Issues Involved in Defining Phonological Awareness and Its Relation to Early Reading
Steven A. Stahl
The University of Georgia
Bruce Murray
Auburn University
The first exposure to the concept of phonological awareness for many educators and psychologists came from the publication of Language by Ear and by Eye, edited by James Kavanaugh and Ignatius Mattingly in 1972. In this volume, Mattingly, Harris Savin, and Donald Shankweiler and Isabelle Liberman discussed the relation between an awareness of phonological segments (then called linguistic awareness) and learning to read. Shankweiler and Liberman’s research drew from the work in speech perception done at the Haskins Laboratories (e.g., Liberman, Cooper, Shankweiler, & Studdert-Kennedy, 1967), which found that spoken words could not be acoustically analyzed into discrete phonological segments because the phonemes, which could be thought of abstractly as separate elements, were “folded” together into units. “Before he can map the visual message to the word in his vocabulary, [the child] has to be consciously aware that the word cat that he knows—an apparently unity syllable—has three separate segments” (Shankweiler & Liberman, 1972, p. 309). Shankweiler and Liberman suggested that difficulties in phonological awareness were at the root of many reading problems.
The basic premise that “phonological awareness is related to reading” has been conclusively established by research in the years since Kavanaugh and Mattingly’s landmark volume. However, we have yet to specify what we mean by the terms of the equation. What kind of “phonological awareness” is related to reading? About what kind of relation are we talking? And how are we defining reading. The purpose of this chapter is to review some of the relevant work related to defining these terms, and to propose a model of how phonological awareness might be related to reading at different stages of the development of both phonological awareness and reading.
DEFINING PHONOLOGICAL AWARENESS
Over the years, phonological awareness has been defined in a number of ways. At first it was defined as a single concept. Researchers created tasks that required what they thought was phonological awareness, assuming that different tasks all reflected general metalinguistic ability. Thus, Liberman (1973) used a tapping task, in which students needed to tap a wooden dowel for every phoneme in a word (see also Liberman, Shankweiler, Fischer, & Carter, 1974); Rosner (1974) had children mentally remove phonemes from a word; and Fox and Routh (1975) had children tell the first sound of a word. All of these tasks require some degree of phonological awareness. With these different tasks, it was little wonder that children appeared to acquire phonological awareness at different ages in different studies!
In retrospect, the tasks varied considerably, involving different processes and different levels of linguistic and metalinguistic knowledge (Stahl & Murray, 1994). In subsequent research, the different tasks were compared empirically, to see which best represents phonological awareness, or whether there were different types of phonological awareness (e.g., Beach, 1992; Yopp, 1988). More recently, phonological awareness has been defined in terms of the content and the processes acting on that content, adding new constructs such as “linguistic complexity” (Treiman, 1985) and “phoneme identity” (Byrne & Fielding-Barnsley, 1989, 1990). Commonly used tasks include:
• Rhyming, either by recognizing rhymes or rhyme production.
• Word-to-word matching tasks, which involve having a child determine whether a series of words begins or ends with the same sound, or which word in a set is the “odd one out.”
• Sound-to-word matching tasks, which ask whether a particular sound, uttered explicitly by the examiner, can be found in a word.
• Full segmentation, requiring a child to articulate separately each phonemic segment in a spoken word or to report the number of segments in a word.
• Partial segmentation (also called phoneme isolation), in which the child is asked to segment one sound. Sometimes this involves separating an onset from a rime (e.g., /k/ /at/) or breaking up a rime (/ka/ /t/).
• Blending—the “flip side” of segmentation—which involves having a child combine sounds that are spoken separately into a word (e.g., recognizing that /k/ /a/ /t/ is cat).
• Deletion and manipulation—in deletion tasks, such as Rosner’s Auditory Analysis Test (1975), a child is told to mentally remove a portion of a word to make another word (e.g., “Say coat. Now say it again without the /k/.”)
In more complex manipulation tasks, children may be asked to remove a phonemic segment and put it elsewhere in the word to make a new word, or to perform other complex manipulations, such as in pig Latin (Savin, 1972). Making sense of the experimental literature about phonological awareness has been muddled by comparisons of these different tasks.
LEVELS OF DIFFICULTY
In a synthesis of the literature on reading acquisition, Adams (1990) theorized that the various tasks used to measure phonological awareness fall into five levels of difficulty. The most primitive level, according to Adams, consists of having an ear for the sounds of words, as revealed by the ability to remember familiar rhymes (e.g., Maclean, Bryant, & Bradley, 1987). A second level consists of the ability to recognize and sort patterns of rhyme and alliteration in words, which requires a more focused attention to sound components; this ability is revealed in oddity tasks (e.g., Bradley & Bryant, 1983). A third level requires familiarity both with the division of syllables into phonemes and with the sounds of isolated phonemes; this level is indicated by blending tasks and syllable-splitting tasks (e.g., isolating initial phonemes). A fourth level of difficulty covers tasks that require full segmentation of component phonemes (e.g., a tapping test; Liberman, 1973). At the fifth and most difficult level are tasks requiring children to add, delete, or otherwise move phonemes and to regenerate the resultant word or pseudoword (e.g., Rosner, 1974).
Simple and Complex Phonological Awareness
Yopp (1988) attempted to resolve empirically the problem of defining phonological awareness. She gave 10 different measures of phonological awareness to a group of kindergartners in order to determine the reliability and relative difficulty of each measure and to assess task validity through correlation with a pseudoword decoding task. Yopp also carried out a factor analysis, which identified two skills influencing test performance: a simple phonemic awareness factor (seen in segmentation, blending, sound isolation, and phoneme counting tests), and a compound phonemic awareness factor (seen in tasks that require holding a sound in memory while performing additional operations). There are some problems with Yopp’s (1988) factor analysis. The first relates to the analysis itself. The first factor found had an eigenvalue of 5.87, accounting for 58.7% of the variance. The second factor had an eigenvalue of 0.94, accounting for an additional 9.5% of the variance, and eigenvalues for other factors drop off slowly, accounting for less and less of the variance. Because the eigenvalue for Yopp’s second factor is beneath the conventional cutoff of 1.0, and because the first factor accounts by itself for the majority of variance, a one-factor solution rather than Yopp’s two-factor solution might be more appropriate. Using somewhat different measures, we found that a single factor, rather than two factors, seemed to best describe our data (Stahl & Murray, 1994). Our basic data looked much like Yopp’s, but, given the eigenvalue of 1.0 criterion, we discounted the second factor. When performances on different tasks (blending, full segmentation, partial segmentation, and deletion) were used for analysis, a single factor accounted for 72.6% of the common variance (eigenvalue = 2.91). When we forced a two-factor solution and attempted various rotations, the first factor was still significantly more pronounced.
Linguistic Complexity
Although Yopp (1988) posited two clearly different levels of phonological awareness, she also noted disparities across the tasks commonly used to assess the construct. Items varied greatly both between and within measures derived from the same type of task. For example, some blending tasks used nonsense words whereas some used real words; some had more short consonant-vowel-consonant (CVC) words, whereas others contained more words with consonant blends.
One important source of variability not controlled in Yopp’s tasks was linguistic level. According to Treiman (1992), syllables seem to break most readily between the onset (any beginning consonants) and the rime (the vowel and any final consonants). The rime is further divided into the vowel nucleus and the coda, or any final consonants. For instance, most people find it easier to divide stamp into /st/ and /amp/ than into other dichotomous parts. This tendency is illustrated by the unintended slips of the tongue (spoonerisms) people construct when they are torn between synonyms by blending the onset of one word with the rime of another (e.g., tons of soil for sons of toil). Treiman (1992) simulated this by asking adults to combine frail and slat into one new word. The majority (62%) responded with frat, which moved the onset of the first word onto the rime of the second; very few responded with frait, which would have required a division within a rime. Because of the significance of onsets and rimes within words, it is more difficult, for example, to delete the initial phoneme in trick than in tick. To delete /t/ from trick involves breaking up the blended phonemes within an onset, but in tick deleting /t/ only requires separating the onset from the rime. Based on Treiman’s notions, it should be easier to break a syllable between an onset and a rime than to separate the coda from the rest of the word, and more difficult still to split cluster onsets such as /sl/ or cluster codas such as /st/ or /mp/.
Yopp used or adapted extant tasks of phonological awareness. Items on the tasks varied considerably in terms of the levels of linguistic analysis required to perform each task. Thus, one cannot be sure whether the performance differences found between tasks resulted from differences in the inherent degrees of task difficulty or from differences in linguistic complexity.
We reexamined the items on Yopp’s (1988) measures. We assigned a weight for each level of linguistic complexity tapped (Stahl & Murray, 1994). Recognition of a rhyme was assigned a value of 1, manipulating onset and a rime a value of 2, manipulating vowel and coda a value of 3, manipulating phonemes within a cluster onset a value of 4, and manipulating phonemes within a cluster coda a value of 5. We rated each item on linguistic complexity and averaged these ratings for each task. When we correlated task ratings with the mean score obtained by Yopp’s subjects on each task, we found a .95 correlation between our post hoc measure of task difficulty and the levels of difficulty obtained by Yopp. This suggested that linguistic complexity may be an important factor in phonological awareness. It also suggested that Yopp’s measures may have confounded linguistic complexity and task.
In our study, we prepared four tasks (blending, full segmentation, partial segmentation, and deletion), each with an equal number of items at different levels of linguistic complexity (Stahl & Murray, 1994). Our design was such that it allowed us to examine the differential effects of both task and linguistic complexity. When we analyzed our measures by linguistic complexity across the different tasks, we found support for our hypothesis that onset-rime separation was easier than breaking up a rime by segmenting vowel and coda, which in turn was easier than breaking a cluster onset or rime into constituent phonemes.
As mentioned earlier, when we used tasks to define phonological awareness, a single factor accounted for 72.6% of the common variance. We also conducted a factor analysis with levels of linguistic analysis used to define phonological awareness. The scores at the various levels, summed across tasks, were used in the analysis. In this analysis, one factor accounted for substantially more of the common variance, 81.7% (eigenvalue = 3.32). We concluded that defining phonological awareness by levels of linguistic complexity better accounted for our data than defining phonological awareness by task, because when the subtests were analyzed by that grouping, the factor analysis accounted for 9% more of the total variance, even though both approaches accounted for high amounts of the total variance and both would be considered to be adequate.
Both the task approach and the linguistic levels approach tend to view phonological awareness mechanistically. That is, according to these views, the task of phonological awareness involves performing some mental manipulation to a spoken word. In contrast, a third approach to defining phonological awareness, phonemic identity (Byrne & Fielding-Barnsley, 1989), views the task as concept formation.
The germ of the concept of phoneme identity was suggested by Lewkowicz (1980), who tried to explain how children could segment syllables into phonemes despite the linguistic problem of the acoustic unity of a syllable. Lewkowicz speculated that children must look for familiar elements in the stretched pronunciations of syllables. What becomes familiar are the vocal gestures and their accompanying sounds that children come to recognize as recycled across words. These vocal gestures come to have identities (from the Latin idem, meaning “same”) when they are recognized as the same vocal gestures used across different words.
The identity view of phonological awareness has important implications for teaching. In identity-based teaching, one teaches the identities of a small number of phonemes in a variety of words. In segmentation training, one teaches isolating the sounds as a skill that one can perform with any word, rather than with only words whose phonemes were mastered. Byrne and Fielding-Barnsley (1989, 1993) used identity teaching and segmentation teaching with samples of preschoolers. The identity group was taught to recognize only the phonemes /s/ and /m/ in words by isolating and stretching these phonemes. This group received guided practice in reliably selecting words beginning or ending with these phonemes (i.e., sound-to-word matching). The segmentation group received practice in isolating the beginning and ending sounds of the same words, using a larger set of phonemes. Both groups learned letter–phoneme correspondences for the letters s, m, b, and f, which appeared in a reading analog task during posttesting. For example, children given the word mow were asked, “Is this sow or mow?” This task involves phonetic cue reading (Ehri, 1991), because participants need only focus on the first letter sound as a cue to correctly respond. They were also asked to respond to words beginning with f and b, for which phoneme identities had not been explicitly taught. Although the groups were not directly compared, possibly because the small sample size lacked power to distinguish performance differences, the identity group appeared to fare somewhat better in leading children to master the phonetic cue reading task. Furthermore, sound-to-word matching scores (an identity measure) were a better predictor of phonetic cue reading ability for identity participants (r = .49) than segmentation scores were for segmentation-trained participants (r = .20) on the phonetic cue reading measure. Byrne and Fielding-Barnsley also observed informally that children seemed to enjoy the identity work more and that it required little corrective feedback.
Five children from each teaching program mastered phonetic cue reading with the untrained letters f and b, suggesting that these children had acquired a larger sense of phoneme awareness that permitted them to use simple correspondence information to quickly get a handle on the identities of these phonemes and succeed in phonetic cue reading.
Murray (1995) replicated the Byrne and Fielding-Barnsley (1989) study using a larger sample size. Murray found that kindergarten children trained to segment spoken words made greater gains than an identity group on measures of phoneme manipulation, such as blending and partial and full segmentation, but that the children with identity training were better able to transfer their knowledge to a measure of phonetic cue reading. The Byrne and Fielding-Barnsley (1989) and Murray (1995) studies suggest strongly that it is important to include phoneme identity in any definition of phonological awareness.
RELATIONSHIPS AMONG PHONOLOGICAL AWARENESS SKILLS
In our earlier study (Stahl & Murray, 1994), we found a clear hierarchy among three tasks—blending, deletion, and partial segmentation. Partial segmentation (which we termed phoneme isolation) seems to be the first task mastered. A series of scattergrams suggested that partial segmentation, at least of an onset and rime, seems to reliably precede the ability to blend and delete phonemes. That partial segmentation was nearly always present in children who could blend and delete phonemes suggests that it is a necessary condition for these manipulations. The sequence of partial segmentation to blending and deletion is complicated by a number of other factors. First, reading ability seems to mediate blending and deletion. The scattergram analysis indicated that some degree of reading ability, as measured by reading words on a preprimer list, also seems to be precede blending and deletion ability. This minimum level was surprisingly high—into the first reader range for both tasks—suggesting that a fairly high level of word recognition is necessary for children to blend and delete phonemes in spoken words. This confirms the findings of Perfetti, Beck, Bell, and Hughes (1987) with similar tasks. Second, we believe that the construct of phoneme identity seems to underlie the ability to perform all of these tasks.
We did not include phoneme identity tasks in our 1994 study, but a later study (Murray, 1995) included a phoneme identity measure.1 He found that children performed similarly on this measure of phoneme identity as they did on a measure of partial segmentation. From this pattern of results, it is possible that partial segmentation measures may be strongly influenced by children’s ongoing development of phoneme identities.
The importance of linguistic complexity suggests a sequence of learning the identity of phonemes in simpler contexts before locating them in more difficult linguistic contexts. For example, recognizing the phoneme /s/ when it serves as an onset, as in sun, is easier than finding it when it is part of a time, as in gas, which would be easier, in turn, then recognizing it as part of a complex onset, such as in star, or as part of a complex rime coda as in wrist.
WHAT DO WE MEAN BY IS RELATED TO?
There are two relationships we can posit that phonological awareness can have to reading: correlational, and necessary but not sufficient. (“Necessary and sufficient” and “sufficient but not necessary” are logically implausible.) These relationships can be demonstrated using scatter grams. At this point, we do not tell you which variables are illustrated but instead do so later, because here we want to concentrate on the relationships rather than what they mean. The scattergram shown in Fig. 3.1 is a classical correlation pattern. The r here is about .79, which is a strong correlation. Notice that both variables rise together. This could suggest that these two measures are measuring the same thing. Because correlation does not imply causality, it could also mean that a third variable, to which both are related, could be causing the relationship.
In contrast, the scattergram in Fig. 3.2 shows what could be a necessary but not sufficient relationship. In this relationship, there were children who could not do either task very well and students who did both tasks well. There were also students who did the first task well but not the second, but few children who did the second well but not the first. Let’s divide this into quadrants. In a correlation, the coefficient would assess the number of children in quadrant B and quadrant C. In this case, as in many cases reported in the literature, this correlation would be high because most of the children in the study fell into one of those two quadrants. However, the real information is in the other quadrants. One can analyze this statistically (we used McNemar’s Test, a nonparametric statistic), but the picture is informative by itself.
In our work with phonological awareness, this type of pattern was the most common for a number of reasons. First, most variables tended to have ceiling effects. There are only 26 letters of the alphabet, children generally either can or cannot segment a word, and so on. Although most research in this area analyzes these variables as if they have a full range and are normally distributed, these are not variables with a full range and their distribution tends to be skewed. Second, these relationships make theoretical sense. From Liberman’s (1973) work onward, it has been assumed that phonological awareness entered into a causal relationship with reading. There is evidence from training studies, from Bradley and Bryant’s (1983) study to their successors, that phonological awareness or lack thereof does seem to be causally related to success or failure in reading. The scattergram analysis shows this most closely. If two variables enter into a causal relationship, especially if one or both variables would tend to have a skewed distribution, then one would expect to see a “necessary but not sufficient” pattern, as in Fig. 3.2.2
FIG. 3.1. Scattergram illustrating a correlational relationship.
FIG. 3.2. Scattergram illustrating a necessary but not sufficient relationship.
The variables in Fig. 3.1 are the measure of phoneme identity and a partial segmentation measure, discussed earlier. Because these covary, one might assume (within the caveat of correlation not implying causality) that these measures are assessing the same or similar skills. Thus, a partial segmentation measure might be measuring phoneme identity knowledge. The variables in Fig. 3.2 are alphabet knowledge and blending. This suggests, as noted previously, that alphabet knowledge (the y-axis) is necessary but not sufficient for the ability to partially segment words (the x-axis), because there are many subjects who know nearly all letters and can segment, some who know few letters and cannot segment, some who know letters but cannot segment, and one who can blend but do not know at least 50 letters of the alphabet.
In this research, regression and correlation have been used extensively to examine these relationships. However, they may be misleading, because variables do not fit together in ways appropriate for correlational analysis.
HOW DO WE DEFINE READING?
There are a number of ways to define reading, hence the confusion about the goals of reading programs (see Stahl, 1997a). Proponents of different instructional models tend to view the concept of reading in different ways. For example, some people tend to view reading purely in terms of decoding, whereas whole language theorists tend to view reading in terms of responding to and appreciating literature. These differing views have led to the acrimonious debates that have typified our field. When we talk about reading, in whatever context, it is important to state exactly what aspect of reading we are talking about.
In relation to phonological awareness, reading has been defined strictly in terms of word recognition. This is appropriate, because phonological awareness has been posited to underlie a person’s learning of sound–symbol relations or of orthography, which in turn underlies the larger purposes of reading.
We assume that readers go through a series of stages as they learn to recognize words. Although we rely on Ehri (1991) for the names of these stages, similar stages have been proposed by Chall (1996), Frith (1985), Gough, Juel, and Griffith (1992), and McCormick and Mason (1986). In all of these models, the development of word recognition goes through roughly three stages, what Ehri (1991) called visual cue reading, phonetic cue reading, and phonological recoding.
Visual Cue Reading
In visual cue reading, children recognize words by purely visual means. Studies of visual cue reading further indicate that readers do not use visual patterns of whole words as retrieval cues, but rather select specific features of a visual array for this purpose (Gough, Juel, & Griffith, 1992). The reader might select a letter shape, a memorized spelling (e.g., for a name), a distinctive logo, a picture, or the bent corner of a card, and ignore the sequence of remaining letters. For example, Gough et al. (1992) asked 4- and 5-year-old children to learn to read four words presented on flash cards. One card in each child’s collection of words to be learned was soiled with a distinct thumbprint in one corner. The children were trained by standard paired-associate learning methods to a criterion of two successive perfect trials. The card with the thumbprint was always learned first. However, when the same word was presented on a clean card, fewer than half of the children could then recognize it. Moreover, when the children were shown a blank card with a thumbprint, nearly all reported the word they had seen on the original thumbprinted card. The same results occurred when they were shown a different word with a thumbprinted corner.
One explanation for these results is that young children with no means of remembering words other than visual cue reading tend to associate words and their meanings with whatever salient visual cue is available—in this case, the thumbprint. It was the thumbprint, rather than the sequence of letters of the word, that elicited their association during the test trials. In a similar demonstration, preschoolers learned to read four dissimilar four-letter words by the same anticipation and correction method, and then were asked to identify the words from half of the spelling. For example, they were asked to recall pony from the letters po, or from ny. Children who could not identify the word from half the spelling were twice as likely to recognize it when shown the other half. This suggests that visual cue readers select a visual feature from part of the word to make their associations, rather than using the entire visual configuration.
Visual cue reading may be used by children to recognize logos. Whole language advocates suggest that children may learn about print initially through examination of logos, such as the arches for McDonald’s. However, the research of Masonheimer, Drum, and Ehri (1984) and Stahl and Murray (1993) suggests that prereaders process logos as pictures and do not analyze print. Masonheimer et al. found that children were insensitive to alterations of the print within a logo, such as placing xepsi within a Pepsi logo. Stahl and Murray found that kindergarten children’s knowledge of the alphabet predicted their recognition of words taken from logos written in block print significantly better than their recognition of the logos themselves.
Although word representations acquired by visual cue reading tend not to be memorable, reliable, or easily learned, paired association of visual cues with word meanings is the standard default strategy used by untaught beginners in their attempts to recognize words (Byrne, 1992). Ehri and Wilce (1985) found that preschoolers and kindergartners who read no words found it easier to read spellings that were visually distinctive (e.g., yMp for turtle) than spellings that used the sounds values of letters (e.g., GRF for giraffe). Children who could read some words found the less visually distinctive sound spellings easier, and learned them more accurately in fewer trials.
Use of a phonological recoding strategy is possible only with syllabic and alphabetic writing systems, in which spellings map the sequential pronunciations of sounds in spoken words. To profit from the advantage offered by alphabetic orthographies, the beginner must gain insight into the alphabetic principle, namely that spellings in written words map onto phonemes in spoken words. Understanding this principle does not come easily to children (Gough & Hillinger, 1980). Byrne (1992) taught preschoolers to recognize pairs of words that differed only in their initial consonants (e.g., fat and bat; fin and bin). Although most of the children learned to identify these words, they did not learn that the initial letter represented a sound found in other words; when asked to distinguish fun and bun or fell and bell, their responses were at chance levels. In contrast, these preschoolers were able to generalize symbols linked to meanings. For example, after learning to identify little boy and big boy and little truck and big truck, they were above chance in differentiating little fish and big fish. This suggests that young children can generalize symbols that refer to meaning, but they are quite unlikely to generalize symbols that encode phonemes.
Phonetic Cue Reading
Children move slowly from visual cue reading to a full analysis of a word’s spelling. The first step is usually a transitional phase that Ehri (1991) called phonetic cue reading. In phonetic cue reading, the beginner uses some letters in words (typically the initial or boundary letters) to generate one or more sounds in the spoken equivalent of the word, thereby narrowing the ranges of choices for contextual guessing. This reduces the memory burden occasioned by the rote associations of visual cue reading, because it replaces the arbitrary link of selected visual cues with a systematic link through letter–sound associations. For instance, instead of remembering that yellow has “two sticks” in the middle, the beginner can access the phoneme /y/ and guess a color name beginning with that sound, presuming he or she has determined from context that the unfamiliar word names a color.
Phonetic cue reading affects children’s reading and writing in a number of different ways. Phonetic cue readers can readily fingerpoint words in memorized text (Ehri & Sweet, 1991), enabling children to pair spoken and printed words. Their misreadings of words tend to preserve initial or boundary sounds (Biemiller, 1970). Also, phonetic cue readers rarely make semantically related misreadings that are not phonetically related (e.g., “sneaker” for Nike), because they use letter-sound information to monitor their reading (Stahl & Murray, 1993). Instead, phonetic cue readers who cannot recognize a word tend to provide no response instead of producing a semantically acceptable guess (Biemiller, 1970). Finally, children’s early invented spellings tend to preserve these letters or their equivalents (Ehri & Wilce, 1987; Gillet & Temple, 1990).
Disadvantages of Phonetic Cue Reading. Although phonetic cue reading offers a decided advantage over the arbitrary associations of visual cue reading, it is not without its drawbacks. In contrast with more skilled readers who can phonologically recode using the entire spelling, children who read via phonetic cues show inconsistency in reading the same word across trials. In one study these beginners could read only about 35% of the words they had read correctly on the previous trial, less than half the number correctly reread by children who could recode most of the spelling sequence (Ehri & Wilce, 1987). Words with similar spellings (e.g., pots, post, spot, stop) require more learning trials for phonetic cue readers than those with dissimilar spellings (McCutchen & McDowell, 1969). Given these limitations, readers who have taken the qualitative leap from visual to phonetic cue reading must still take the quantitative steps toward fuller analysis and use of complete spelling sequences in phonological recoding.
Phonological Recoding
Another means of accessing the meanings of printed words is phonological recoding; that is, generating an approximate pronunciation of a word from spelling cues to retrieve meaning. (This is also called “cipher” reading; Gough et al., 1992). Phonological recoding serves as a self-teaching mechanism for word recognition as well as a backup mechanism for skilled readers to access less familiar words (Jorm & Share, 1983; Share, 1995). Expert phonological recoding involves generating pronunciations from entire spellings of words. Whatever teaching method is used to lead children into independent word recognition, it is increasingly clear that would-be readers of alphabetic languages must learn to reproduce spoken words by using complete spellings as maps to pronunciation (Adams, 1990; Ehri, 1991; Jorm & Share, 1983).
Children use a number of means to identify words. Sometimes children use a “sounding-out”strategy, involving sequential translation of letters to sounds. The ability to sound words out develops from a slow, attention-demanding, and laborious process to one that is rapid, at least partially automatic, and relatively effortless. More skillful phonological recoding probably develops as children learn to recognize syllabic or intrasyllabic word parts as units (e.g., when they learn that the -tion suffix reliably encodes the syllable /shun/). Such sophisticated spelling knowledge is characteristic of the orthographic stage of reading development (Frith, 1985) and probably represents an instance of chunking smaller units into larger ones (Rozin & Gleitman, 1977). Just as children discard their training wheels as they learn to maintain balance on a moving bicycle, novice readers put aside laborious sounding and blending routines as they begin to perceive larger orthographic units.
Other times children use an analogy strategy, as described by Goswami (chap. 2, this volume). In such a strategy, the child uses a known word to decode an unknown word. We suggest that the successful use of an analogy strategy is dependent on a child’s full encoding of the known word (Gaskins, Ehri, Cress, O’Hara, & Donnelly, 1996). Without such full encoding, the child is likely to confuse that word with others with similar rime patterns, and make errors that would be classified as phonetic cue errors. It is only at the orthographic stage that a child is able to choose effectively and consistently among analogues.
Development of Recoding Skill. Monaghan (cited in Ehri, 1991), in a longitudinal study of recoding acquisition, described phases she observed in the development of the skill of pronouncing pseudowords. At first, the children articulated the sounds represented by letters but could not blend. Next, they learned to sound out and blend, but their efforts were slow and effortful. In a transitional stage, they learned to recode in a quiet voice or by silently moving their lips before saying the word. Finally, they could read the pseudowords rapidly without moving their lips. Monaghan’s observations suggest that the skill of phonological recoding develops from a slow overt process to a fast silent one. Moreover, skilled recoding depends on moving from sounding and blending based on single letters to mentally dividing a word into pronounceable multiletter units (e.g., consonant and vowel digraphs) and recognizing the signaling influence of nearby letters (e.g., a final e). More sophisticated knowledge of orthographic constraints is characteristic of later stages of word recognition development (Frith, 1985).
DEFINING READING IN THE PHONOLOGICAL AWARENESS LITERATURE
We reviewed a selected set of studies to examine how reading has been assessed in phonological awareness studies, including both training studies and correlational studies. In the 24 training studies we looked at, 14 used a measure of decoding, including pseudoword decoding. Eight used a measure of word recognition. Only one used a measure of phonetic cue reading. Measures of connected text reading were used in four studies.
Word recognition involves a series of increasingly more sophisticated knowledge of letters and sounds and how they map unto the speech stream. However, word recognition in isolation is not the same as word recognition in context. Adams and Huggins (1985) found that children can read words somewhat more accurately in context than they can in isolation. This effect has been found by a number of other researchers, although different researchers have found different magnitudes of the effects (cf. Goodman, 1983; Nicholson, 1991). In context, word recognition processes interact with comprehension-driven processes. This interaction is suggested to be more pronounced with younger or less competent readers (Stanovich, 1980), such as those who are the participants in phonological awareness studies.
Reading words in isolation is a more focused test of the effects of phonological awareness on reading than contextual measures. We assume that phonological awareness most directly affects the learning of sound–symbol relationships and their use in decoding and word recognition. Reading connected text does involve decoding, but also involves some use of contextual information (e.g., Adams & Huggins, 1985). This may be especially true for beginning readers. If phonological awareness training has strong effects on measures of decoding but lesser effects on measures of oral reading of connected text, this would support the view that phonological awareness specifically affects sound–symbol knowledge. Using broader measures of reading would also enable richer models of reading acquisition, including the use of contextual information and how it interacts with phonological and orthographic information.
ALPHABET KNOWLEDGE AND PHONOLOGICAL AWARENESS
In the research on phonological awareness, there is an unstated assumption that some level of phonological awareness combined with alphabet knowledge is necessary for children to learn to decode. There is an assumption, in Adams (1990) among others, that alphabet knowledge is separate from phonological knowledge. In our research and others, however, we found that alphabet knowledge seems to precede the simplest level of phonological awareness that we tested. We discovered evidence of this in a number of studies. Lomax and McGee (1987) tested a multicomponent model of emergent reading. They examined 81 3-, 4-, 5-, and 6-year-olds on a variety of measures, using LISREL analysis to examine relationships among variables. They found that the children’s “graphic awareness” factors (consisting of measures of letter orientation, letter discrimination, and word discrimination) accounted for 99% of the variance in their phonemic awareness factor (consisting of two word-to-word matching tasks, one with beginning sounds and one with final sounds, and an auditory discrimination task). These tasks seem rather different than those that we would use. In their meta-analysis of the effects of phonological awareness training, Wagner and Rashotte (1993) found that only those programs that combined phonological awareness training with alphabet recognition (and, consequently, spelling) had a significant effect on reading achievement (e.g., Bradley & Bryant, 1983). (An exception is the study by Lundberg, Frost, & Petersen, 1988, which found significant effects on a reading measure from a program that did not include letter training.)
Other studies more directly tested the effects of alphabet knowledge on phonological awareness. Wagner, Torgesen, and Rashotte (1994), in a longitudinal study of the relations between early reading and reading-related abilities, found that letter-name knowledge appeared to have a causal relationship with phonological processing. Stahl and Murray (1994), using scattergram analysis, found that alphabet knowledge seems necessary for phonological awareness. They found only one child who could successfully recognize and manipulate onsets and rimes but who was unable to recognize at least 45 of 54 letter forms. This finding suggests that knowledge of letter identities may be necessary for phonological awareness, although, of course, not sufficient.
Why might this be? One explanation might be that having a concrete referent, such as a letter, may make it easier to understand an abstract entity such as a phoneme (e.g., Hohn & Ehri, 1993). This could be tested by using another concrete mediator, such as the symbols developed by Calfee, Chapman, and Venezky (1972). Another explanation might be that phonological awareness and alphabet knowledge may both be mediated through exposure to alphabet books. Such books typically include both letter-name information and phonological information about initial sounds (b is for bear). It may be that children who are read alphabet books, and thus understand how b is for bear, will learn both letter names and be able to isolate phonemes.
We pursued this second possibility through an experimental study. We gave three treatments to different groups of prekindergarteners, a total of 42 students in three classes. In the first group, the teacher read conventional alphabet books. In the second, the teacher read books chosen to contain the letter names only, without reference to the sound values, such as Martin and Archambault’s (1989) Chicka-Chicka-Boom-Boom. The third group, a control, read only storybooks. All groups gained in print concepts and letter knowledge over the course of the study. The conventional alphabet group made significantly greater gains in phoneme awareness than did the group that read the letter-name books without sound values, suggesting that conventional alphabet books may be one route to the development of phonological awareness. However, these gains were relatively small in an absolute sense. In addition, the storybook reading class, for reasons we believe were irrelevant to the treatment, also made significant gains in phonological awareness.
This study provides tentative support for the notion that alphabet books can mediate phonological awareness. The notion that b is for bear or m is for mouse may require children to construct the concept of phoneme identity, which in turn enables the children to partially segment sounds from spoken words. We suggest, along with Yaden, Smolkin, and MacGillivray (1993), that in order for children to understand how b could stand for bear they must begin to look at words phonologically. Yaden et al., in an ethnographic study of children’s alphabet book reading, found evidence that they went through a period of disequilibrium when listening to alphabet books. This disequilibrium was resolved when they were able to focus on the phonological, rather than semantic, aspects of words.
HOW DOES PHONOLOGICAL AWARENESS RELATE TO READING?
Table 3.1 shows the possible definitions of phonological awareness, is related to, and reading discussed so far. It is not enough to posit that phonological awareness is related to reading; we must posit which aspect of phonological awareness is related to which aspect of reading in which way.
Our tentative model is shown in Fig. 3.3. Working from data from several of our studies, confirmed by findings of others in the field, we can find some probably causal links using the logic of our scattergrams. As noted earlier, alphabet knowledge seems to be related to phoneme identity, possibly through alphabet book reading (Murray, Stahl, & Ivey, 1996). Alphabet knowledge may mediate phoneme identity learning through either the process of confronting why b might be for bear. Children also may use alphabet letters as concrete mediators for phonemes. Hohn and Ehri (1993) directly compared phonological awareness training with and without simultaneously training subjects on letter names. They found that the letter-name training produced somewhat superior performance on measures of phonological awareness. As noted earlier, Wagner and Rashotte (1993), in a meta-analysis of phonological awareness training studies, found that only those training programs that included some letter or spelling training had a significant effect on children’s acquisition of written words. In contrast, Read, Yun-Fei, Hong-Kin, and Bao-Qing (1986) found that literate adult readers of a non-alphabetic language had difficulties with complex phonological awareness tasks. In addition, Morais, Cary, Alegria, and Bertelson (1979) found that illiterate adults could not similarly perform complex phonological awareness tasks. Both sets of researchers suggested that their research emphasizes the key role of an alphabetic language for developing phonological awareness. We suggest that it is the alphabet itself that is responsible for that relation.
Possible Relationships Among Terms
| Phonological Awareness | Is Related To | Reading |
| Rhyming | Correlation | Visual cue reading |
| Partial segmentation | Necessary but not sufficient | Alphabet knowledge |
| Identity | Phonetic cue reading | |
| Blending | Phonological recoding | |
| Deletion | ||
| Full segmentation |
FIG. 3.3. Possible relations among phonological awareness and reading.
The alphabet may also mediate the development of phonological awareness through children’s invented spelling. As children move from prephonetic spelling, in which the letters used bear no relation to the sounds in the words (xgrs for giraffe), to early phonemic spellings, where the children use the initial and sometimes the final consonant sounds to represent the whole word (b or br for bear), they are beginning to analyze the spoken word. Having a letter to represent that sound seems to be the beginning of the establishment of phonemic identity. The relationship between phonological awareness and invented spelling may be reciprocal. Tangel and Blachman (1992) found that a phonological awareness training program improved children’s invented spelling as well.
The concept of phoneme identity is closely related to the ability to segment a part of a spoken word, as evidenced by scattergrams in the Murray (1995) study, shown in Fig. 3.1 above. We see the relation between phoneme identity and partial segmentation as more correlational. Conceptually, it makes sense to see these two abilities as covarying; that is, as children learn the identities of more phonemes, they are better able to isolate their sounds in spoken words.
Partial segmentation (as well as alphabet knowledge and other print concepts) seems to underlie the ability to read words. We speculate that partial segmentation is needed for phonetic cue reading, because noticing initial letters in words and connecting those to sounds seems highly similar to segmenting initial sounds in spoken words. In addition, Byrne and Fielding-Barnsley (1989) and Murray (1995) found that phoneme identity seemed to be necessary for phonetic cue reading.
Phoneme awareness also seems to influence fingerpoint reading, another important precursor to the development of word recognition. Ehri and Sweet (1991) found that phoneme awareness, as measured by success on a training task, accounted for a significant amount of variance in fingerpointing to words in a memorized book. When this variable was in the equation, other variables commonly used in early reading studies, such as alphabet knowledge and word recognition, failed to contribute significant amounts of variance.
Phonetic cue reading, in turn, would lead to more advanced knowledge of words, such as the cipher reading discussed by Ehri (1991) and Gough, Juel, and Griffith (1992). This advanced knowledge, in its own turn, would enable children to perform more complex phonological awareness tasks, such as blending, deletion, and full segmentation. A number of researchers—including Beach (1992), Perfetti, Beck, Bell, and Hughes (1987), and Stahl and Murray (1993)—found that the relationship between phoneme awareness and reading is a reciprocal one, with success with simple phoneme awareness tasks, such as segmentation of initial consonants, preceding simple word recognition, but that some recognition of words typically precedes more advanced phonological awareness abilities, such as deletion.
What we have tried to do is to simplify a greatly complex body of literature. To us, the notion of phoneme identity is a crucial one, because it seems to be a bridge between alphabet knowledge and phonological awareness and phonological awareness and early decoding. However, phoneme identity is a relatively new concept. Whether its central role holds up is yet to be determined.
When we talk to a class of inservice or preservice teachers, we always try to clearly distinguish phonological awareness from phonics. We have not always been successful. Students in our classes, despite our best efforts, still confuse the two constructs. Up until recently, the two seemed to be clearly distinct—phonological awareness had to do with spoken words, phonics with written words. However, the concept of phoneme identity seems to be midway between the notions of phonics and the more mechanical notions of phonological awareness. The evidence for this view is that alphabet knowledge seems to be implicated in the development of phoneme identity. Thus, what we have been calling “phonological awareness” may indeed be something akin to the processes of learning letter–lsound relations and using them to decode words. At the least, the distinctions between phonological awareness and phonics may be finer than many think.
Second, if one looks at the factors that we have identified as crucial for the development of phonological awareness (and, through phonological awareness, word recognition), it seems that all of them are ordinarily fostered in a home environment with a literacy press. As Adams (1990) described such a home, it contains not only alphabet books, but magnetic letters on the refrigerator, alphabet blocks, Sesame Street, and other media for learning about the alphabet. The children are read to from a variety of storybooks for a significant amount of time each day. Through the modeling of reading and alphabet play, children develop not only alphabet knowledge and phoneme identity knowledge, but also a broader knowledge of the functions and forms of reading and literacy. The presence of so many activities that foster literacy in the stereotypical home with a strong literacy press may suggest why all of these early reading skills are so strongly correlated, and why factor analyses tend to produce single factor solutions.
Stressing the key role of the home, however, begs the question of why some children fail to read in supportive environments. It also begs the question of how to teach children who are significantly behind their peers because of the early lack of phonological awareness (Stahl, 1997b).
The model shown in Fig. 3.3 suggests that all of these basic literacy skills grow together—that they are entwined rather than separate skills that should be taught through targeted programs. Instead, the most desirable way of developing these early literacy skills is through a supportive literacy environment in a home with a strong literacy press (McClain, 1995). Many homes do not provide such a press. Adams (1990) contrasted her reading to her son with parents and children studied by Teale (cited in Adams, 1990). She observed that, by reading to her son for 30 to 45 minutes per day, her child will come to first grade with over a 1,000 hours of literacy experience that children who have not been read to at home will lack. Making up these 1,000 hours becomes a difficult if not insurmountable job for the school. Understanding how key early literacy skills fit together can make early instruction more efficient and help us bridge this yawning gap.
REFERENCES
Adams, M. J. (1990). Beginning to read: Thinking and learning about print. Cambridge, MA: MIT. Press.
Adams, M. J., & Huggins, A. W. (1985). The growth of children’s sight vocabulary: A quick test with education and theoretical implications. Reading Research Quarterly, 20, 262–281.
Beach, S. A. (1992). Toward a model of the development of reader resources in the emergence and acquisition of literacy skill. Unpublished doctoral dissertation, University of California at Riverside.
Biemiller, A. (1970). The development of the use of graphic and contextual information as children learn to read. Reading Research Quarterly, 6, 75–96.
Bradley, L., & Bryant, P. E. (1983). Categorizing sounds and learning to read—a causal connection. Nature, 301, 419–421.
Byrne, B. (1992). Studies in the acquisition procedure for reading: Rationale, hypotheses, and data. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition. Hillsdale, NJ: Lawrence Erlbaum Associates.
Byrne, B., & Fielding-Barnsley, R. (1989). Phonemic awareness and letter knowledge in the child’s acquisition of the alphabetic principle. Journal of Educational Psychology, 81, 313–321.
Byrne, B., & Fielding-Barnsley, R. (1990). Acquiring the alphabetic principle: A case for teaching recognition of phoneme identity. Journal of Educational Psychology, 82, 805–812.
Byrne, B., & Fielding-Barnsley, R. (1993). Evaluation of a program to teach phonemic awareness to young children: A 1-year follow-up. Journal of Educational Psychology, 85, 104–111.
Calfee, R., Chapman, R., & Venezky, R. (1972). How a child needs to think to learn to read. In L. W. Gregg (Ed.), Cognition in learning and memory (pp. 139–182). New York: Wiley.
Chall, J. S. (1996). Stages of reading development (2nd ed.). New York: Harcourt Brace.
Ehri, L. C. (1991). Development of the ability to read words. In R. Barr, M. L. Kamil, P. B. Mosenthal, & P. D. Pearson (Eds.), Handbook of reading research (Vol. II, pp. 383–417). White Plains, NY: Longman.
Ehri, L. C., & Sweet, J. (1991). Fingerpoint-reading of memorized text: What enables beginners to process the print. Reading Research Quarterly, 26, 442–462.
Ehri, L. C., & Wilce, L. S. (1985). Movement into reading: Is the first stage of printed word learning visual or phonetic? Reading Research Quarterly, 20, 163–179.
Ehri, L. C., & Wilce, L. S. (1987). Cipher versus cue reading: An experiment in decoding acquisition. Journal of Educational Psychology, 79, 3–13.
Fox, B., & Routh, D. K. (1975). Analyzing spoken language into words, syllables, and phonemes: A developmental study. Journal of Psycholinguistic Research, 4, 331–342.
Frith, U. (1985). Beneath the surface of developmental dyslexia. In K. E. Patterson, J. C. Marshall, & M. Colheart (Eds.), Surface dyslexia: Neuropsychological and cognitive studies of phonological reading (pp. 301–330). Hillsdale, NJ: Lawrence Erlbaum Associates.
Gaskins, I. W., Ehri, L. C., Cress, C., O’Hara, C., & Donnelly, K. (1996). Procedures for word learning: Making discoveries about words. The Reading Teacher, 50, 312–328.
Gillet, J. W., & Temple, C. (1990). Understanding reading problems (3rd ed.). Glenview, IL: Scott-Foresman.
Goodman, K. S. (1983). A linguistic study of cues and miscues in reading. In L. M. Gentile, M. L. Kamil, & J. S. Blanchard (Eds.), Reading research revisited (pp. 187–192). Columbus, OH: Merrill.
Gough, P. B., & Hillinger, M. L. (1980). Learning to read: An unnatural act. Bulletin of the Orton Society, 30, 179–196.
Gough, P. B., Juel, C., & Griffith, P. L. (1992). Reading, spelling, and the orthographic cipher. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 35–48). Hillsdale, NJ: Lawrence Erlbaum Associates.
Hohn, W. E., & Ehri, L. C. (1983). Do alphabet letters help prereaders acquire phonemic segmentation skill? Journal of Educational Psychology, 75, 752–762.
Jorm, A. F., & Share, D. L. (1983). Phonological recoding and reading acquisition. Applied Psycholinguistics, 4, 103–147.
Kavanaugh, J. F., & Mattingly, I. C. N. (1972). Language by ear and by eye. Cambridge, MA: MIT Press.
Lewkowicz, N. K. (1980). Phonemic awareness training: What to teach and how to teach it. Journal of Educational Psychology, 72, 686–700.
Liberman, A. M., Cooper, F. S., Shankweiler, D. P., & Studdert-Kennedy, M. (1967). Perception of the speech code. Psychological Review, 74, 431–461.
Liberman, I. Y. (1973). Segmentation of the spoken word and reading acquisition. Bulletin of the Orton Society, 23, 65–77.
Liberman, I. Y., Shankweiler, D., Fischer, F. W., & Carter, B. (1974). Reading and the awareness of linguistic segments. Journal of Experimental Child Psychology, 18, 201–212.
Lomax, R. G., & McGee, L. M. (1987). Young children’s concepts about print and reading: Toward a model of reading acquisition. Reading Research Quarterly, 22, 237–256.
Lundberg, I., Frost, J., & Peterson, O.-P. (1988). Effects of an extensive program for stimulating phonological awareness in preschool children. Reading Research Quarterly, 23, 263–284.
Maclean, M., Bryant, P., & Bradley, L. (1987). Rhymes, nursery rhymes, and reading in early childhood. Merrill-Palmer Quarterly, 33, 255–281.
Martin, B, Jr., & Archambault, J. (1989). Chicka-chicka-boom-boom. New York: Simon & Schuster.
Masonheimer, P. E., Drum, P. A., & Ehri, L. C. (1984). Does environmental print identification lead children into word reading? Journal of Reading Behavior, 16, 251–271.
McClain, V. P. (1997). Literacy press: How families promote literacy. Unpublished doctoral dissertation, The University of Georgia, Athens.
McCormick, C. E., & Mason, J. M. (1986). Intervention procedures for increasing preschool children’s interest in and knowledge about reading. In W. H. Teale & E. Sulzby (Eds.), Emergent literacy: Writing and reading (pp. 90–115). Greenwich, CT: Ablex.
McCutcheon, B. A., & McDowell, E. E. (1969). Intralist similarity and acquisition and generalization of word recognition. The Reading Teacher, 23, 103–107.
Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech as a sequence of phones arise spontaneously? Cognition, 7, 323–331.
Murray, B. A. (1995). Winch better defines phoneme awareness: Segmentation skill or identity knowledge? Unpublished doctoral dissertation, University of Georgia, Athens.
Murray, B. A., Stahl, S. A., & Ivey, M. G. (1996). Developing phoneme awareness through alphabet books. Reading and Writing: An Interdisciplinary Journal, 8, 307–322.
Nicholson, T. (1991). Do children read words better in context or in lists: A classic study revisited. Journal of Educational Psychology, 83(4), 444–450.
Perfetti, C. A., Beck, I. L., Bell, L., & Hughes, C. (1987). Phonemic knowledge and learning to read are reciprocal: A longitudinal study of first grade children. Merrill-Palmer Quarterly, 33, 283–319.
Read, C., Yun-Fei, Z., Hong-Kin, N., & Bao-Qing, D. (1986). The ability to manipulate speech sounds depends on knowing alphabetic writing. Cognition, 24, 31–44.
Rosner, J. (1974). Auditory analysis training with prereaders. The Reading Teacher, 27, 379–384.
Rosner, J. (1975). Helping children overcome learning difficulties. New York: Walker.
Rozin, P., & Gleitman, L. (1977). The structure and acquisition of reading. In A. S. Reber & D. L. Scarborough (Eds.), Toward a psychology of reading (pp. 1–76). Hillsdale, NJ: Lawrence Erlbaum Associates.
Savin, H. B. (1972). What the child knows about speech when he starts to learn to read. In J. F. Kavanaugh & I. C. Mattingly (Eds.), Language by ear and by eye (pp. 319–326). Cambridge, MA: MIT Press.
Shankweiler, D., & Liberman, I. Y. (1972). Misreading: A search for causes. In J. F. Kavanaugh & I. G. Mattingly (Eds.), Language by eye and by ear (pp. 293–317). Cambridge, MA: MIT Press.
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, 55, 151–218.
Stahl, S. A. (1997a). Models of reading instruction: An introduction. In S. A. Stahl & D. A. Hayes (Eds.), Instructional models in reading (pp. 1–29). Hillsdale, NJ: Lawrence Erlbaum Associates.
Stahl, S. A. (1997b). Teaching children with reading problems to recognize words. In L. Putnam (Ed.), Readings on language and literacy: Essays in honor of Jeanne S. Chall (pp. 131–154). Cambridge, MA: Brookline Books.
Stahl, S. A., & Murray, B. A. (1993). Environmental print, phonemic awareness, letter recognition, and word recognition. In D. J. Leu & C. K. Kinzer (Eds.), Examining central issues in literacy research, theory, and practice: Forty-second yearbook of the National Reading Conference (pp. 227–233). Chicago: National Reading Conference.
Stahl, S. A., & Murray, B. A. (1994). Defining phonological awareness and its relationship to early reading. Journal of Educational Psychology, 86, 221–234.
Stanovich, K. E. (1980). Toward an interactive-compensatory model of individual differences in the development of reading fluency. Reading Research Quarterly, 16, 32–71.
Tangel, D. M., & Blachman, B. A. (1992). Effect of phoneme awareness instruction on kindergarten children’s invented spellings. Journal of Reading Behavior, 24, 233–262.
Torgesen, J. K., & Bryant, B. R. (1994). Test of phonological awareness. Austin, TX: Pro-Ed. Treiman, R. (1992). The role of intersyllabic units in learning to read and spell. In P. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 85–106). Hillsdale, NJ: Lawrence Erlbaum Associates.
Treiman, R. (1985). Onsets and rimes as units of spoken syllables: Evidence from children. Journal of Experimental Child Psychology, 39, 161–181.
Wagner, R. K., & Rashotte, C. A. (1993, April). Does phonological awareness training really work? A meta-analysis. Paper presented at annual meeting, American Educational Research Association, Atlanta, GA.
Wagner, R. K., Torgesen, J. K., & Rashotte, C. A. (1994). Development of reading-related phonological processing abilities: New evidence of bidirectional causality from a latent variable longitudinal study. Developmental Psychology, 30, 73–87.
Yaden, D. B., Smolkin, L. B., & MacGillivray, L. (1993). A psychogenetic perspective on children’s understanding about letter associations during alphabet book readers. Journal of Reading Behavior, 25, 43–68.
Yopp, H. K. (1988). The validity and reliability of phonemic awareness tests. Reading Research Quarterly, 23, 159–177.
__________
lThe instrument we used was a modification of Torgesen and Bryant’s (1994) Test of Phonological Awareness (TOPA). The TOPA is designed to be a word-to-word matching test with illustrations of the words to be matched. Children are directed, for example, to notice pictures of leg, lamp, hand, and fish, and to “mark the one that begins with the same sound as leg” To make the test more explicit, the examiner pronounced the target phoneme in isolation, telling children that “lamp begins with /l/,” and asking them to “mark the one that begins with /l/, as in leg” By this modification, the TOPA became a sound-to-word matching test, operationalizing the identity view of phoneme awareness as knowledge of particular phoneme identities across words.
2If two variables are both skewed, as is typical in this type of research, then the presence of a third variable could cause this pattern. Thus, as in other basically correlational research, one cannot impute causality. However, this pattern is consistent with a “necessary but sufficient” pattern.