseudoword/word word -reading accuracy. In addition, the time from the appearance of the pseudoword/word to the end of articulation (i.e. reaction time + response duration) was used as a measure of pseudoword/word-reading fluency.
Text reading. Children read a short passage (189 words) presented on a sheet of paper.
The text is part of a nationally standardized assessment battery (Lindeman, 2000). Children were instructed to read as quickly and accurately as possible. The percentage of correctly read words was used as a measure of text-reading accuracy. Furthermore, the average number of words read correctly per minute was used as a measure of text-reading fluency.
Reading pseudoword text. In Grade 3, children read aloud a short passage comprising 38 pseudowords. The pseudoword text was generated from a story text by replacing 0-3 letters inside syllables so that the words and structure of the sentences resembled real Finnish in form but had no meaning. The percentage of correctly read pseudowords was used as a measure of pseudoword-text accuracy. The amount of pseudowords read in one minute was the measure of pseudoword-text reading fluency.
Spelling accuracy. Spelling to dictation was assessed for 10 words and 24 pseudowords.
All 34 items contained four syllables and were presented separately via headphones. The number of correctly written items was used as a measure of spelling accuracy. Cronbach's alpha was .71.
Word-list reading fluency. A nationally standardized reading test 'Lukilasse ' (Häyrinen, Serenius-Sirve, & Korman, 1999) was used as the fourth measure of reading fluency. The child was asked to read aloud a list of words of increasing difficulty, printed on a sheet of paper. The number of correctly read words within two minutes was transformed into a standard score according to the manual.

Results
The distributions of all measures were examined prior to analyses. Children's accuracy of text and pseudoword reading showed somewhat skewed distributions. There was, however, variation at the lower end of the distributions that we consider informative to the identification of dyslexia and consequently, the measures were included into analyses. A few outliers were identified in children's RAN-scores and also in parents' spelling, reading fluency, and reading accuracy. These outliers were moved to the tails of the distributions.  6.16, p < .001, η 2 =.26) indicated group differences. As expected, the measures used for diagnosis differed between the FR-D and FR-ND groups. With respect to preschool measures, the FR-D performed poorer than FR-ND group in phonological awareness, RAN, verbal short-term memory, and letter knowledge. Their vocabulary, however, did not differ significantly. Table 2 reports the means and standard deviations and the ANOVA group comparisons of the parental skills. The MANOVA group comparisons of parental reading skills (F(5, 93) = 2.61, p < .05, η 2 = .12) but not of cognitive skills (F(7, 68) = 1.70, p > .05, η 2 =.15) indicated group differences The affected parents of the FR-D group scored significantly lower than the parents of the FR-ND in pseudoword reading and spelling accuracy, in rapid word recognition, text-reading fluency, and vocabulary. There were no significant differences between the FR-D and FR-ND groups on parental text-reading accuracy, phoneme deletion, RAN, verbal short-term memory, or block design.

Prediction of Children's Skills
Next, we examined the associations between children's skills and parental skills in order to examine whether we could use the skill profile of the affected parent to predict that of the child.
Note that for the correlations and regression analyses all scores were transformed so that higher scores correspond to better performance than lower scores. Table 3 reports the correlations between the measures. Multiple stepwise regression analysis was used to predict children's skills from parental skills. Because text-and pseudowordreading accuracy were slightly skewed, these were log-transformed prior to regression analyses. Table 4 reports the regression analyses for the prediction of children's reading and spelling accuracy and Table 5 reports the results for reading fluency. All parental measures were entered using the stepwise method, which adds significant predictors one by one, starting with the strongest predictor.
The results show that between 10 and 25% of the variance in children's reading and spelling accuracy, and between 8 and 16% of the variance in children's reading fluency can be predicted by the skills of their parent with dyslexia. The strongest predictor of children's accuracy in text-reading, pseudoword reading, and spelling was parental spelling. The strongest predictor of children's pseudoword text accuracy was, however, parental pseudoword reading accuracy. The parental predictors of children's Grade 3 reading fluency (see Table 5), varied across measures. The strongest predictor of text-reading fluency was parental digit span, the strongest predictor of pseudoword text fluency was parental RAN, and the strongest predictors of list reading fluency were phoneme deletion and rapid word recognition. Pseudoword reading fluency was not significantly predicted by any of the parental measures.
Next, we examined whether parental skills predict children's reading and spelling, even after accounting for children's preschool skills. Regression analyses using the stepwise method were used with children's skills entered in Step 1 and parental skills in Step 2. The results for the accuracy and fluency measures are presented in Tables 6 and 7, respectively. Interestingly, the parental measures were significant predictors of the children's accuracy scores in text, pseudoword text, and spelling, but not in pseudoword-reading accuracy or any of the fluency measures, after controlling for the children's preschool skills. Letter knowledge at age 5 was the strongest predictor of all Grade 3 skills except pseudoword text-reading fluency, which was predicted only by children's RAN. After accounting for children's preschool skills, parental spelling predicted children's Grade 3 spelling and text-reading accuracy and parental pseudoword-reading accuracy predicted children pseudoword text-reading accuracy.

Discussion
Although having a parent with dyslexia is a strong risk factor for dyslexia (e.g. Elbro et al., 1998;Pennington & Lefly, 2001;Puolakanaho et al., 2007;Snowling, et al., 2003;van Bergen et al., 2011, in press), not all children with a family risk (FR) develop dyslexia. The present study examined the possibility of using measures obtained from the parent with dyslexia to predict their children's skills. Group comparisons of children with and without dyslexia (diagnosed in Grade 3) showed firstly, that the FR children with dyslexia were impaired across all language and literacy skills prior to school entry. Secondly, FR-D group parents showed significantly poorer reading and spelling than FR-ND group parents. Third, the regression analyses showed that children's reading and spelling skills can be predicted by parental skills. Reading and spelling accuracy were predicted significantly by parental reading and spelling accuracy measures, even after accounting for children's language and literacy skills prior to school entry.
Finally, there were no differences between the groups with and without dyslexia in gender of the child or the parent with dyslexia, parental education, home literacy environment, or in children's reading interest.
Our findings thus showed that the cognitive, reading, and spelling difficulties of the FR-D children were linked to their dyslexia status: the FR-D children were impaired across all literacy-related skills. Group differences in Grade 3 reading and spelling were expected by definition, and group differences in reading precursors (letter knowledge, phonological awareness, RAN, and verbal short-term memory) were consistent with previous findings of these measures being good predictors of reading development (e.g. Byrne, 1998;de Jong & van der Leij, 1999;Pennington & Lefly, 2001;Puolakanaho et al., 2008;Snowling et al., 2003;Landerl & Wimmer, 2008). The FR-D and FR-ND groups did not, however, differ in vocabulary at age 5.5 years of age. This finding concurs with our previous findings in the same sample concerning the very early language development of children with and without dyslexia in Grade 2 (AUTHORS), but differs from the findings of Snowling et al. (2003) who showed that the FR children without reading disabilities had better oral language skills than the FR children with reading disabilities . They suggested that good oral language may have operated as a compensatory mechanism. The definition of reading disabilities in Snowling et al. (2003) was, however, based on a composite measure that also included comprehension, which may explain why there was a significant difference in vocabulary.
The comparison of reading and spelling skills of the affected parents of the FR children with and without dyslexia showed that the parents of the FR-D children were significantly poorer than those of the FR-ND children in pseudoword-reading accuracy, rapid word recognition, spelling accuracy, and text reading fluency. These parental differences indicate that the FR-D children have a higher liability for dyslexia than the FR-ND children, which is very interesting given that all of these parents have dyslexia. The FR-D and FR-ND group parents did not, however, differ to a significant degree in text reading accuracy, phoneme deletion, RAN, verbal short-term memory, vocabulary, or block design. Two previous studies have also found parental skill differences between FR-D and FR-ND groups: van Bergen et al. Our second and third questions concerned the prediction of children's reading and spelling from parental skills. The regression analyses showed that children's Grade 3 reading and spelling skills were significantly predicted by parental skills. Children's accuracy measures were predicted by parental accuracy measures, even after accounting for children's language and literacy skills prior to school entry. Children's fluency measures, on the other hand, were predicted by parent's cognitive skills and rapid word recognition. However, parental measures did not explain additional variance in children's reading fluency after accounting for children's preschool language and (pre)literacy skills. It is worth noting that knowing how well an expectant parent with dyslexia can spell ten pseudowords is predictive of their child's reading and spelling accuracy ten years later, even after controlling for children's phonological awareness, RAN, letter knowledge, verbal short-term memory, and vocabulary at age 5 to 6.5 (0-2 years prior to school entry). It should also be noted that these associations between the skills of the parents with dyslexia and their children are not likely to be due to differences in children's home environment or reading interest because the children with and without dyslexia did not differ on these variables. Overall, our findings suggest that the use of parental skills in families with a history of dyslexia may be of assistance when evaluating whether and which literacy difficulties their children may face. Early anticipation of children's liability for literacy difficulties may provide important opportunities to offer the children at highest risk preventive support programs to minimize the challenges during early school age (e.g. Lyytinen, Erskine, Kujala, Ojanen, & Richardson, 2009;Saine, Lerkkanen, Ahonen, Tolvanen, & Lyytinen, 2011).    (1,70) = 12.86*** (1,74) = 8.09** (1,72) = 8.55** (2,73) = 11.72*** *p < .05, **p < .01, ***p < .001. (1,74) = 7.58** (1,72) = 6.46* (2,73) = 6.86** Note. The variance of children's pseudoword-reading fluency variance was not explained significantly by any of the parental measures. *p < .05, **p < .01, ***p < .001. (3,64) = 9.68*** (1,70) = 32.11*** (2,76) = 14.78*** (2,69) = 12.19 *** *p < .05, **p < .01, ***p < .001. (1,69) = 17.89 *** (1,73) = 16.88 *** (1,68) = 11.67** (1,70) = 33.60*** *p < .05, **p < .01, ***p < .001.