development of phonological sensitivity in 2- to 5-year...

Journal of Educational Psychology1998, Vol. 90, No. 2, 294-311

Copyright 1998 by the American Psychological Association, Inc.0022-0663/98/$3.00

Development of Phonological Sensitivity in 2- to 5-Year-Old Children

Christopher J. Lonigan, Stephen R. Burgess, Jason L. Anthony, and Theodore A. BarkerFlorida State University

This study examined phonological sensitivity in 238 children from middle- to upper-incomefamilies and 118 children from lower-income families across different levels of linguisticcomplexity. Children ranged in age from 2 to 5 years. Overall, the results indicated that aschildren increased in age, phonological sensitivity both increased in absolute terms andbecame more stable. Significant social class differences in growth of phonological sensitivitywere also obtained. Phonological sensitivity at different levels of linguistic complexity (e.g.,syllables, phonemes) was substantially interrelated at each age and predicted word readingability in older children independently of language skills and letter knowledge. These resultsindicate that phonological sensitivity can be assessed in young preschool children and thatlower levels of phonological sensitivity may serve as developmental precursors to higherlevels of phonological sensitivity.

The development of phonological processing is an impor-tant precursor to the acquisition of early reading skills (e.g.,Adams, 1990; Wagner & Torgesen, 1987). Phonologicalprocessing refers to the use of phonological information(i.e., the sounds of one's language) in processing written andoral language. A growing body of research indicates thatindividual differences in one form of phonological process-ing, phonological sensitivity, are causally related to thenormal acquisition of beginning reading (e.g., Bryant,MacLean, Bradley, & Crossland, 1990; Wagner & Torgesen,1987; Wagner, Torgesen, & Rashotte, 1994). Children whoare better at detecting and manipulating syllables, rhymes, orphonemes are quicker to learn to read, and this relation ispresent even after variability in reading skill due to factorssuch as IQ, receptive vocabulary, memory skills, and socialclass is partialed out (Bryant et al., 1990; Wagner &Torgesen, 1987; Wagner et al., 1994).

The majority of evidence linking phonological sensitivityin prereaders with the development of reading has comefrom studies that have assessed children's phonologicalsensitivity at the point of school entry but prior to formalreading instruction. For example, Share, Jorm, MacLean,and Mathews (1984) found that children's phonologicalsensitivity, measured at the beginning of kindergarten with aphoneme segmentation task, was the single best predictor of

Christopher J. Lonigan, Stephen R. Burgess, Jason L. Anthony,and Theodore A. Barker, Department of Psychology, Florida StateUniversity.

Joseph K. Torgesen and Peter E. Bryant provided helpfulcomments on earlier versions of this article. We also wish toacknowledge the contribution of the child care centers, directors,and personnel who assisted with this project as well as the childrenand parents who made it possible. We also express our thanks toSarah Dyer, Crystal Carr, Tracy Ferguson, Kimberly Ingram,Danielle Karlau, Nikki Sutton, Emily Shock, and other students atFlorida State University for their assistance with data collection.

Correspondence concerning this article should be addressed toChristopher J. Lonigan, Department of Psychology, Florida StateUniversity, Tallahassee, Florida 32306-1270. Electronic mail maybe sent to [email protected].

reading achievement both by the end of kindergarten and bythe end of first grade. A consistently strong connectionbetween phonological sensitivity and learning to read hasbeen reported across diverse studies (e.g., Bradley & Bryant,1983, 1985; Juel, Griffith, & Gough, 1986; Lundberg,Olofsson, & Wall, 1980; Mann, 1984; Mann & Liberman,1984; Stanovich, Cunningham, & Cramer, 1984; Wagner etal., 1994). A number of experimental studies have alsodemonstrated that training children in phonological sensitiv-ity results in increases in their rate of reading acquisition(e.g., Brady, Fowler, Stone, & Winbury, 1994; Byrne &Fielding-Barnsley, 1991; Wagner, 1996). These studiesindicate that phonological sensitivity measured prior to theonset of formal reading instruction is strongly and causallyrelated to later word recognition and reading comprehension.

Stanovich (1992) noted that there is little consensusconcerning the terminology to use when referring to variousaspects of phonological ability. Although some investigatorsreserve the use of the term phonological awareness to referto the ability to manipulate words at the level of phonemes,others use the term to refer to the constellation of tasksrequiring sensitivity to speech sounds at various levels (e.g.,phonemes, onset-rime, syllables). Because explicit aware-ness of speech sounds does not appear to be required for theconnection to reading and because there is no consensus onwhat level of phonological skill is a prerequisite to acquisi-tion of alphabetic literacy, Stanovich suggested that the termphonological sensitivity be used to refer to the more globalset of processing abilities that require sensitivity to speechsounds. Phonological sensitivity can be viewed as a hierar-chy of sensitivity or as varying along levels of linguisticcomplexity (Adams, 1990; Stanovich, 1992). Higher levelsof phonological sensitivity require more explicit analyses ofsmaller sized phonological units (e.g., phonemes), and moreprimitive levels of phonological sensitivity require a moreshallow level of analysis of larger sound units (e.g., syl-lables). In this view, phoneme segmentation, phonemecounting, and phoneme reversal tasks represent the higher

294

DEVELOPMENT OF PHONOLOGICAL SENSITIVITY 295

level of sensitivity, whereas rhyming or syllable segmenta-tion tasks represent a more primitive level.

Sensitivity to phonemes is often assumed to have specialstatus in the relation between phonological sensitivity andreading both because it is at this level that graphemescorrespond to speech sounds in reading and because indi-vidual phonemes do not have separable physical reality(e.g., Liberman, Cooper, Shankweiler, & Studdert-Kennedy,1967). Although a variety of tasks have been developed toassess phonological sensitivity, those that are often usedwith school-age children (e.g., phoneme tapping and pho-neme counting) are extremely difficult for children youngerthan age 5 or 6 years (e.g., Adams, 1990; Bryant et al., 1990;Wagner et al., 1987). The finding that young childrenperform poorly on these tasks of phonological sensitivityprior to school entry led some to conclude that the highestlevels of phonological sensitivity were mainly a by-productof learning to read (e.g., Morais, 1991).

Evidence suggests that there is a developmental hierarchyof children's sensitivity to linguistic units at different levelsof complexity. Children achieve syllabic sensitivity earlierthan they achieve sensitivity to phonemes (Fox & Routh,1975; Liberman, Shankweiler, Fischer, & Carter, 1974), andchildren's sensitivity to intrasyllabic units (i.e., onset-rime)also precedes sensitivity to phonemes (Treiman, 1992).However, it is not clear if sensitivity to syllables and tointrasyllabic units represent processing abilities that aredistinct from sensitivity to phonemes or if these abilities aredevelopmental precursors to phoneme sensitivity. For in-stance, Yopp (1988) found evidence for two highly corre-lated processing factors in kindergarten children: a simplephonological sensitivity factor that included measures involv-ing the manipulation of phonemes and detection of rhymeand a factor for tasks that required holding a sound inmemory while performing an operation. In contrast, Hoien,Lundberg, Stanovich, and Bjaalid (1995) found separatefactors for phoneme sensitivity, syllabic sensitivity, andrhyme sensitivity in both Norwegian preschool children(mean age 83 months) and first-grade children (mean age 94months), and scores on all three factors independentlypredicted reading abilities for the older group of children.However, it is difficult to interpret the results of Hoien et al.in the context of developing phonological sensitivity be-cause of the age of the children, the fact that scores onseveral measures were near ceiling, and the fact that onlyone task defined the rhyme sensitivity and syllabic sensitiv-ity factors. Stahl and Murray (1994; see also Stanovich et al.,1984) reported that different factor loadings of phonologicalsensitivity tasks were most likely the result of the differentlinguistic complexity between tasks and not necessarilydifferent phonological abilities.

There have been only a small number of studies that haveexamined phonological sensitivity in preschool-age chil-dren. Most of these studies have been limited both by smallsample sizes at each age level studied and by the use of onlyone or two measures of phonological sensitivity. Theselimitations preclude examination of developmental trendswithin or relations between phonological sensitivity tasks atdifferent levels of linguistic complexity. For instance, Fox

and Routh (1975) asked 50 children (10 at each age from 3 to7 years) to segment sentences into words, words intosyllables, and syllables into phonemes. They found that evensome of the 3-year-olds could segment syllables into pho-nemes and that all segmenting abilities increased with age.Both Lenel and Cantor (1981) and Smith and Tager-Flushberg (1982) found age-related performance differenceson a forced-choice rhyme matching task with preschool-agechildren. Chaney (1992) administered several phonologicalsensitivity tasks (i.e., rhyme matching, sentence segmenting,phoneme blending) to 43 three-year-old children but did notreport the relations between performance on the differenttasks; however, a composite phonological variable wascorrelated with both age and language scores. Using confir-matory factory analysis with a group of 105 four- andfive-year-old children, Wagner et al. (1987) found thatmeasures of syllabic sensitivity (i.e., syllable counting,syllable elision, and syllable blending) were related torhyme tasks and to phonological memory tasks indepen-dently of general cognitive ability.

In one of the more extensive studies to date, MacLean,Bryant, and Bradley (1987) gave a group of 66 three-year-old children a rhyme detection task and a knowledge ofnursery rhyme task. In subsequent assessments, these chil-dren completed an alliteration detection task, an alliterationproduction task, a rhyme production task, a segmenting task,and additional rhyme detection tasks. The results of the firsttwo assessments, when the mean age of the children wasunder 4 years, indicated that approximately 25% of thechildren could detect, produce, or both detect and producerhyme and alliteration. When the children averaged 54months of age, their ability to read 12 simple high-frequencywords was assessed. Analyses comparing readers and non-readers on initial rhyme detection, alliteration detection, andknowledge of nursery rhyme scores revealed that readershad scored significantly higher than nonreaders even whencontrolling for differences in age and IQ. Bryant et al. (1990)reported additional data on these children who performedphoneme deletion and phoneme tapping tasks when theyaveraged 71 months of age, and reading and spelling testswhen they averaged 79 months of age. Rhyme and allitera-tion oddity tests that had been administered when thechildren were 55 months of age were correlated with thephoneme deletion and phoneme tapping tasks (averager = .48). In a series of hierarchical multiple regressions,scores on these rhyme and alliteration tasks significantlyadded to the prediction of reading and spelling scores evenafter the mother's educational level; the child's age, IQ, andreceptive vocabulary; and the score on one of the threephoneme deletion and phoneme tapping tasks were takeninto account.

Despite their limitations, these studies indicated that somedegree of phonological sensitivity can be measured inchildren as young as 3 years of age, that early phonologicalsensitivity is related to oral language, that the phonologicalabilities measured by rhyme and alliteration detection tasksat age 3 years share a component of what is measured byhigher level phonological sensitivity tasks such as phonemedeletion and phoneme tapping and that phonological sensi-

296 LONIGAN, BURGESS, ANTHONY, AND BARKER

tivity measured during the preschool period is related to thelater development of reading. The fact that rhyme andsyllabic sensitivity measures correlate with measures ofphonemic sensitivity suggests that these measures tap thesame underlying phonological sensitivity process at differ-ent levels of linguistic complexity (e.g., Adams, 1990;Stanovich, 1992). These results suggest that the lower levelsof phonological sensitivity measured by rhyme and syllabicsensitivity tasks may be developmental precursors to higherlevels of phonological sensitivity rather than abilities unre-lated to later reading; however, additional examination ofpreschool children's ability to perform tasks at differentlevels of linguistic complexity and the relations betweenthese tasks at different ages is required.

Questions concerning the early development of phonologi-cal sensitivity, the measurement of phonological sensitivityin preschool-age children, and the coherence of tasksdesigned to measure phonological sensitivity during thepreschool years are becoming increasingly important, be-cause several recent studies have indicated that individualdifferences in phonological sensitivity are relatively stableacross time from kindergarten on (e.g., Byrne, Freebody, &Gates, 1992; Torgesen & Burgess, in press; Wagner et al.,1994). As noted by Wagner et al. (1994), this stabilityhighlights the value of early screening of phonologicalsensitivity to identify children who may be at risk forreading difficulties. In addition, because it appears thatdifferences in phonological sensitivity may be present priorto school entry and relatively stable, the search for potentialcauses or early correlates of the development of phonologi-cal sensitivity might best be directed toward preschool-agechildren (e.g., Whitehurst & Lonigan, 1998).

In addition to examining the development of children'sphonological processing skills, a few studies have examinedsocioeconomic status (SES) differences in phonologicalsensitivity in prereaders as an attempt to explain SESdifferences in reading achievement. There are substantialdifferences in the reading and writing ability of children as afunction of the economic level of their parents (NationalAssessment of Educational Progress, 1991), and SES is oneof the strongest predictors of performance differences inchildren at the beginning of first grade (Entwisle & Alex-ander, as cited in Alexander & Entwisle, 1988, p. 99).However, several studies conducted with children prior toformal reading instruction have yielded mixed results con-cerning SES differences in phonological sensitivity. Forinstance, although MacLean et al. (1987) found effects ofsocial class and parental education on 3-year-old children'sability to identify rhyme and alliteration, these differenceswere reduced when children's IQ scores were controlled.Raz and Bryant (1990) reported significant social classdifferences on rhyme oddity tasks in forty 5-year-old andforty 6-year-old children from higher and lower SES samplesmatched on IQ; however, they did not find social classdifferences on an initial phoneme isolation task in theyounger group of children when they were 4 years old.These findings have been interpreted as indicating that SESdifferences in preschool children reflect underlying differ-ences in general and verbal abilities. In contrast to these

findings, however, Bowey (1995) reported that SES differ-ences on rhyme or final phoneme matching measuresremained even after controlling for performance IQ andlanguage in a sample of forty-eight 5-year-old childrenselected to represent extreme SES groups. Given the impor-tance of phonological sensitivity to reading and the possibil-ity that lower levels of phonological sensitivity are develop-mental precursors to higher levels of phonological sensitivity,including phonemic sensitivity, it is crucial to examinewhether SES differences are present on these early measuresand when they emerge. Such information will both helpguide intervention efforts to ameliorate the negative impactof poverty on educational achievement and provide cluesconcerning the origins of phonological sensitivity during thepreschool years.

To address limitations of previous studies with preschool-age children, this study was designed to examine thedevelopment of preschool-age children's performance ontasks designed to assess phonological sensitivity, to examinethe interrelatedness of children's phonological sensitivityacross different levels of linguistic complexity, and toinvestigate potential SES differences in phonological sensi-tivity during the preschool years. Our strategy was to utilizeseveral different measures of phonological sensitivity thatwere likely to be within the capabilities of preschool-agechildren. Performance on these tasks was examined cross-sectionally in a relatively large sample of preschool childrenfrom lower and higher income families. We includedmeasures of syllabic and intrasyliabic sensitivity becausechildren's sensitivity to these linguistic units may serve asdevelopmental precursors to phonemic sensitivity and be-cause these are likely to be tasks that can be performed bypreschool-age children. Several indicators of phonemicsensitivity were also included. On the basis of prior research(e.g., MacLean et al., 1987; Wagner et al., 1994), wehypothesized that the performance of very young childrenwould be low and variable across these measures, but thatperformance would both increase and become more consis-tent across tasks in older children. Because previous studieshave found a relation between measures of phonologicalsensitivity and children's language skills (e.g., Bowey, 1995;Chaney, 1992; Fox & Routh, 1975; Smith & Tager-Flushberg, 1982), standardized measures of oral languagewere used to control for variance in phonological sensitivitytasks that was due to language abilities.

Method

Participants

Two groups of preschool-age children were used in this study.One group of children consisted of 238 preschoolers who were apart of a longitudinal study. These children were recruited throughchild care centers serving middle- to upper income families. Thesechildren ranged in age from 25 to 70 months (M = 50.4 months,SD - 11.71 months). The majority of the children were white(93%) and 52% were girls. The second group of children consistedof 118 preschoolers who were a part of an intervention study forchildren at risk for language and reading difficulties because ofconditions associated with poverty. These children were recruited


from child care centers serving children from low-income families(i.e., families eligible for state-subsidized child care). This group ofchildren ranged in age from 25 to 64 months (M = 45.6 months,SD = 10.10 months). The majority of the children were AfricanAmerican (82%) and 47% were girls. Questionnaire data collectedon a subsample of both groups indicated significant differences inmother's education between the middle-income (Mdn = 16 years)and lower-income {Mdn — 12 years) samples, F(l, 224) = 57.81,p < .001. Only children who had completed both the phonologicalsensitivity and language assessments were included in thesesamples. An additional 23 children (5 from the lower-incomegroup) who had incomplete data on one or both sets of measureseither because they refused continued participation or because theyhad left the center they were attending prior to the completion ofthe assessments were excluded.

Measures

Four tasks (rhyme oddity detection, alliteration oddity detection,blending, and elision) were used as the main measures of phonologi-cal sensitivity and were administered to all children. Two additionaltasks, rhyme matching and category oddity detection, were admin-istered to a subgroup of the children to determine whetherperformance on the two phonological oddity tasks reflected phono-logical sensitivity or the task demands of the oddity procedure.Each of these six tasks consisted of both practice and test trials.Practice trials were followed by correction, explanation, andreadministration if the child gave an incorrect answer, or confirma-tion and explanation if the child gave the correct answer. There wasno feedback on any of the test trials. Subgroups of the olderchildren in the middle-income sample and of children in the lower-income sample were administered a test of letter knowledge and anenvironmental print test. Children were asked to name all 26 lettersof the alphabet shown to them printed in uppercase on individual3 x 5 in. (7.6 X 12.7 cm) index cards presented in a random order.On the environmental print task, children were shown photographsof labels and signs and asked to say what the sign said; childrenwere also asked to read the word from the sign out of itsenvironmental context (i.e., as printed text only). Children's abilityto read these words was used as an index of word reading ability.

Children also took standardized tests of oral language. Receptivelanguage was assessed with the Peabody Picture VocabularyTests—Revised (PPVT-R; Dunn & Dunn, 1981). Some of theolder children in the middle-income sample were administered theGrammatical Understanding subtest of the Test of LanguageDevelopment: Primary (TOLD; Newcomer & Hammill, 1988)instead of the PPVT-R. Scores on this subtest of the TOLD arecorrelated strongly with scores on the PPVT-R in this age group(/• — ,73), indicating that both tests measure similar abilities.Expressive language was assessed using the Expressive One-WordPicture Vocabulary Test—Revised (EOWPVT-R; Gardner, 1990)and the Grammatical Closure subtest of the Illinois Test ofPsycholinguists Abilities (ITPA-GC; Kirk, McCarthy, & Kirk,1968). These tests are normed and validated for use with childrenof this age, and indices of internal consistency for each test are high(e.g., split-half reliabilities: PPVT-R = .80, TOLD = .76,EOWPVT-R = .94, ITPA-GC = .86). Scores on the EOWPVT-Rand the ITPA-GC were highly correlated (r = .64); consequently,an expressive language composite score was computed and used insubsequent analyses to reduce the number of variables.

Rhyme oddity detection. This task was patterned after the taskdeveloped by MacLean et al. (1987) with the use of their word lists.Two of the three words in each set rhymed (e.g., fish, dish, book;sail, boot, nail). Children were presented with three pictured wordsthat were named by the examiner. Children were then asked to

select the one not rhyming with (or that "did not sound the sameas," or was "different than") the other two. The task consisted of 2practice trials and 11 test trials that were administered to allchildren. The position of the odd word across trials was randomlydetermined and was the same for all children.

Alliteration oddity detection. This task was also patterned afterthe one used by MacLean et al. (1987) with the use of their wordlists. Two of the words in each set began with the same singletononset (e.g., toad, toaster, girl; bed, hair, bell). Children werepresented with three pictured words that were named by theexaminer. Children were then asked to select the one that did notsound the same at the beginning of the word (or that was "differentat the start of the word"). The task consisted of 2 practice trials and11 test trials that were administered to all children. The position ofthe odd word across trials was randomly determined and was thesame for all children.

Blending. On this task, children were required to combineword elements to form a new word. There were 3 practice trials and22 test trials. The practice items and the first 12 test trials werepresented both verbally and with pictures; the final 10 test trialswere presented verbally only (see Appendix). The initial items inboth picture and nonpicture trials required blending single-syllablewords1 to form compound words. Later items in both picture andnonpicture trials consisted of syllables or phonemes. For pictureditems involving compound words, the examiner showed the childtwo pictures and named them (e.g., "This is a cow, and this is aboy.") and then asked the child what word would be produced if heor she said them together (e.g., "What [word] do you get when yousay cow . . . boy together?"). For the picture trials involvingblending of nonword-syllables or phonemes, the examiner showedthe child three pictures, one of which corresponded to the correctanswer, and named each prior to the trial; the child could respond tothe item by saying the word or pointing to the correct picture. Allpractice items required the blending of compound words. Theexaminer emphasized the nature of the task during the practicetrials by putting the pictures together while presenting the trial.Testing was discontinued after a child missed five consecutivetrials.

Elision. On this task, children were required to say a wordminus a specific sound. There were 2 practice trials and 17 testtrials. The practice items and first 9 test trials were presented bothverbally and with pictures; the final 8 test trials were presented onlyverbally (see Appendix). Initial items within both picture andnonpicture trials consisted of compound words, and the child wasrequired to say what the word would be with one part of it deleted.Later items in both picture and nonpicture trials consisted ofnonword-syllables or phonemes. All practice items used com-pound words. For pictured items, the examiner showed the childtwo pictures and named them (e.g., "This is bat, and this is man")and then asked the child to say the compound (i.e., "batman") priorto being asked to delete part of it. For the picture trials involvingsyllables or phonemes, the examiner showed the child threepictures, one of which corresponded to the correct answer, andnamed them prior to the trial, and the child could either say theword or point to the correct picture. The examiner emphasized thenature of the task during administration of the practice trials byphysically removing the picture of the word to be deleted. Test-ing was discontinued after a child missed five consecutive trials.

1 Blending and elision tasks conducted with compound wordsrepresent a special case of syllabic sensitivity rather than wordsensitivity because the components of the compound consist ofsingle-syllable words.


Rhyme matching. This task used the same word list andpictures as the rhyme oddity detection task. Children were pre-sented with a picture on a small card and had to indicate with whichof two additional pictured words it rhymed. The examiner namedall three pictures before and during a trial. All children wereadministered 2 practice trials and 11 test trials.

Category oddity detection. This task followed the same presen-tation and administration format as the two sound oddity detectiontasks; however, instead of children selecting the odd sound theywere required to select the odd object out of three (see Appendix).Three pictured objects were presented to the child. Two of thepictures represented the same category of object (e.g., dog); thethird picture represented a different category of object (e.g., cow).The examiner did not name the pictures but simply showed thechild the pictures and asked which one did not match (or was notthe same as the others). All children were administered 2 practicetrials and 11 test trials.

Procedure

After obtaining informed consent from parents, a trained exam-iner administered the assessment measures to the children. Allassessments were conducted individually in a separate room or in aquiet location in the children's child care centers. The four mainmeasures of phonological sensitivity were administered in a singlesession that lasted approximately 20 min. Typically, language,letter knowledge, and environmental print tests were administeredto the 4- and 5-year-old children in the middle-income sampleduring this same session. For the 2- and 3-year-old children in themiddle-income sample and for all children in the lower-incomesample, language tests and the letter knowledge and environmentalprint tests were administered during separate sessions within 2weeks of administration of the phonological sensitivity measures.Only a subset of the children were administered the rhymematching and category oddity tasks. These tasks were administeredin a separate session within 2 to 3 weeks of the administration ofthe four main phonological sensitivity measures.

Results

Preliminary Analyses

There were no differences between girls and boys inoverall performance on the four main phonological tasks (allps > .12). Gender did not enter into any interactions withage group (all ps > .21), SES group (all ps > .49), or theAge Group X SES Group interaction (allps > .18). Conse-quently, all analyses were conducted with boys and girlscombined. Children from the middle-income sample wereolder than the children from the lower-income sample, F(l ,354) = 14.54,/? < .001, Scores on each of the phonologicalsensitivity tasks were correlated with children's ages(rs = .38, .43, .60, and .66 for rhyme oddity, alliterationoddity, blending, and elision respectively; all ps < .001.).Children were divided into four age groups (i.e., 2-, 3-, 4-,and 5-year-olds) to examine age and SES group differences.The number of children in each age group for the middle-income and lower-income groups as well as the mean age ofchildren in each age group is shown in Table 1. An analysisof variance (ANOVA) on chronological age showed that theAge Group X SES Group interaction was not significant,F(3, 348) = 1.26, p — .29, indicating that the change in agesbetween age groups was equivalent for middle-income andlower-income samples.

A series of analyses of covariance (ANCOVA) revealedthat as a group, children from the middle-income sampleperformed significantly better than the children from thelower-income sample on the rhyme oddity task, F(l , 353) =17.58, p < .001, the alliteration oddity task, F(l, 353) =7.11, p = .008, the blending task, F(l , 353) = 64.64, p <.001, and the elision task, F(\y 353) = 57.77,p < .001, withchronological age used as a covariate. Children from thelower-income sample had lower receptive language scores(M = 79.09, SD = 17.66) than children from the middle-

Table 1Descriptive Statistics for Children's Ages and Scores on Phonological Sensitivity Tasks for Each Age Group and Sample

Variable

nAge (months)

MSD

Rhyme oddityMSD

Alliteration oddityMSD

BlendingMSD

ElisionMSD

2-year-olds

35

30.913.02

4.51a1.58

3.17a1.34

0.97a2.99

0.89a1.62

Middle-income sample

3-year-olds

56

41.793.63

4.13a1.25

3.48a1.51

3.68b6.05

1.98a

3.52

4-year-olds

82

54.073.42

5.43^2.51

4.68b2.10

11.99C7.67

7.71b

4.64

5-year-olds

65

61.732.60

7.09b2.51

5.97C2.83

16.08d6.02

10.77c

4.25

2-year-olds

20

30.852.74

4.10a,b1.89

2.90a1.21

0.30a,b0.57

0.40a

0.88

Lower-income sample

3-year-olds

45

41.093.41

3.89,,,1.40

3.53a1.46

l-22a3.69

l-07a

1.83

4-year-olds

38

52.343.12

4.84b1.70

3.71^1.47

2.16b4.29

2.71b

3.14

5-year-olds

15

61.731.44

4.27^1.39

4.27b1.16

1.93b4.68

2.73b

3.65

Note, Means for variables within middle-income and lower-income groups that do not share a common subscript differ at the/? < .05 level.


income sample (M= 101.11, SD = 14.25), F(l, 354) -160.09, p < .0001, and these differences were evident ineach age group (all ps < .001). Children from the lower-income sample also had lower expressive language scores(M = 86.21, SD = 12.59) than children from the middle-income sample (M = 110.74, SD = 21.12), F(l , 354) -135.33, p < .0001, and these differences were evident ineach age group (ps < .001 for 3-, 4-, and 5-year-olds,p = .04 for 2-year-olds). Scores on the phonological sensitiv-ity tasks were correlated with receptive language scores(rs = .22, .19, .39, and .38 for rhyme oddity, alliterationoddity, blending, and elision respectively, allps < .001) andexpressive language scores (rs = .34, .31, .47, and .50 forrhyme oddity, alliteration oddity, blending, and elisionrespectively, allps < .001).

Age and SES Differences on Task Performance

Descriptive statistics for each of the four main phonologi-cal sensitivity measures are shown in Table 1. Generally,age-related performance differences on each of the taskswere evident. The youngest children's performance waslow; however, there were substantial increases in perfor-mance on all of the tasks between the ages of 3 and 4 years.Because of the significant differences in the language scoresof the middle-income and lower-income groups, a series of 4(age group) X 2 (SES group) ANCOVAs, controlling forreceptive and expressive language scores, was used toexamine differences on the four main phonological sensitiv-ity tasks.2 Neither covariate entered into a significantinteraction with age group (all ps > .06),3 SES group (allps > .24), or the Age Group X SES Group interaction (allps > .18); consequently, the covariates met the homogeneityof regression assumption of ANCOVA.

On the rhyme oddity task there was a significant AgeGroup X SES Group interaction, F(3, 346) = 3.30, p = .02.Trend analysis revealed a significant Age Group X SESGroup interaction for the linear trend, F( 1, 346) = 4.08, p =.04, and the quadratic trend, F(l, 346) = 5.86, p = .02. Inthe middle-income group there were significant linear, F(l,346) = 37.95, p < .0001, and quadratic, F(l, 346) = 14.44,p < .001, trends in rhyme oddity scores across age group,but not in the lower-income group (ps > .08). Contrastsbetween age groups for both SES groups are shown in Table1. Rhyme oddity scores for children in the middle-incomesample were higher than those of the lower-income samplein the 5-year-old group (p < .0001) but not in the 2-, 3-, or4-year-old groups (ps > .54).

On the alliteration oddity task, there was a significant effect ofage group, F(3, 346) = 12.80, p < .001, but no SES group orAge Group X SES Group interaction (ps > .29). Trend analysisrevealed a significant linear trend in alliteration oddity scoresacross age group, F(l, 346) = 36.12,p < .0001. Follow-upcontrasts indicated that scores for the 2-year-olds did notdiffer from those of the 3-year-olds (p = .14), but all otherage groups differed from each other (all ps < .02). Allitera-tion oddity scores were higher in the middle-income samplethan in the lower-income sample for 5-year-olds (p < .001)but not for the 2-, 3-, or 4-year-olds (ps > . 11).

On the blending task, there was a significant Age Group XSES Group interaction, F(3, 346) = 12.49, p < .001. Trendanalysis revealed a significant Age Group X SES Groupinteraction for the linear trend, F(l , 346) = 30.36, p <.0001. Although there were significant linear trends in boththe middle-income, F(l, 346) = 179.83, p < .0001, andlower-income, F(l, 346) = 10.18, p = .002, samples, theeffect was stronger in the middle-income sample. There wasa significant cubic trend for the middle-income sample, F(3,346) = 5.38,p = .02. As indicated in Table 1, all age groupsin the middle-income sample differed from each other (allps < .007), whereas the effect in the lower-income groupwas due mostly to the older age groups. Children in themiddle-income sample scored higher than children in thelower-income sample in the 4- and 5-year-old groups(ps < .001) but not in the 2- or 3-year-old groups (ps > .19).

On the elision task, there was a significant Age Group XSES Group interaction, F(3, 346) = 8.29, p < .001. Trendanalysis revealed a significant Age Group X SES Groupinteraction for the linear trend, F(l , 346) = 18.92, p <.0001. Similar to the results for the blending task, there weresignificant linear trends in both the middle-income, F(l ,346) = 208.23, p < .0001, and lower-income, F(l , 346) =21.34, p < .0001, samples, but the effect was stronger in themiddle-income sample. There was a significant cubic trendfor the middle-income sample, F(3, 346) = 8.99, p = .003.As noted in Table 1, most age groups in the middle-income(ps < .0001 for significant effects) and lower-income samples(ps < .003 for significant effects) differed from each other.Elision scores were higher in the middle-income samplethan in the lower-income sample for both the 4- and5-year-old groups (ps < .001) but not for the 2- or 3-year-old groups (ps > .13).

The percentage of children within each age group scoringabove chance (i.e., 6 or more correct by binomial probabilityatp < .05) on the rhyme oddity and alliteration oddity tasksis shown separately in Table 2 for the middle-income andlower-income groups. For the middle-income group, thenumber of children scoring above chance was associatedwith age for both the rhyme oddity task, X2(3» N = 238) =

2 Because of moderate positive skew in the four phonologicalsensitivity variables, two additional sets of analyses were con-ducted to determine the robustness of the results reported below. Inthe first set of analyses, the phonological sensitivity variables weretransformed using a square-root transformation to normalize theirdistributions and the ANCOVAs reported below were repeatedusing the transformed variables. In the second set of analyses,nonparamateric statistics were used to examine age and SES groupdifferences on the phonological sensitivity variables. Both sets ofanalyses reproduced the results reported for the ANCOVAs on theuntransformed variables, with the exception that SES group effectswithin the different age groups were larger in nonparametricanalyses, which did not control for children's language skills.

3 Because of the near-significant interaction of the covariates andage group on the blending task that was due to inclusion ofreceptive language scores as a covariate, a separate ANCOVA forthe blending task was conducted without using receptive languagescores as a covariate. Results identical to those reported belowwere obtained.


Table 2Percentage of Children at Different Ages Scoring Above Chance on Rhyme Oddity and Alliteration Oddity Tasksand Percentage Able to Complete Items on Blending and Elision Tasks at Different Levels of Linguistic Complexity

Age group

Middle-income sample2-year-olds3-year-olds4-year-olds5-year-olds

Lower-income sample2-year-olds3-year-olds4-year-olds5-year-olds

n

35568265

20453815

Rhymeoddity

25.714.339.075.4

25.013.336.820.0

Alliterationoddity

0.08.9

34.149.2

5.08.9

10.513.3

Words

14.739.381.793.8

25.024.450.026.7

Blending

Syllables

8.625.069.589.2

0.06.7

13.213.3

Phonemes

2.921.450.080.0

0.06.75.36.7

Words

42.939.392.7

100.0

20.042.265.860.0

Elision

Syllables

5.819.669.583.1

5.04.4

26.320.0

Phonemes

2.910.745.172.3

0.04.4

18.46.7

51.73, p < .001; p = .43, and the alliteration oddity task,X2(3, AT = 238) = 40.34, p < .0001; p = .41. For the lower-income group, age was not related to the number of childrenscoring above chance on the rhyme oddity task, x2(3,N = 118) = 6.43, p < .10; p == .10, or to the alliterationoddity task, x2(3,A^= 118) = 0.83,/) = .85; p = .08. In the5-year-old group, more middle-income children than lower-income children scored above chance on the rhyme odditytask, x2(l, N = 80) = 16.43, p < .001. More middle-incomethan lower-income children scored above chance on thealliteration oddity task in the 4-year-old group, x2U»N = 120) = 7.41, p = .006, and 5-year-old group, x2(l,N = 80) = 6.43, p = .01. These results indicate that amoderate percentage of the 2- and 3-year-old children wereable to detect rhyme and a smaller percentage could detectalliteration. Older children were more likely to score atabove-chance levels, and this effect was stronger in themiddle-income sample.

Both the blending task and the elision task included itemsat different levels of linguistic complexity (i.e., words,syllables, phonemes). To examine children's performance onthe blending and elision tasks at these different levels oflinguistic complexity, we calculated separate scores foritems involving words (blending items 1-9, 13-17; elisionitems 1-6, 10-13), syllables (blending items 10, 11, 18;elision items 7, 8,14,15), and phonemes (blending items 12,20-22; elision items 9, 16, 17). The percentage of childrenwithin each age group with scores above zero on thesedifferent component scores is shown in Table 2 for themiddle-income and lower-income samples.4

On the blending task, the number of children in themiddle-income sample with scores of one or higher wasassociated with age for the word-level items, X2(3* N —238) = 89.83, p < .0001; p = .59; the syllable-level itemsX2(3, N = 238) = 88.75, p < .0001; p = .60; and thephoneme-level items, x2(3, N - 238) = 70.81, p < .0001;p — .55. In contrast, the number of children in the lower-income sample with scores of one or higher was notassociated with age for the word-level items, X2(3> N —118) = 7.30, p < .07; p = .14; syllable-level items, x2(3,N = 118) = 3.57, p = .31; p = .17; or phoneme-level items,X2(3, N = 118) = 1.39,/? = .71; p = .06, on the blending

task. More children in the middle-income sample than in thelower-income sample scored one or higher on the word-levelitems in the 4- and 5-year-old groups (ps < .001), on thesyllable-level items in the 3-, 4-, and 5-year-old groups(ps < .01), and on the phoneme-level items in the 3-, 4-, and5-year-old groups (ps < .04).

On the elision task, age was strongly associated with thenumber of children in the middle-income sample who scoredone or higher on the word-level items, x2(3, N = 238) =92.20, p < .0001; p = .57, the syllable-level items, x2(3,AT = 238) = 88.77, p < .0001; p = .59, and the phoneme-level items, x2(3, N = 238) = 70.11, p < .0001; p = .54.Age was more weakly associated with the number ofchildren in the lower-income sample who scored one orhigher on the word-level items x2(3, N= 118)= 12.56, p =.006; p - .30, the syllable-level items x2(3, N= 118) =10.25, p = .02; p = .25? and the phoneme-level items X2(3.N = 118) = 7.70, p < .06; p = .18. More children in themiddle-income sample than in the lower-income samplescored one or higher on the word-level items in the 4- and5-year-old groups (ps < .001), on the syllable-level items inthe 3-, 4-, and 5-year-old groups (ps < .02), and on thephoneme-level items in the 4- and 5-year-old groups(ps < .005). These results indicate that even a small tomoderate percentage of the 2- and 3-year-old children wereable to manipulate sounds at the syllable- and phoneme-level. Older children were more likely to be able tomanipulate syllables and phonemes, and this effect wasstronger in the middle-income sample.

Internal Consistency and Interrelatednessof Phonological Sensitivity Tasks

Because of the SES differences between the middle-income and lower-income groups, differences in the groups'scores on the phonological sensitivity measures, and the

4 Because of zero variance in several cells for the different levelsof linguistic complexity in the lower-income sample, the paramet-ric analyses reported for the total scores on the blending and elisiontasks could not be conducted because of violations of the homoge-neity of variance assumption of ANOVA, ANCOVA, or both.


Table 3Internal Consistency Indices (Alphas) for Four Phonological Sensitivity Tasksat Different Ages

Sample

Middle-income sample2-year-olds3-year-olds4-year-olds5-year-olds

Lower-income sample2-year-olds3-year-olds4-year-olds5-year-olds

Rhyme

c

c

.63

.67

.30c

.12c

Full measures

Alliteration

c

c

.44

.74

c

c

c

c

i Blending

.94

.96

.96

.93

.14

.96

.94

.96

Elision

.74

.92

.89

.88

.54

.77

.84

.90

Components

Blending8

.93

.81

.83

.72

d

.93

.74

.91

Elisionb

.63

.81

.73

.66

.52

.58

.51

.65

"Correlation between word-level syllable items on blending task and nonword syllable- andsubsyllable-level items on blending task. bCorrelation between word-level syllable items on elisiontask and nonword syllable- and subsyllable-level items on elision task. cAverage item to itemcorrelation negative; unable to compute internal consistency. dZero variance for scores on syllableand phoneme-level items; unable to compute correlation.

differences in the groups' overall oral language skills,internal consistency and correlational analyses were con-ducted separately for the middle-income and lower-incomesamples.

Middle-income sample. Indices of internal consistencyfor each of the four main phonological tasks for each of thefour age groups in the middle-income sample are shown inTable 3. As indicated in the table, the internal consistency ofthe blending task and the elision task was high for each ofthe four age groups. Because many of the children in theyounger age groups were not administered the later items onthese tests, the indices of internal consistency might havebeen inflated by the high proportion of failed (or nonadmin-istered) items. To assess this possibility, we calculated twoadditional sets of reliability estimates. The first alternativecalculated the reliability for the items up to the maximumscore achieved by a child in that age group. This procedureresulted in estimates of reliability that were similar to those

for the full test for both the blending (as = .94 and .96 for2- and 3-year-olds, respectively) and the elision (as = .70and .89 for 2- and 3-year-olds, respectively) tasks. Thesecond alternative calculated the reliability for the first fiveitems on the tests because all children completed theseitems. This procedure also resulted in similar estimates ofthe reliability for the blending task (as = .87 and .89 for2- and 3-year-olds, respectively) and for elision with the3-year-olds (a = .80). The reliability estimate for elisionwith the 2-year-olds was reduced (a = .57).

The internal consistency of the two oddity tasks wassubstantially lower than it was for the blending and elisiontasks for all age groups. The internal consistencies of therhyme oddity and alliteration oddity tasks were particularlypoor for the 2- and 3-year-old children. For both tasks, theestimate of internal consistency could not be calculatedbecause the average interitem correlation was negative. Forboth older age groups, the rhyme oddity task had acceptable

Table 4Partial Correlations Controlling for Receptive and Expressive Language Scores BetweenScores on Blending and Elision Tasks at Different Levels of Linguistic Complexityin Middle-Income Sample

Correlated variables

nBlending task

Word with syllableWord with phonemeSyllable with phoneme

Elision taskWord with syllableWord with phonemeSyllable with phoneme

Fullsample

7g*#*49***53***

.60***

.53***

.56***

2-year-olds

35

97***.40*.37*

44***^j***

43***

Age group

3-year-olds

56

84***73***81***

_75***72***73***

4-year-olds

82

g3*#*57***.50***

67***57***.58***

5-year-olds

65

71***_5Q***jj7***

.51***

.50***

.55***

Note. Partial correlations for full sample were determined controlling for receptive language,expressive language, and chronological age.*p<.05 . ***/>< .001.


Table 5Correlations Between Tests of Phonological Sensitivity and Language Standard Scoresfor 2-Year-Old (Above Diagonal) and 3-Year-Old Children (Below Diagonal)in Middle-Income Sample

1.2.3.4.5.6.

Task

Rhyme oddityAlliteration oddityBlendingElisionReceptive languageExpressive language

1

—.19.02.06

-.17-.21

2

-.04—.15.06

-.12.05

3

.15

.01—

.37**

.11

.15

4

-.34*-.09

.06—.23.15

5

-.17-.16-.04

.23—69***

6

-.21.05.15.1569***—

Note, n = 35 for 2-year-olds; n = 56:*p < .05. **p < .01. ***p < .001.

internal consistency, and the alliteration oddity task hadmarginal (i.e., for the 4-year-olds) to acceptable internalconsistency. Table 3 also shows the correlations between theitems that involved compound words with items that in-volved nonword syllables or phonemes on the blending andelision tasks. For each age group, these correlations werehigh and were not substantially different from the alphas forthe entire task. These results suggest that both the manipula-tion of single-syllable words and the manipulation ofnonword syllables and phonemes involve similar phonologi-cal processing abilities. Table 4 shows partial correlationsfor blending and elision scores at the different levels oflinguistic complexity (i.e., word, syllables, phonemes) aftercontrolling for receptive and expressive language scores.Scores at the different levels of linguistic complexity weresubstantially correlated for the sample as a whole and withineach of the four age groups, and the correlations betweenperformance at the different levels of linguistic complexitywere not due to variance common to both phonologicalsensitivity tasks and language measures.

Correlations between the four phonological tasks and orallanguage scores for each of the four age groups in themiddle-income sample are shown in Table 5 and Table 6.Preliminary analyses examining the distribution of variablesin the different age groups indicated that the distribution ofblending and elision scores had moderate positive skew inthe two younger age groups. Although transformationsnormalized the distributions, correlations using the trans-formed variables were virtually identical to those using the

untransformed variables; consequently, untransformed vari-ables were used in subsequent analyses. Correlations be-tween phonological tests for the 2-year-olds were low andnonsignificant (average r = —.04). The average correlationbetween the tests was higher for the 3-year-olds (averager = .15), and an examination of the four tests as single itemsin a phonological processing test battery yielded an alpha of.37. Correlations between phonological sensitivity and orallanguage were generally low and nonsignificant for the2- and 3-year-olds. For the 4-year-olds (average r = .29) andthe 5-year-olds (average r = .30), the four phonologicaltests were moderately intercorrelated and yielded moderatealphas (a = .52 and a = .57 for the 4- and 5-year-olds,respectively). Correlations between phonological sensitivityand oral language for the 4- and 5-year-olds were moderatefor most tasks and tended to be strongest for expressivelanguage and the nonoddity tasks.

The pattern of correlations suggested that, generally,scores on the two oddity tasks were more related to eachother than to the blending and elision tasks, and the blendingand elision tasks were more highly related to each other thanto the oddity tasks. Principal-components analyses of thefour tests supported this observation. A two-factor solution(Factor 1 = Oddity Tasks, Factor 2 = Nonoddity Tasks) wasobtained for the 3- and 4-year-olds, whereas a one-factorsolution was obtained for the 5-year-olds. Separate correla-tions between word-level items and nonword-level items(i.e., syllables, phonemes) on the blending and elision tasksshowed relations similar to that of the full measures with the

Table 6Correlations Between Tests of Phonological Sensitivity and Language Standard Scoresfor 4-Year-Old (Above Diagonal) and 5-Year-Old Children (Below Diagonal)in Middle-Income Sample

Task

1. Rhyme oddity2. Alliteration oddity3. Blending4. Elision5. Receptive language6. Expressive language

1

—39***.1245***.1139***

24Q***

—.17.23.25*.36***

3

.12

.25*—

42***.23.37***

4

.22*

.26*45***

—.38**47***

5

41***.28*.30**.28*

—40***

6

.29**

.28*

.26*

.26*

.34**—

Note. « = 82 for 4-year-olds; n = 65 for 5-year-olds.*p<.05. **p<-01. ***/>< .001.


Table 7Correlations Between Tests of Phonological Sensitivity and Language Standard Scoresfor 2-Year-Old (Above Diagonal) and 3-Year~Old Children (Below Diagonal)in Lower-Income Sample

Task

1. Rhyme Oddity2. Alliteration Oddity3. Blending4. Elision5. Receptive Language6. Expressive Language

1

—- .06-.11- .06-.07- .07

2

.05—

-.06- .08-.15- .35*

3

.17

.05—

-.05-.08-.08

4

.10- .01- .25—

-.19.14

5

.07

.02

.21

.19—

49***

6

.06

.33

.08

.38t

.76***—

Note, n = 20 for 2-year-olds; n = 45 for 3-ytp < .10 (marginally significant). *p < .05.

rhyme and alliteration scores for each age group. Theseresults support the findings of the item analyses of the fourtests and indicate that the tests were more reliable measuresof phonological sensitivity for the 4- and 5-year-olds thanthey were for the 2-year-olds.

Lower-income sample. Estimates of the internal consis-tencies of the four main phonological measures for childrenin the lower-income sample are shown in Table 3. Estimatesof internal consistency could not be calculated for thealliteration oddity task because the average interitem correla-tions were negative. Estimates of internal consistency for therhyme oddity task were either low or could not be calculatedbecause the average interitem correlations were negative.With the exception of the 2-year-olds' data, estimates ofinternal consistency for the blending and elision tasks weresubstantial. These estimates were relatively unaffected bythe alternative methods of calculating internal consistency(i.e., maximum score item and first five items). Table 3 alsoshows the correlations between the items that involvedcompound words with items that involved nonword syl-lables or phonemes on the blending and elision tasks for thelower-income sample. Similar to the results with the middle-income sample, these correlations were high to moderateand were not substantially different from the alphas for theentire task, suggesting that both types of items requiresimilar phonological processing abilities. Partial correla-tions for blending and elision, after controlling for age andreceptive and expressive language scores, showed thatscores at the different levels of linguistic complexity weresubstantially correlated for the lower-income sample as awhole (blending: rs = .81, .59, and .65 for word withsyllable, word with phoneme, and syllable with phoneme,respectively; elision: rs = .40, .51, and .69 for word withsyllable, word with phoneme, and syllable with phoneme,respectively; ps < .00I).5

Correlations between the four phonological tasks and orallanguage for each of the four age groups in the lower-incomesample are shown in Table 7 and Table 8. Correlationsbetween tests for the 2-year-olds and 3-year-olds were lowand nonsignificant (average r = .02 and - .07 for the 2- and3-year-olds, respectively). The average correlation betweenthe tests was higher for the 4-year-olds (average r = .10)and 5-year-olds (average r = .45). An examination of thefour tests as single items in a phonological processing test

battery either yielded low alphas or could not be calculatedfor the three younger age groups but was high for the oldestchildren (as = .14, .16, and .70 for the 2-, 4-, and 5-year-olds, respectively). Similar to the middle-income sample,correlations between phonological sensitivity and oral lan-guage were higher for older children and tended to bestrongest for the nonoddity tasks. Scores on the blending andelision tasks also tended to be more strongly related to eachother and to the oddity tasks in the older children. Separatecorrelations for the items that involved compound words anditems that involved nonword syllables or phonemes on theblending and elision tasks were not different from those forthe full measures with the rhyme and alliteration scores.

The lower-income sample was used to determine thetest-retest reliability of the phonological sensitivity tasks.The four phonological tasks were readministered approxi-mately 6 weeks after the initial administration. Test-retestreliability for each of the four tasks was moderate for thecombined 4- and 5-year-old groups (rs = .51, .30, .53, and.53 for the rhyme oddity, alliteration oddity, blending, andelision tasks, respectively; all ps < .05.). There was noevidence of test-retest reliability for the combined 2- and3-year-old groups (rs = .03, - . 0 3 , .18, and .14 for therhyme oddity, alliteration oddity, blending, and elision tasks,respectively; all/?s > .44).

Task Analysis

To determine whether the younger children's performanceon the oddity tasks was due to the task demands or to level ofphonological processing abilities, we administered twoadditional tasks to a subsample of the younger children. Thefirst alternative task was a rhyme matching task that wasadministered to 110 of the 2- through 4-year-olds from themiddle-income sample. This task was used to provide aneasier task that assessed rhyme identification without thepotential confusion caused by having to select the odd (i.e.,nonrhyming) word. Estimates of internal consistency for thistask (a = .07, a = .33, and a = .66 for the 2-, 3-, and

5 Because of zero variance at higher levels of linguistic complex-ity in some ages (see Table 2), correlations were not calculated forthe separate age groups in the lower income sample. For ages withnonzero variance, results were similar to the full sample results.


Table 8Correlations Between Tests of Phonological Sensitivity and Language Standard Scoresfor 4-Year-Old (Above Diagonal) and5-Year-Old Children (Below Diagonal)in Lower-Income Sample

1.2.3.4.5.6.

Task

Rhyme oddityAlliteration oddityBlendingElisionReceptive languageExpressive language

1

—.22.53*.40

-.25.03

2

.35*—.55*.15.07

-.15

3

-.05-.05

—.69**.47t.25

4

.17

.11

.04—

.54*

.33

5

.00-.06

.04

.13—.55*

6

.14- .04

.26

.41**

—

Note, n = 38 for 4-year-olds; n =•\p < .10 (marginally significant).

15 for 5-year-olds.*/><.O5. **/><.01. ***/><.001.

4-year-olds, respectively) were slightly higher than those ofthe rhyme oddity task (see Table 3). Correlations betweenthe rhyme matching task and the other nonrhyme tasks werehigher than the correlations between the rhyme oddity taskand the other nonrhyme phonological tasks for both therestricted sample (i.e., only the children who were adminis-tered the rhyme matching task) and the full sample (seeTable 9); however, these differences were not significant.Rhyme matching scores were more highly correlated withlanguage scores than were rhyme oddity scores, and severalof these differences were significant (i.e., receptive andexpressive for 2-year-olds, receptive for 4-year-olds). AnANCOVA controlling for language scores revealed a stronglinear effect of age of the rhyme matching task, F(l, 105) =13.25, p < .001. Follow-up contrasts revealed that the2-year-olds (M = 5.47, SD = 1.72) scored lower than boththe 3-year-olds (M = 6.90, SD = 1.89) and the 4-year-olds(M = 7.46, SD = 2.44), but the 3-year-olds did not scorelower than the 4-year-olds (p < .01 for significant con-trasts). Children in each age group scored above chancelevel on the rhyme matching task (13.3%, 40.4%, and 53.6%of the 2-, 3-, and 4-year-olds scored above chance levels; 8or higher by binomial probability zip = .05), and age wassignificantly related to above-chance performance, x2(2,N = 110) = 10.82, p = .004; p = .31. Similar to the othertasks, even a small percentage of the 2-year-old childrenexhibited some level of phonological sensitivity on therhyme matching task.

The second alternative task was a category oddity taskthat determined whether children's performance was signifi-cantly affected by the nature of the rhyme and alliteration

oddity tasks (i.e., having to select the odd item in a set). Asubsample of 38 children, 21 of whom were from the lower-income sample, (age: M - 51.5 months, SD = 9.4) com-pleted the category oddity measure. The internal consistencyof the measure was high (a = .82), and performance ap-proached ceiling for the majority of children (M = 8.3,SD = 2.8). Only 21% of the children correctly identifiedfewer than 6 of the odd category items. In contrast, themajority of these children could not correctly identify theodd rhyme item (83% scored 6 or lower) or the oddalliteration items (100% scored 6 or lower). Scores on thecategory oddity task were not related to scores on the rhymeoddity task (r = .17, p = .33) or the alliteration oddity task(r - - .05 , p = .79). These results suggest that children'sperformance on the rhyme oddity and alliteration odditytasks was not related to the nature of the oddity task butinstead reflected phonological sensitivity.

Relation to Reading

Partial correlations between scores on the four measuresof phonological sensitivity, letter knowledge, and scores onthe text reading task are shown in Table 10 for the subgroupof 115 four- and five-year-old children from the middle-income sample and the subgroup of 42 children from thelower-income sample who were administered the letterknowledge and reading measures. All but one child in themiddle-income sample knew at least some letter names(M = 19.81, SD = 7.38; range = 0-26) and 38% could readat least one word {M = 1.09, SD = 2.16; range = 0-11).

Table 9Correlations Between Rhyme Matching Task and Other Measures of PhonologicalSensitivity and Standardized Language Scores at Different Agesin Middle-Income Sample

Age group

2-year-olds (n = 30)3-year-olds (n == 52)4-year-olds (n - 28)

Rhymeoddity

.06-.11

.44*

Alliterationoddity

- .10.01.07

Blending

.08•24f.45*

Elision

.07

.35*

.47*

Receptivelanguage

.40*

.06

.38*

Expressivelanguage

.41*

.21

.33t

tp < .10 (marginally significant). *p < .05.


None of the children from the lower-income sample couldread any of the words, but 62% could identify some letters(M = 4.29, SD = 6.50; range = 0-25). Partial correlations(controlling for age) indicated that scores on all of thephonological sensitivity measures, including scores for thedifferent levels of linguistic complexity on the blending andelision tasks, were significantly related to letter knowledgeand the ability to read words in the middle-income sampleand to letter knowledge in the lower-income sample. In themiddle-income sample, the ability to blend phonemes wasmore strongly related to word reading than the ability toblend words, f(l 12) = 2.55, p < .05, but there were no othersignificant differences between the correlations of phonologi-cal sensitivity with word reading or letter knowledge. Forthe middle-income sample, the significant relation betweenphonological sensitivity and word reading remained evenafter controlling for age, receptive language, expressivelanguage, and letter knowledge. There were no significantdifferences for these partial correlations between wordreading and scores on the blending or elision tasks at thedifferent levels of linguistic complexity (allps > .10).

Discussion

Overall, these results indicate that as children increase inage, their performance both increases in absolute level andbecomes more stable across situations in which phonologi-cal sensitivity is required. As expected, very young chil-dren's performance on tasks designed to assess phonologicalsensitivity was relatively low. Although their average perfor-mance was low, there was evidence that a number of the2- and 3-year-old children in this study demonstratedphonological sensitivity at all levels of linguistic complex-ity. Approximately one quarter of the 2-year-old childrenand slightly more than one third of the 3-year-old childrenperformed above chance levels on one of the rhyming tasks.

Additionally, a moderate percentage of the 2- and 3-year-oldchildren were able to perform some of the blending andelision items, including items at the nonword syllable andphoneme level. These results replicate and extend thosereported by MacLean et al. (1987). In their sample, 21% of3-year-olds performed above chance on their rhyme odditytask, and 38% scored above chance on their alliterationoddity task. To our knowledge, the present study is the firstdemonstration of phonological sensitivity in children youn-ger than 3 years of age on tasks designed to measure theseabilities explicitly.

The results of this study demonstrate a strong developmen-tal trend in performance on phonological sensitivity tasks inchildren 2 to 5 years of age from middle-income families.The pattern of results indicates that the development ofphonological sensitivity is not a continuous function repre-sented as a steady increase in ability with increasing age.Although trend analyses indicate that there were lineareffects of age on the tasks, significant higher order trendssuggest that growth in phonological sensitivity was acceler-ated in the older groups. There were substantial increases inchildren's abilities to perform the rhyme oddity, blending,and elision tasks between 3 and 4 years of age. On theblending and elision tasks, these changes were evident at alllevels of linguistic complexity (i.e., words, syllables, pho-nemes). In contrast to results for the middle-income sample,children in the lower-income sample did not appear toexperience this same increased rate of growth of phonologi-cal sensitivity.

The developmental trends in the present study might beexplained in several ways. It could be that children'sphonological sensitivity truly improves as they grow olderfor a variety of reasons such as physical, linguistic, andcognitive maturation, or because of the cumulative effects,or changing quality, of their home literacy, reading, or

Table 10Partial Correlations Between Measures of Letter Knowledge and Text Readingwith Measures of Phonological Sensitivity

Middle-income sample (n = 115)Lower-income

sample (n = 42)

Task or component

Rhyme oddityAlliteration oddityBlending

Total scoreWord-level itemsSyllable-level itemsPhoneme-level items

ElisionTotal scoreWord-level itemsSyllable-level itemsPhoneme-level items

Letter knowledge*

.18f

.19*

.23*

.21*

.15

.23*

.17f

.19*

.09

.15

Texta

.20*

.27**

.29**

.22*

.24**

.38***

.33***

.30*

.23*32***

Textb

.10

.18t

.21*

.13

.18t

.26**

.26**

.23*

.18f

.25**

Letter knowledge*

-.05- .10

.41**

.40**

.40**

.01

.19

.12

.19

.32*aPartial correlations controlling for chronological age. bPartial correlations controlling for chrono-logical age, receptive language standard score, expressive language standard score, and letterknowledge.tp < .10 (marginally significant). *p < .05. **p < .01. ***p < .001.


language environments. It is also possible that this growth isa function of cognitive maturity unrelated to phonologicalsensitivity that allows children to understand and performthe specific tasks more efficiently. A strong version of thislatter possibility is made less likely by the findings thatchildren could perform at a high level on a nonphonologicaloddity task and that performance on this task was unrelatedto performance on the phonological oddity tasks.6 It is alsounlikely that the developmental trends are explained com-pletely by cognitive maturation. Wagner and colleagues(Wagner et al., 1987; Wagner, Torgesen, Laughon, Sim-mons, & Rashotte, 1993; Wagner et al., 1994) demonstratedthat phonological processing abilities are distinct fromgeneral cognitive ability.

Several prior studies have reported a relation betweenchildren's phonological sensitivity and their language abili-ties (e.g., Bowey, 1994; Chaney, 1992; Smith & Tager-Flushberg, 1982; Wagner et al., 1993,1994). Consistent withthese findings, Fowler (1991) suggested that the develop-ment of phonological sensitivity reflects development of theunderlying phonological representation of the lexicon. Asmore words have to be represented, phonological representa-tion becomes more efficient than other forms of representa-tion. Although it is clear that higher levels of phonologicalsensitivity (i.e., phonemic sensitivity) require more thanlexical development, as evidenced by studies of adults andchildren who cannot read and lack phonemic sensitivity butwho have apparently normal lexical skills (e.g., see Adams,1990), it is possible that early lexical development sets thestage for the development of lower levels of phonologicalsensitivity. The finding in this study that oral languagescores were significantly correlated with phonological sensi-tivity scores in the middle-income sample lends support tothis hypothesis; however, the lower correlations betweenoral language and phonological sensitivity in the youngerchildren suggest that more than oral language skills arerequired for the emergence of phonological sensitivity evenat lower levels of linguistic complexity.

Estimates of internal consistency and correlations be-tween tasks from the present study indicate that phonologi-cal sensitivity as a construct becomes more cohesive aschildren mature and as their level of performance on tasksdesigned to measure phonological skills increases. Thepresent study demonstrated that children as young as fouryears of age begin to show stability in their phonologicalsensitivity across different tasks and across time (as indi-cated by the test-retest correlations in the lower-incomesample). In contrast, younger children vary more in theirperformance within and across different phonological sensi-tivity tasks. These findings are consistent with the results ofWagner et al. (1994), who reported that the correlationsbetween blending and elision increased from kindergarten(r — .53) to second grade (r = .65). Estimates of reliabilityand correlations between phonological sensitivity tasks forthe 5-year-old children in this study were slightly lower thanthe estimates reported by Stanovich et al. (1984) or Yopp(1988); however, the children in their studies were approxi-mately one year older than the oldest group of children inthis study. The lower estimates of internal consistency on the

oddity measures likely reflects the chance element in thesetasks that may result in increased error variance. That is, lowscores on the oddity measures may reflect either a low levelof phonological sensitivity or chance responding; conse-quently, these tasks may be relatively insensitive to theearliest stages of phonological sensitivity.

Children in the lower-income sample generally performedsubstantially lower than children from the middle-incomesample. Consistency within and between tasks was lower inthe lower-income sample than in the middle-income sample;however, consistency within and between tasks appeared tofollow the levels of absolute scores on the measures for bothgroups. Although MacLean et al. (1987) and Raz and Bryant(1990) did not find social class differences in preschool-agechildren once variance due to general cognitive or linguisticabilities was controlled, both Raz and Bryant and Bowey(1995) reported significant social class differences in groupsof 5- and 6-year-old children even after controlling forcognitive and language abilities. The present results indicatethat substantial social class differences in phonologicalsensitivity are present from an earlier age. Consistent withprevious studies, children from lower and higher SESgroups differed on rhyme and alliteration oddity tasks at 5years of age. Children from lower and higher SES groupsdiffered in raw scores on the blending and elision tasks from4 years of age. Examination of the percentage of childrenfrom each group who were able to complete phonologicalsensitivity tasks at different levels of linguistic complexityrevealed SES differences at 3 years of age. These differencesdid not appear to be the results of differences in thechildren's language abilities because differences remainedafter controlling for oral language. Moreover, despite thefact that oral language scores were lower for the lower SESsample at age 2 and 3 years, there were no differences inmean phonological sensitivity scores at these ages.

There are at least two possible reasons for the discrepancybetween studies examining social class differences. The firstreason concerns the tasks used to measure phonologicalsensitivity. For example, Bowey (1995) found that socialclass differences on a measure of rhyme oddity, but not onmeasures of phoneme and sound identity, were reduced aftercontrolling for performance IQ and language. In the presentstudy, the earliest SES differences were obtained on theblending and elision tasks. Given the generally poor reliabil-ity of the oddity tasks, it is possible that earlier differencesare obscured by error at low levels of performance. Thesecond reason concerns the different levels of social classinvestigated. In the present study, the families of children inthe lower-income sample were significantly disadvantagedin terms of both family income and parents' educationalachievement relative to the families of children in themiddle-income sample. All of the families in the lower-income sample were eligible for at least one form of public

6 Although the category oddity task does not rule out thepossibility of involvement of phonological memory as a significantfactor in children's performance on the phonological odditymeasures, the use of pictures to represent all words makes thememory component of the phonological oddity measures small.


assistance (i.e., subsidized child care), and many weresingle-parent households. In contrast, the majority of fami-lies in the middle-class sample consisted of two-parenthouseholds in which both parents were employed and bothhad completed at least some college. All families in Bowey'ssample and all but five families in MacLean et al.'s (1987)sample were two-parent households. It is likely that largersocial class differences result in larger group differences onmeasures of phonological sensitivity.7

The reasons for social class differences on tasks designedto measure phonological sensitivity are not clear; however,Raz and Bryant's (1990) results indicated that social classdifferences in phonological sensitivity were related to laterdifferences in word decoding skills. The results of thepresent study indicate that social class differences are smallor absent in very young children; however, differencesbetween the middle-income and lower-income samplesincreased across the ages included in this study as demon-strated by the significant age X SES group interactions forthree of four phonological sensitivity measures. One pos-sible explanation for these apparently different developmen-tal trajectories involves the types and frequency of exposureto literacy-related activities to which children are exposed. Alarge literature has documented differences in book owner-ship, shared-reading frequency, and other potentially instruc-tive literacy activities between higher and lower social classgroups (e.g., Feitelson & Goldstein, 1986; Heath, 1982;McCormick & Mason, 1986; Raz & Bryant, 1990; Teale,1986; Whitehurst & Lonigan, 1998). For example, McCor-mick and Mason (1986) reported that 47% of their sample ofpublic aid parents reported no alphabet books in the home incontrast to only 3% of their sample of professional parents.Evidence suggests that the development of phonologicalsensitivity is both a cause and a consequence of learning toread (e.g., Perfetti, Beck, Bell, & Hughes, 1987; Wagner etal., 1994), and several recent studies have reported a relationbetween higher levels of phonological sensitivity and pre-school-age children's knowledge of the alphabet (e.g.,Bowey, 1994; Johnston, Anderson, & Holligan, 1996; Stahl& Murray, 1994; Wagner et al., 1994). Consistent with thesefindings, Murray, Stahl, and Ivey (1996) demonstrated thatexposure to alphabet books with letter-sound informationresulted in more gains in phonological sensitivity thanexposure to either alphabet books without letter-soundinformation or story books. These results suggest thatmovement from lower to higher levels of phonologicalsensitivity may require either informal or formal instructionin print and alphabet knowledge, situations that are lesscommon for younger children and children from lower-income families.

Results of the present study suggest that lower levels ofphonological sensitivity (i.e., syllables) are developmentalprecursors to higher levels of phonological sensitivity (i.e.,phonemes). Although a moderate percentage of the youngerchildren in this study could manipulate speech sounds inword-level syllables, fewer could perform similar operationson intrasyllabic units, nonword syllables, or phonemes. Incontrast, many of the older children could manipulatespeech sounds at the word, syllable, and phoneme levels.

The fact that children's abilities to perform blending orelision on the word-level syllables were correlated with theirabilities to perform blending or elision at the nonwordsyllable level and phoneme level even after taking intoaccount oral language ability indicates that items at thesedifferent levels of linguistic complexity tap the same phono-logical processing ability. Moreover, the ability to performthese operations at the different levels was correlated withsensitivity to intrasyllabic-level units measured by rhymeand phoneme-level units measured by alliteration tasks inthe older children. Of course, strong statements concerningthe status of lower levels of phonological sensitivity (i.e.,word-level syllable on blending and elision) as developmen-tal precursors to higher levels of phonological sensitivity arelimited by the cross-sectional design of the present study;however, the present results indicate that further examina-tion of this possibility is clearly warranted.

Additional evidence that phonological sensitivity tasksacross the dimension of linguistic complexity tap the sameunderlying phonological processing ability was provided bythe relations between phonological sensitivity tasks atdifferent levels of linguistic complexity and other reading-related abilities. Children's scores on all phonologicalsensitivity tasks were correlated with letter knowledge, andthis relation was consistent across the different levels oflinguistic complexity in the blending and elision tasks forthe middle-income sample. Limited variance at the higherlevels of linguistic complexity precluded comparison acrosslevels in the lower-income sample. These findings areconsistent with results of other studies that have foundchildren's letter knowledge to be related to sensitivity tophonemes (e.g., Bowey, 1994; Johnston et al., 1996; Stahl &Murray, 1994); however, these findings also indicate thatsensitivity to syllables (including word-level syllables) andintrasyllabic units are similarly associated with letter knowl-edge. Phonological sensitivity at both higher and lowerlevels of linguistic complexity (i.e., syllables, onset-rime,phonemes) was related to text reading abilities, and theassociation between phonological sensitivity and text read-ing was not due to general language skills or letter knowl-edge. These results call into question the special status inrelation to reading generally afforded to phonemic sensitiv-ity, at least for young children.

The results of this study as well as others (e.g., Bryant etal., 1990; Chaney, 1992; Fox & Routh, 1975; MacLean etal., 1987; Wagner et al., 1987) indicate that it is possible tomeasure some level of phonological sensitivity in preschool-age children. The fact that Bryant et al. (1990) found thatyoung children's performance on rhyme and alliterationtasks predicted later reading performance, as well as find-ings of concurrent relations between reading and measuresof lower levels of phonological sensitivity in this study,

7 Because social class and race were confounded in this sample,it is possible that the differences between the lower-income andmiddle-income samples reflected racial differences; however, theabsence of early differences and a likely mechanism for theemergence of later differences (e.g., print exposure) suggests thatsocial class provides the more parsimonious explanation.


suggests that the tasks used in the present study will alsopredict the children's later reading achievement. Consistentwith this expectation, Burgess, Lonigan, Anthony, andBarker (1996) reported that scores on the rhyme oddity,alliteration oddity, blending, and elision tasks used in thepresent study were significant predictors of 4- and 5-year-old children's word reading abilities one year later.

It is clear that additional work on the assessment ofphonological sensitivity in preschool-age children is re-quired. Although the results of this study demonstrated that anontrivial number of 2- and 3-year-old children exhibitedsome degree of phonological sensitivity, there were sizablefloor effects on all of the tasks. The rhyme matching task hadthe least severe floor effect; however, it is likely that ceilingeffects would occur with older children (e.g., Stanovich etal., 1984). In the present study, these floor effects somewhatlimit the conclusions that can be drawn. Floor effects mighthave reduced the correlations between tasks by masking trueabilities, and they may have resulted in lower estimates ofreliability because of reduced variance and increased errordue to true but low abilities being masked below the chancelevel of responding. Floor effects may also have masked truedifferences in phonological sensitivity abilities between ageand SES groups. That is, the absence of significant differ-ences between the middle-income and lower-income groupsat younger ages and between the 2- and 3-year-olds mayhave reflected the relative insensitivity of the tasks, ratherthan an absence of ability differences.

Because individual differences in phonological sensitivityappear to be relatively stable from an early age, efforts toidentify the origins of phonological sensitivity and screeningof children for phonological sensitivity deficits are likely tobe most productive during the preschool years. The presentstudy suggests that children in the upper end of thedistribution of phonological sensitivity abilities can beidentified relatively early; however, more sensitive tasks willbe required to adequately differentiate younger children inthe middle and lower end of the distribution. Additionalexamination of the relation between lower levels of phono-logical sensitivity, higher levels of phonological sensitivity,and the development of reading is also required. The resultsof the present study indicate that measures of syllabic,intrasyllabic, and phonemic sensitivity tap the same underly-ing phonological processing ability. Whereas phonemicsensitivity may be the penultimate skill required for transla-tion of the alphabetic code, lower levels of phonologicalsensitivity are both related to phonemic sensitivity andmeasurable in preschool-age children.

References

Adams, M. J. (1990). Beginning to read: Thinking and learningabout print. Cambridge, MA: MIT Press.

Alexander, K. L., & Entwisle, D. R. (1988). Achievement in thefirst 2 years of school: Patterns and processes. Monographs ofthe Society for Research in Child Development, 53(2, Serial No.218).

Bowey, J. A. (1994). Phonological sensitivity in novice readers andnonreaders. Journal of Experimental Child Psychology, 58,134-159.

Bowey, J. A. (1995). Socioeconomic status differences in preschoolphonological sensitivity and first-grade reading achievement.Journal of Educational Psychology, 87, 476-487.

Bradley, L., & Bryant, P. E. (1983). Categorizing sounds andlearning to read—A causal connection. Nature, 301, 419-421.

Bradley, L., & Bryant, P. (1985). Rhyme and reason in reading andspelling. Ann Arbor, MI: University of Michigan Press.

Brady, S., Fowler, A., Stone, B., & Winbury, N. (1994). trainingphonological awareness: A study with inner-city kindergartenchildren. Annals of Dyslexia, 44, 26-59.

Bryant, P. E., MacLean, M., Bradley, L. L., & Grassland, J. (1990).Rhyme and alliteration, phoneme detection, and learning to read.Developmental Psychology, 26, 429-438.

Burgess, S. R., Lonigan, C. J., Anthony, J. L.s & Barker, T. (1996,April). Predictors of the development of emergent literacy skillsin preschool-aged children; Evidence from a longitudinal inves-tigation. Paper presented at the annual meeting of the Society forthe Scientific Study of Reading, New York, NY.

Byrne, B., & Fielding-Barnsley, R. F. (1991). Evaluation of aprogram to teach phonemic awareness to young children.Journal of Educational Psychology, 82, 805-812.

Byrne, B., Freebody, P., & Gates, A. (1992). Longitudinal data onthe relations of word-reading strategies to comprehension,reading time, and phonemic awareness. Reading ResearchQuarterly, 27, 141-151.

Chaney, C. (1992). Language development, metalinguistic skills,and print awareness in 3-year-old children. Applied Psycholin-guistics, 13, 485-514.

Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture VocabularyTest—Revised. Circle Pines, NM: American Guidance Service.

Feitelson, D., & Goldstein, Z. (1986). Patterns of book ownershipand reading to young children in Israeli school-oriented andnonschool-oriented families. Reading Teacher, 39, 924-930.

Fowler, A. E. (1991). How early phonological development mightset the stage for phoneme awareness. In S. A. Brady & D. P.Shankweiler (Eds.), Phonological processes in literacy (pp.97-117) Hillsdale, NJ: Erlbaum.

Fox, B., & Routh, D. K. (1975). Analyzing spoken language intowords, syllables, and phonemes: A developmental study. Journalof Psycholinguistic Research, 4, 331-342.

Gardner, M. F. (1990). Expressive One-Word Picture VocabularyTest—Revised. Novato, CA: Academic Therapy.

Heath, S. B. (1982). What no bedtime story means: Narrative skillsat home and school. Language in Society, 11, 49-76.

Hoien, T., Lundberg, I., Stanovich, K. E., & Bjaalid, I. (1995).Components of phonological awareness. Reading and Writing:An Interdisciplinary Journal, 7, 171-188.

Johnston, R. S., Anderson, M , & Holligan, C. (1996). Knowledgeof the alphabet and explicit awareness of phonemes in preread-ers: The nature of the relationship. Reading and Writing: AnInterdisciplinary Journal, 8, 217-234.

Juel, C , Griffith, P., & Gough, P. (1986). Acquisition of literacy: Alongitudinal study of children in first and second grade. Journalof Educational Psychology, 78, 243-255.

Kirk, S. A., McCarthy, J. J., & Kirk, W. D. (1968). Illinois test ofpsycholinguistic abilities. Urbana, IL: University of IllinoisPress.

Lenel, J. C , & Cantor, J. H. (1981). Rhyme recognition andphonemic perception in young children. Journal of Psycholinguis-tic Research, 10, 57-67.

Liberman, A. M., Cooper, F. S., Shankweiler, D., & Studdert-Kennedy, M. (1967). Perception of the speech code. Psychologi-cal Review, 74, 431-461.

Liberman, I. Y., Shankweiler, D., Fischer, F. W., & Carter, B.(1974). Explicit syllable and phoneme segmentation in young


children. Journal of Experimental Child Psychology, 18, 201-212.

Lundberg, I., Olofsson, A., & Wall, S. (1980). Reading and spellingskills in the first school years predicted from phonemic aware-ness skills in kindergarten. Scandinavian Journal of Psychology,21, 159-173.

MacLean, M., Bryant, P., & Bradley, L. (1987). Rhymes, nurseryrhymes, and reading in early childhood. Merrill-Palmer Quar-terly, 33, 255-282.

Mann, V. A. (1984). Longitudinal prediction and prevention ofearly reading difficulty. Annals of Dyslexia, 34, 117-136.

Mann, V. A., & Liberman, I. Y. (1984). Phonological awareness andshort-term memory. Journal of Learning Disabilities, 17, 592-599.

McCormick, C. E., & Mason, J. M. (1986). Intervention proceduresfor increasing preschool children's interest in and knowledgeabout reading. In W. H. Teale & E. Sulzby (Eds.), Emergentliteracy: Writing and reading (pp. 90-115), Norwood, NJ:Ablex.

Morais, J. (1991). Constraints on the development of phonologicalawareness. In S. A. Brady & D. P. Shankweiler (Eds.), Phonologi-cal processes in literacy (pp. 5-27). Hillsdale, NJ: Erlbaum.

Murray, B. A., Stahl, S. A., & Ivey, M. G. (1996). Developingphoneme awareness through alphabet books. Reading andWriting, 8, 307-322.

National Assessment of Educational Progress (1991). The 1989-90national assessment of reading and literature. Denver, CO:Author.

Newcomer, P. L., & Hammill, D. D. (1988). Test of LanguageDevelopment—Primary (2nd ed.). Austin, TX: Pro-Ed.

Perfetti, C. A., Beck, I., Bell, L., & Hughes, C. (1987). Phonemicknowledge and learning to read are reciprocal: A longitudinalstudy of first grade children. Merrill-Palmer Quarterly, 33,283-319.

Raz, I. S., & Bryant, P. (1990). Social background, phonologicalawareness and children's reading. British Journal of Developmen-tal Psychology, 8r 209-225.

Share, D. L., Jorm, A. E, MacLean, R., & Mathews, R. (1984).Sources of individual differences in reading acquisition. Journalof Educational Psychology, 76, 1309-1324.

Smith, C L., & Tager-Flushberg, H. (1982). Metalinguistic aware-ness and language development. Journal of Experimental ChildPsychology, 34, 449-468.

Stahl, S. A., & Murray, B. A. (1994). Defining phonologicalawareness and its relationship to early reading. Journal ofEducational Psychology, 86, 221-234.

Stanovich, K. E. (1992). Speculations on the causes and conse-quences of individual differences in early reading acquisition. InP. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Readingacquisition (pp. 307-342). Hillsdale, NJ: Erlbaum.

Stanovich, K. E., Cunningham, A. EM & Cramer, B. B. (1984).Assessing phonological awareness in kindergarten children:Issues of task comparability. Journal of Experimental ChildPsychology, 38, 175-190.

Teale, W. H. (1986). Home background and young children'sliteracy development. In W. H. Teale & E. Sulzby (Eds.),Emergent literacy: Writing and reading (pp. 110-121). Nor-wood, NJ: Ablex.

Torgesen, J. K., & Burgess, S. R. (in press). Consistency ofreading-related phonological processes throughout early child-hood: Evidence form longitudinal-correlational and instruc-tional studies. In J. Metsala & L. Ehri (Eds.), Word recognition inbeginning reading. Hillsdale, NJ: Erlbaum.

Treiman, R. (1992). The role of intrasyllabic units in learning toread and spell. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.),Reading acquisition (pp. 107-143). Hillsdale, NJ: Erlbaum.

Wagner, R. K., Balthazor, M., Hurley, S., Morgan, S., Rashotte, C ,Shaner, R., Simmons, K., & Stage, S. (1987). The nature ofprereaders' phonological processing abilities. Cognitive Develop-ment, 2, 355-373.

Wagner, R. K. (1996, April). Does phonological awareness train-ing enhance children's acquisition of written language? Ameta-analysis. Paper presented at the annual meeting of theAmerican Educational Research Association, New York.

Wagner, R. K., & Torgesen, J, K. (1987). The nature of phonologi-cal processing and its causal role in the acquisition of readingskills. Psychological Bulletin, 101, 192-212.

Wagner, R. K., Torgesen, J. K., Laughon, P., Simmons, K., &Rashotte, C. A. (1993). The development of young readers'phonological processing abilities. Journal of Educational Psy-chology, 85, 83-103.

Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1994).Development of reading-related phonological processing abili-ties: New evidence of bidirectional causality from a latentvariable longitudinal study. Developmental Psychology, 30,73-87.

Whitehurst, G. J., & Lonigan, C. J. (1998). Child development andemergent literacy. Child Development, 69, 848-872.

Yopp, H. K. (1988). The validity and reliability of phonemicawareness tests. Reading Research Quarterly, 23, 159-177.

(Appendix follows)


Appendix

Test Items Used in Different Phonological Sensitivity and Control Tasks

A. Items used for blending task

Item presentation

Practice items1. cow — boy2. tooth — brush3. fire — man

Items administered with pictures1. foot —ball2. shoe — string3. star —fish4. sand — box5. doll — house6. wheel — chair7. finger — nail8. rain — bow9. horse — shoe

10. mom — N11. bro — ther12. /h/ — /a;/ — N

Items administered without pictures13. base —ball14. air — plane15. sea — shell16. jump — rope17. light —bulb18. dad —/ i /19. sis — ter20. Pol — ivl — N21. /g/ —/ou/22. ZtJ — hjJ

Answer

cowboytoothbrushfireman

footballshoestringstarfishsandboxdoll housewheelchairfingernailrainbowhorseshoemommy (mommy, brother, keys)a

brother (bear, brother, bird)hat (bat, house, hat)

baseballairplaneseashelljumpropelightbulbdaddysisterbatgotoy

B. Items used for elision task

Answer

batmat

sundogcardoorstove

Item presentationPractice items

1. Say batman. Now say batman without saying man.2. Say doormat. Now say doormat without saying door.

Items administered with pictures1. Say sunshine. Now say sunshine without saying shine.2. Say watchdog. Now say watchdog without saying watch.3. Say streetcar. Now say streetcar without saying street.4. Say doorbell. Now say doorbell without saying bell.5. Say stovetop. Now say stovetop without saying top.6. Say carpet. Now say carpet without saying pet. (car, pet, cat)3 car7. Say candy. Now say candy without saying dee. (cup, can bear) can8. Say rattle. Now say rattle without saying til. (rake, duck, rat) rat9. Say time. Now say time without saying /m/. (tie, line, tree) tie

Items administered without pictures10. Say orange juice. Now say orangejuice without saying orange, juice11. Say playground. Now say playground without saying ground, play12. Say stop sign. Now say stopsign without saying stop. sign13. Say target. Now say target without saying get. tar14. Say dandy. Now say dandy without saying dee. dan15. Say battle. Now say battle without saying til. bat16. Say dime. Now say dime without saying /m/. die17. Say heat. Now say heat without saying /t/. he

DEVELOPMENT OF PHONOLOGICAL SENSITIVITY

Appendix (continued)

311

Practice 1.Practice 2.

1.2.3.4.5.6.7.8.9.

10.11.

C. Items used for category oddity task*

redtriangleyellowovalheartmoonwestapplecatshirtcarplanesock

redsquareorangerectangleheartsunwestbananadogpantscarbirdshoe

greentriangleorangerectanglestarmoonnorthappledogshirtcowplaneshoe

"Items in parentheses in blending and elision tasks refer to names of pictures shown to children inmultiple-choice format to maintain pictured presentation of items. bItems with identical nameswere represented by a different representation of the category.

Received January 28,1997Revision received October 21, 1997

Accepted October 30, 1997

Low Publication Prices for APA Members and Affiliates

Keeping you up-to-date. All APA Fellows, Members, Associates, and Student Affiliatesreceive—as part of their annual dues—subscriptions to the American Psychologist andAPA Monitor. High School Teacher and International Affiliates receive subscriptions tothe APA Monitor, and they may subscribe to the American Psychologist at a significantlyreduced rate. In addition, all Members and Student Affiliates are eligible for savings of upto 60% (plus a journal credit) on all other APA journals, as well as significant discounts onsubscriptions from cooperating societies and publishers (e.g., the American Association forCounseling and Development, Academic Press, and Human Sciences Press).

Essential resources. APA members and affiliates receive special rates for purchases ofAPA books, including the Publication Manual of the American Psychological Association,and on dozens of new topical books each year.

Other benefits of membership. Membership in APA also provides eligibility forcompetitive insurance plans, continuing education programs, reduced APA convention fees,and specialty divisions.

More information. Write to American Psychological Association, Membership Services,750 First Street, NE, Washington, DC 20002-4242.

development of phonological sensitivity in 2- to 5-year...

Documents