Thursday, March 24, 2011

Acquiring the Alphabetic Principle: A Case for Teaching Recognition of Phoneme Identity (1990)

Byrne, B. & Fielding-Barnsley, R. (1990). Journal of Educational Psychology, 82(4), 805-812.

Given: an understanding of the phonological structure of language, a.k.a. PA, initiates an understanding of the alphabetic principle, the prerequisite to reading (Bradley & Bryant, 1983; Byrne & Fielding-Barnsley, 1989; Juel, Griffith, & Gough,1986; Tunmer, Herriman, & Nesdale, 1988).

This article addresses "some questions of detail concerning phonemic awareness—in particular, how it might be efficiently taught,whether various phoneme classes might present special problems, and whether phonemic awareness "spreads" unaided from the phonological locality in which it was first generated."

Experiment 1
Dealt with teaching preschoolers phoneme identity and the phoneme's position in the word (beginning and ending). Children trained to identify /m/ and /s/. Later /f/ and /b/ were introduced. The research question addressed was whether children trained in identity on the first two sounds would show evidence of the alphabetic principle through successfully transferring their knowledge to /f/and /b/.

Training: The experimenter first named items presented to children and said they started or ended with a target  phoneme. This was repeated twice (three exposures). The child was then asked to name the items 3 times while the experimenter reminded the child that "all the words are /?/ words." This training lasted 4 days.
Testing followed immediately after each training day - the child was presented with two pictures, provided the names of the pictured items, told "one is an /?/ word and one is not" then was asked "Which is the /?/ word?"
12 pairs of words were used while six were presented after each training session.

Days 5 and 6, were termed the alphabetic stage of the experiment. Children were trained to read the words sat and mat to a criterion of six correct responses. They were also taught the the phonemes associated with the graphemes for /s/ and /m/ (same criterion). Transfer was tested by presenting new words while being asked to choose one with a target phoneme: Is this "mow" or "sow" ? The eight transfer items were "sum, mad, sow, met, mum, sad, mow and set." On day 6,the same procedure as day 5 continued with the phoneme-grapheme instruction for f and b with a transfer task. Words targeted for phonetic cue reading: "fun, big, fell, bat, bun, fig, bell, and fat." (p.806)

Results: "children can be trained to notice the identity of phonemic segments in words."

  • 15 of 16 subjects scored better than chance on some of the identity tests
  • 11 reached criterion on six or more of the eight phonemes
  • children no worse at identity of final consonants than initial
  • high correlations among eight identity tasks means the concept of phoneme identity for a single sound can transfer to all sounds. 
  • children with a high level on phonemic awareness completed the transfer task 

". . . a single aspect of phonemic awareness, sensitivity to phoneme identity, was sufficient to support acquisition of the alphabetic principle (p. 807)."

Implications for teaching: reading curriculum can be limited to select phonemes rather than  systematic coverage or all phonemes.

Experiment 2


Experiment 2 examined the same issues as Experiment 1 but focused on segmentation. Did thorough training in segmentation alone lead to acquisition of the alphabetic principle? The second question addressed transfer, would children who successfully acquired /s/ and /m/ perform well on /f / /b/ transfer without segmentation training for these phonemes? A third inquiry involved position effect (initial vs. final). The effect did not exist for phoneme identity, but was hypothesized that it could occur in the segmentation task.


Results: identity performance predicted total alphabetic transfer whereas segmentation did not.

Segmentation of the initial and final consonants of words was taught to preschoolers. There was no child who failed to respond correctly to some degree.  But performance on segmentation task was less stable on than it was on the identity task. The results from Experiment 1 showed that once the principle
of phoneme identity was understood, it generalized to untaught phonemes, but this did not apply to segmentation in Exp. 2. However, segmentation training did make children aware of phoneme identities. Thus, a benefit of segmentation is that it results in phoneme identity


Article's purpose is the "fine-tuning of instructional materials designed to teach phonemic awareness."

Phoneme identity tasks differ based on phoneme position in word. Final phonemes are harder to detect than initial phonemes.

Friday, March 4, 2011

(2003) Phonemic Awareness Instruction: Contribution of Articulatory Segmentation to Novice Beginners' Reading and Spelling

(Castiglioni-Spalten and Ehri, 2003, pp.25-52) Scientific studies of reading vol 7


PA Facts noted by C-S & E, 2003:


Prereaders are unaware of individual phonemes in words 
(Adams, 1990; Ehri, 1979; Griffith & Olson, 1992; Liberman,  Shankweiler, Foscher, &Carter 1974; Stahl & Murray, 1994; Wagner & Torgesen, 1987).


"Beginning readers need to learn how distinguish phonemes in spoken words and how they are linked to graphemes in the spellings of words” (C-S & E, 2003, p. 25).


These studies show that PA and letter knowledge predict reading ability:
     Juel, Griffith, & Goff, 1986; Share, Jorm, Maclean, & Matthews, 1984)
Segmenting support over "other" tasks:
     Cunningham, 1990; Ludberg, Frost, & Peterson, 1988)
Using letters to segment phonemes, more benefits in learning to segment, read, and spell:
     Ball & Blachman, 1991; Bradley & Bryant, 1983, 1985;Byrne & Fielding-Barnsley, 1989; Ehri & Wilce,   
    1987, Hohn & Ehri, 1983; Tangel & Blachman, 1992, 1995; Uhry & Shepherd, 1993)
Meta-analysis verifications of the above knowledge:
     Bus & van IJendoorn, 1999; Ehri et al., 2001)


The separate contribution of markers and articulatory pictures was not determined until this study.


Methods: 
3 groups: mouth treatment used articulatory pictures of mouth movements; ear treatment - used blocks; and control - no treatment  
Urban setting (2 schools, lower SES, children weren't taught to read in K. 
Criteria for inclusion: 
name 13/17 letters, segment one phoneme within 1 of the 10 words, but no more than 3, read only 1 cvc pseudoword and 10 or fewer primer words, identified picture-words on Peabody Pic Vocab, performed about the same as 2 other children to form a triplet with similar scores on segmentation, word reading & vocab. Members of triplets randomly assigned to 3 groups. 1 pretest session, 3-6 training sessions, 2 post-test sessions.


pretests:
  • letter naming (capital letters)
  • segmenting 8 words (modified Yopp, 88)
  • 5 Pseudowords 
  • Spell 5 words
  • read 22 high frequency words
  • Peabody Pic Vocab-Rev

Instruction:


Correspondence relationship between blocks and sounds taught. Segmentation, pronounce separate sounds in words and pseudowords, and position blocks for sounds. The meaning of the words was provided with a picture card while pseudowords were represented with people's faces.
Children had to reach a criterion of 8 consectutive correct responses or have completed 6 training sessions. word patterns used were cv, vc, cvc, ccv vcc ccvc, cvcc. The last task (exit task) A picture card was shown, the experimenter said what it was, provided a defining sentence, thent he child repeatedthe names. Finally, the child was shown pictures again, had to name the object and segment it in a frame (3 or 4 segments).


Researcher used a mirror to help children identify mouth moves.
The control group did regular kindergarten work with no PA or reading instruction. They memorized and sang the alphabet.


Posttests


administered a day after training ended and a week later, by another researcher, blind to participant grouping.
1. segmented words into phonemes using pretest words + 7 new words/delayed test had 14 new words
2. pseudoword decoding, same as pretest, only on day after posttest
3. spelling pretest readministere on both posttests
4. different word were taught for learning to read words: 8 s words on potst 1 and 8 f words on potst 2. Five trials of practicing reading (no words had been taught but sounds had been taught). 
Experimenter read the word the first time on an index card, provided a defining sentence, andthe child repeated it. On subsequent trials, word presented on a card, if incorrect, researcher provided the word for each trial.  Word learning was measured by counting the number of words read correctly over 5 trials, scoring only the first response.


ANOVAs verified treatment groups were well matched on pretests. 
Reliability of posttest assessed with Chronbacks alph
ANOVAs used for effects of PA on outcomes under 3 conditions at 2 times. Post hoc Tukey pairwise comparison found effect sizes of 1.47 for mouth group and 1.53 for ear group on segmenting task.


Theory at work: Connectionist model proposed by Plaut et al. 1996.


Rack et al., Laing & Hulme (1999) showed that word learning is a "connection-forming process, referred to as direct mapping hypothesis.


Ehri's theory: beginners read words by "forming connections between some leters in spellings and phonetic cues in pronunciations." Children become sensitive to "articulatory gestures" providing them "better access to phonetic cues in word if . . .  motoric gestures" are used (Liberman & Mattingly, 85) (in C-S & Ehri, 2003, p. 46).