Monday, April 25, 2011

Defining Phonological Awareness and Its Relationship to Early Reading ~ Stahl, S. & Murray, B (1994)

Journal of Educational Psychology, 86(2), 221-234.

Purpose of the Study

Twofold: Examine the importance of "linguistic complexity and task differences in measuring phonological awareness." Examine the relationship of PA to early reading skills in according to PA tasks.


Early studies (Liberman et al, 1974) children needed to be able to reflect on sounds in words to later map the associated graphemes. Children without this knowledge have difficulty learning to read (Savin, 1972; Stanovich, 1986) Correlational evidence that is both concurrent and predictive (Adams, 1990 review) is extensive (Maclean, et al., 1987; Perfetti et al., 1987).

Adams also examined studies that suggested letter knowledge and PA are the strongest predictors of reading acquisition.  Lomax and McGee (1987) studied children aged 3-6 and noted a developmental sequence from concepts of print (CP), graphic awareness, phoneme awareness, grapheme-phoneme (G-P) correspondence, and word reading. CP factor was assoicated with G-P correspondence. CP measured knowledge of print use, vocabulary of print (word, letter, sentence), and behavior with print (tracking left to right, top to bottom, etc.). PA assessed on 3tasks--deciding if two words were the same or different words, isolating initial consonant, isolating final consonant. Yopp (1988) referrs to isolation tasks as simple phoneme awareness.

Training Studies

Bradley & Bryant (1983) grouped children into a sorting group (Hawthorne control), a sorting with letters (Experimental), and an untreated control. Prereaders in the experimental group were 9 months ahead in reading skills compared to the H group and 12.5 ahead of the Control.  PA and letter training facilitated spelling. PA also had a significant effect on early reading without letter training. Lundberg (1988) found that PA training of kindergarteners without letters led to gainis in PA and significantly effected spelling and reading achievement in 1st and 2nd grade.

Some researchers believe there is reciprocal causation between PA & reading achievement - Ehri & Wilce (1986) determined that children who could read responded differently on phoneme awareness tasks that involved identifying /t/ or /d/ in a PA task. Suggests that orthographic knowledge influenced choices. Levels of PA are likely the basis for reciprocal causation since some tasks can be done without print knowledge while others are advanced, requiring more sophisticated understanding of identities and orthography.


Demographics: 113 K-1 children in public and parochial schools participated. Mixed SES & racial background. Even boys to girls.


  • Tests of phonological awareness (researcher constructed for 4 tasks - blending, segmenting, phoneme isolation, phoneme deletion; 4 levels of word complexity - onset-rime CVC, vowel-coda CVC, cluster onset CCVC, and cluster coda CVCC.)
  • Written language measures 
    • Clay letter recognition
    • Johns informal reading inventory (1991) adequate retell rather than questions for comp measure
    • Environmental print logo reading
    • Spelling measure (Tangel & Blachman, 1992)
  • Memory measure
    • Digit span subtest of the WISC-R (1974) 
  • Procedure
    • Children tested individually (30-40 min) for logos, PA measures, word identification, IRI, & digit span.
    • Children rewarded with a gift
    • Spelling test administered in small groups at another time (0nly 85/113 participated in spelling test)

Phoneme isolation was the easiest task followed by blending, deletion, and segmentation. (All statistically significant differences between each other)

Easiest linguistic level was analyzing onsets and rimes, followed by vowels and codas, then cluster codas and cluster onsets (all statistically significant differenes between measures.)

61% of errors were in manipulating cluster onsets or codas for isolation, segmentation and deletion tasks. Nasal blends (nk, nd, mp) and liquid blends (ld) were easies to break up. 

Many children dropped the final consonant due to Southern dialect influence. 
Both ways of defining PA, by task or level of complexity, yielded a single common factor, however, the notion of levels of complexity accounted for more variance as the common factor.

PA & Reading outcomes were affected by children's prior knowledge of the alphabet and the fact that a number of the participants could not read any words. The distribution was highly skewed. However, it was found that the ability to separate an onset from a rime seemed to aid children with word recognition. Also, the ability to recognize words might aid children's ability to analyze rimes into vowel and coda, while the ability to break up a rime and word recognition were correlated as statistically significant.

Task differences - only one of the tasks had a similar relationship with reading/ Phoneme isolation, the easiest task, distinguished children who could rad from children who could not. Children could not break up blends but recognized them as single units. 

PA and spelling - there were strong correlations between PA and spelling onset-rime awareness and spelling, e.g., (p<.01). Findings suggested that a spelling task is an easier measure of PA than oral PA measures.

Possible explanation: invented spelling for children who know letter names, minimizes the need to memorize, compared with oral PA tasks. Children can spell the remainder of a word when deleting an onset, for instance.


The notion of levels of linguistic complexity rather than difficulty of task, seems to better describe phonological awareness. The ability to manipulate onsets and rimes in syllables relates strongly to reading with adequate letter recognition skills. Isolating a phoneme from the beginning or end of a word, the easiest PA task, seems crucial to reading concluded from the evidence that nearly all the children who could not complete this task did not achieve preprimer instructional level on the IRI. 

Knowledge of letter names help children manipulate onsets and rimes, enabling word recognition. Basic word recognition possibly aids higher levels of PA. Cite's similar results in Barron, 1991. Adams suggested children may learn letter names from alphabet books or the singing ABCs, then learn to match letters with sounds. Consonants contain the phonemes associated with them with some exceptions, so letter names may provide clues in identifying their sound value. Separating an onset from a rime, however, requires more than these two skills.

Phonetic cue reading is considered a stage in reading development when the initial or final consonant provided information to help identify a word. Stahl and Murray suggest that this is a "precursor to developing a rudimentary sight vocabulary." The ability to read well is always evident among children who can decode pseudowords.

Learning about phonemes introduces a child to a new way of thinking about language, providing insight into the alphabetic principle.


The range of reading abilities in this sample may be representative of the population but it makes it difficult to confirm outcomes as valid since some had rather sophisticated knowledge of written language while others barely knew their letters and no correspondences. There was no treatment, just all examination of correlations, so there cannot be any conclusions about causal relations. It does suggest further study into looking at tasks or linguistic complexity as we refine the construct of PA.

No comments:

Post a Comment