Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Wiley and International Reading Association are collaborating with JSTOR to digitize, preserve and extend
access to Reading Research Quarterly.
http://www.jstor.org
? InternationalReadingAssociation7031-4620/91/US$1.25+ .00
LinneaC. Ehri
JenniferSweet
Universityof California,Davis
of emergent literacy have revealed Yaden & Templeton, 1986). Very young chil-
Studies
that childrenbecome readersmuch earlier dren become readers as they gain experience
than is commonly recognized- in fact, long be- listening to stories and learning how they are
fore they can read novel words and stories on structured semantically, in terms of ideas, as
their own (Sulzby, 1985; Teale & Sulzby, 1987; well as visually, in terms of appearance on the
were given on Day 1, prior to the fingerpoint- the title of this nursery rhyme was written in
readingtraining. capitalletters.
In addition, a story of 8 lines was creat-
1. Reading target words. Subjects were ed by modifying one verse of a poem by
asked to read 12 words from the story that Olive A. Wadsworth entitled "Over in the
would be used for the fingerpoint-readingtrain- Meadow."Wordswere changedto make the ini-
ing. Each word was printedin capital letters on tial letters more varied, to make the text less fa-
a flash card. The words were blue, old, dived, miliar,andto includemore wordswith an initial
fishes, and, mother,jungle, the, swim, where, letter that symbolized the initial sound in the
down, and two. Five picture cards were mixed word as well as a sound in the name of the letter
in with the word cards to minimize nonreaders' (e.g., M, named "em"'containing/m/, the first
sense of failure. Subjects who could read more sound in Mother). The 8 lines were printed in
than one target word were excluded from the capitalletters, one line per page:
study. Two of the subjects who were retainedin
the sample were able to read one target word DOWNIN THEJUNGLE
each; the remainingsubjectscould not readany WHERETHERIVERRUNSBLUE
targetwords. LIVEDAN OLDMOTHERFISH
AND HERLITTLEFISHESTWO.
2. Naming letters. Subjects named 26 up- "SWIM," SAIDTHEMOTHER.
"WEARESWIMMING" SAIDTHETWO.
percase letters presentedin randomorder on a
card. Scores of subjects ranged from 0 to 26; SO THEY PADDLED AND THEYDIVED
25 percent knew 0-10 letters, 22 percent knew WHERETHERIVERRUNSBLUE.
11-21 letters, and 53 percent knew 22-26 let-
ters. Each line was accompaniedby a relevant pic-
ture in color. The 8 pages formed a soft-cover
3. Reading preprimerand Slosson words. booklet. Capitalletters were used because they
Subjects were shown 11 preprimer words on are more familiarto preschoolers.
flash cards mixed with 5 picture cards. If the
child read at least one word, then he or she was Training.On Day 1, each child was first in-
shown 11 more preprimerwords mixed with 3 troduced to fingerpoint-readingusing the pic-
picture cards. Subjects who read more than 2 ture of "Peter,Peter,PumpkinEater"along with
words were given the Slosson OralReadingTest the printedtitle. The experimenterread the title
(1963). Subjects who scored above the pre- twice, pointing to each word as she said it. She
primer level on this test were excluded from explainedthat she knew which wordsto pointto
the study. Scores of subjects who remained in by attendingto the correspondencebetween ini-
the study ranged from 0 to 9 preprimerwords tial letters and sounds in the words. Then the
correct; 67 percent read 0 words, 28 percent child attemptedto point to each word as he or
read 1-3 words, and the 2 remaining subjects she spokethe four-wordtitle. If this attemptwas
read 6 and 9 words, respectively. Thus, the unsuccessful, the experimentermodeled how to
sample comprised a large group of nonreaders point at the words until the child was able to do
(n = 24) and a small group of novice readers so correctly, up to a maximum of three times.
(n = 12). Most children (86%) were successful within
two trials.
Fingerpoint-reading training Next, the experimenterreadthe story aloud
Materials. Two sets of materialswere used while the subject looked at the pictures in the
for the trainingin fingerpoint-reading.The ma- booklet. Then she read the story again more
terials for the introductorytask consisted of an slowly while pointingher finger at the words as
illustrationof the nursery rhyme "Peter,Peter, the child watched. The child and experimenter
PumpkinEater"and a sheet of paper on which then readthe book together in chorus, and both
pointed to the words- with the experimenter 1. Withthe bookclosed, the child was asked,
pointing above the line, and the child below the "Werethe wordsin thatbookjust like the
line. The child then fingerpoint-readthe first wordswe readbeforein the otherbook?"If
line alone. The experimenterand child read the the child answered"No,"the experimenter
story together again in chorus using finger- asked,"Howweretheydifferent?"
2. With the book open, the experimenter
pointing. Then the child fingerpoint-readthe pointedto the first line of printandasked,
whole story alone. This was followed by one fi- "Dothesewordslookjust likethewordswe
nal choral reading. readbefore?" If thechildanswered"No,"the
On Day 2, the story was read five times. experimenter asked,"Howweretheydiffer-
On Trials 1, 2, and 4, the experimenterand the ent?"
child read the story in chorus while pointing to 3. Withthe firstbookopenandpositionedon
the words. On Trials 3 and 5, the child finger- top of the secondbook so thatthe original
point-readthe story alone. andalteredlinesof printwerevisible,theex-
perimenter said, "Lookat thesewords.Are
Taskperformance. Performanceon the fi- thesejustlikethose?"
nal solo readingof the 8 lines of text was scored
for three characteristics:(a) correctly reciting Childrenwere assigned 4 points for spontane-
each line, (b) pointing to words in each line, ous recognition of the altered text before any
and (c) voice-point matching (i.e., pointing to questions were asked, 3 points for answering
words at the same time as they were spoken). that the text had been changed when the book
The scoring criteriafor the first two characteris- was closed, 2 points for recognitionwith the al-
tics were as follows: If subjectsrespondedto a tered book open, 1 point for recognition upon
line perfectly,they received a score of 3; if they seeing the two books open side-by-side, and 0
respondedcorrectlyto only some of the words, points for no recognition.
they received a score of 1; if their responsewas 2. Reading isolated lines of text. Subjects
totally incorrect, they received a score of 0. were asked to read three lines from the text
Nonlinear values were adopted to lend greater
(Lines 3, 1, and 7, in that order) with the pic-
weight to perfectperformances.Scores were to- tures removed.
taled for the 8 lines, for a maximum score of
24. The scoring criterion for the voice-point 3. Reading isolated target words. This was
matching measure was the numberof lines re- a repeat of the Day 1 pretest. Subjects were
spondedto perfectly (maximumscore = 8). asked to read 12 words from the story printed
on flashcards,with 5 picturecards mixed in.
Posttests
Six tasks measuringletterand word knowl- 4. Readingtargetwordsin text. The experi-
edge were given after the fingerpoint-reading menter turned to various pages of the text,
task (i.e., the last solo reading)on Day 2. pointed to words in the text out of order, and
asked subjects to read the words. These were
1. Recognizingaltered text. Subjects were the same 12 target words that were tested in
shownan alteredversion of the story booklet, in isolation.
which the pictures, the length of the words, and
the line length remainedthe same as in the orig- 5. Recognizing words in text. The experi-
inal, but new letters were substituted.For ex- menteropened the book and pronouncedwords
ample, the first line, DOWNIN THEJUNGLE, on severalpages, and asked the subjectto point
was changed to WSWSAL WOH OEWNIH. to these words. Five of the target words used
Childrenwere askedto readthe first four pages. above plus 5 new words were used. (Subjects
The experimenternoted whetherthe child spon- were not betterat recognizingpreviouslytested
taneously commented on the altered text, then than untestedwords, indicatingthat testing did
asked three questions: not teach the words.)
Fingerpoint-reading
Recitinglines verbatim 24 17.39 5.06 72% 0% 14%
Pointingto words 24 10.47 7.74 44% 6% 3%
Voice-pointmatching 8 1.83 2.36 23% 44% 3%
Recognizingalteredtext 4 1.47 1.34 37% 19% 17%
Readingwordsin isolation 12 0.50 1.11 4% 78% 0%
Readingwordsin text 12 2.11 2.94 18% 58% 0%
Locatingwords in text 10 2.56 2.22 26% 19% 0%
Note. N = 36.
aTheminimum(notmaximum) of subjectswitha perfectscoreis thepercentage
numberof learningtrialswas4. Thepercentage achieving
criterionin theminimum of trials;thepercentage
number thatneverreachedcriterion.
of subjectswitha zeroscoreis thepercentage
Measure 1 2 3 4 5 6 7 8 9 10 11 12
1. Age - -.01 .12 .11 .31 .23 .07 .38 .31 -.04 .25 .15
2. Naming letters - .46 .08 .64 .50 .34 .47 .48 .61 .55 .54
3. Readingpreprimerwords - .08 .47 .70 .05 .57 .51 .51 .72 .63
4. Letterdiscrimination - .16 .07 .42 .13 .29 .28 .26 .07
5. Correctphonemes segmented - .83 .32 .64 .53 .44 .55 .53
6. Blends segmentedcorrectly - .28 .70 .60 .47 .68 .58
7. Recitinglines verbatim - .26 .38 .46 .22 .18
8. Pointingto words - .89 .38 .61 .57
9. Voice-pointmatching - .48 .61 .48
10. Recognizingalteredtext - .56 .55
11. Readingwords in text - .51
12. Locatingwords in text
Note. r > .32 is significantly
differentfrom0 atp <.05; r > .40 is significantly
differentfrom0 atp <.01.
were not good at segmenting entire blends (M than to reciting accuracy (r = .38). This find-
= 28% correct). Most of the blends that were ing indicatesthatthe majorproblemin learning
correctly segmented consisted of two pho- to fingerpoint-readis not memorizing the text
nemes, either CV or VC (M = 43%). Se- but pointingto the word segments.
quences of threephonemeswere much harderto In the regressionanalysis on scores for re-
segment (M = 13%). citing the text accurately,letter discrimination
ability and letter-name knowledge accounted
Fingerpoint-reading performance for unique variance. However,as shown in Ta-
Three aspects of subjects'ability to learn to ble 3, the amountof varianceaccountedfor was
fingerpoint-read the story were measured in fairly small, .27. What explains this relation-
their final solo readingattempt:accuracyat re- ship is not clear.
citing each line of text verbatim, accuracy at The best regression models for the word
pointingto the wordsin each line, and ability to pointing and voice-point matching measures
coordinate these two behaviors to match print were identical, which is not surprisingbecause
with speech. From the mean percentagecorrect the two measures were highly correlated (see
values in Table 1, it is apparentthat subjects Table 3). One independentvariable explained
learnedto recite the text fairly well. They were the variance: phonemic segmentation skill as
less able to point to the words in each line and measuredby the number of blends segmented
least able to match their voice with their point- correctly. This finding suggests that phonemic
ing. In fact, 44 percent of the subjects scored segmentationis a critical skill enabling begin-
zero on the measure of voice-point matching. ners to learn to fingerpoint-readfamiliartext. It
Inspection of the children's shortcomings re- may be that the hardestpart of the fingerpoint-
vealed that subjects would often slide their fin- reading task is finding the appropriatephone-
gers along beneath the line of text ratherthan mic segments in speech to match up with the
pointing to individual words in the line. How- printedletters.
ever, this behavior was apparentonly in sub- To verify this interpretation,we examined
jects without much skill in fingerpoint-reading; scores on the measures of phonemic segmen-
none of the subjects who fingerpoint-read at tation and voice-point matching. Figure 1
least one line accuratelyever did this. presents a scatter plot of these scores. Data
As shown by the correlations in Table 2, points are distinguishedfor two groups of sub-
accuracy at voice-point matching was more jects: weak segmenters (those who segmented
highly related to pointing accuracy (r = .89) mainly two-phoneme blends) and strong seg-
menters (those who segmented some CV and segmentation skill may emerge before finger-
VC words plus at least 2 of the three-phoneme point-readingand may be a necessary but not
blends). The distributionof the weakest scores sufficient condition for its development. Note
is especially interesting(see lower left cornerof that the phonemic segmentationskill examined
figure): All 8 subjectswho correctlysegmented here is a rudimentaryform: segmenting two-
only 1 or 0 blends failed to fingerpoint-readany phonemeblends.
lines correctly. However, of the 16 subjects in After the fingerpoint-readingpractice, sub-
total who did not fingerpoint-readany lines cor- jects were also given a version with the letters
rectly, 69 percent were able to segment from I alteredand were asked to read it. Subjectsvar-
to 8 blends. These data suggest that phonemic ied in their recognitionthat the print had been
Figure 1
Scatterplot of scores on phonemicsegmentationtask and fingerpoint-readingtask (with datapoints
distinguishedby whethersubjectswere able to segment at least two 3-phonemeblends)
g
Length of Blends
O CV/VC(N=18)
"8 U CV/VC+ at least
0- 7 - 2 CVC/VCC/CCV
"13 (N=12)
r- 60-
o ? ?
"13 5 0
S2- o
S 0
80 0
0- 000 000 0O
S 00 0
E 0 2 4 6 8 10 12 14 16 18 20 22 24
zNumber of Blends Segmented into Phonemes
changed: 6 noticed it spontaneously as they knowledge was the critical skill underlyingper-
were reading, 2 mentioned it when they were formance, very likely because it enabled sub-
questioned with the book closed, 2 noticed it jects to distinguish and rememberelements of
when questionedwith the book open, 18 did not the original letterconfiguration.
see it until the two books were presentedside- The distribution of scores on these two
by-side for comparison, and 8 never noticed it measures- detecting altered text and naming
at all. Thus, 26 subjects, or 72 percent, exhib- letters-were plotted in a scatter diagram, pre-
ited no memory for the letter configuration sented in Figure 2. Interestingly,letter knowl-
in the text they had practiced reading. This edge was highly tied to success or failurewhen
percentage is similar to results reported by subjects did not have the two texts side-by-side
Masonheimer et al. (1984), who found that to compare. The subjects who detected the al-
about two thirdsof the subjectswho could read terationspontaneouslyor when asked with the
environmentalprint failed to detect the differ- book either closed or open were subjects who
ence between original and altered signs (e.g., had named the majority of the letters. In con-
Xepsi vs. Pepsi), even when the signs were pre- trast, those who never recognizedthe alteration
sented side-by-side. knew fewer than 18 letters, and most of these
In the regressionanalysis for scores on rec- subjects knew fewer than 10 letters. This indi-
ognizing altered text, the best model included cates that near-masteryof letterswas crucial for
only one factor, letter-nameknowledge, which complete success in the detection task. Letter
accounted for 37 percent of the variance (see knowledge was less importantamong subjects
Table3). These results suggest that letter-name who detectedthe alterationwhen they could see
Figure 2
Scatterplot of scores on the letter-namingtask and alteredtext task
Low
0 4 0 @ Spontaneous
c)
.2 3 - Bookclosed
o2 Bookopen
x 1 * ** * *** : : Sidebyside
) 0 * * * ** Norecognition
d)
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Numberof Letters Named
the books side-by-side; these subjects varied was not centralto this task (r = .08).
greatly in their letter knowledge (see Figure 2). After the altered text task, subjects were
This is not surprising,because the information asked to read three different lines without the
needed to make the comparisonwas at hand, so pictures. Only 7 subjects(19%) read any of the
letter memory was not essential. lines correctly,and none readmorethan2 lines,
One subjectwith weak letter knowledgere- indicatingthat readinglines out of context was
sponded to this task in a way that revealedthat very difficult. All but one of the subjects who
he had not paid attentionto the letters during were able to readat least one line had displayed
fingerpoint-reading. During the latter part of word reading ability on the pretest, indicating
the task, when the altered and original lines that word readingability facilitatedmemory for
of print were exposed together, this subject re- lines of text.
positionedthe books in orderto see both sets of
pictures before answering the question about Word reading after training
whetherthe words were the same or different. Three word reading tests were given fol-
The alteredprinttask was similarto the let- lowing the fingerpoint-reading task. When
ter discrimination task in that both required words from the text were presentedin isolation,
readers to compare letters in two print speci- only 8 subjects(22 %)readany wordscorrectly,
mens. However,performanceswere not correl- and these subjects read no more than 4 apiece
ated (r = .28, p >.05), revealing that the (out of 12). All but one of these subjects had
ability to detect letter differences when two read at least one preprimerword on the pretest.
words are seen side-by-side did not account The low scores on this task reveal that even
for variance in performanceon the alteredtext novice word readersdid not learn to read many
task. Moreover,even subjects with weak letter individual words out of context as a result of
knowledgeperformedwell in the letter discrim- fingerpoint-reading practice. Several factors
ination task, indicating that letter knowledge may be responsible, including subjects'lack of
success in fingerpoint-reading many lines of read any preprimerwords in the pretest. All of
text, the rudimentarynature of these subjects' the subjects who read any words either in text
word readingskill, the greatersalience of other or in isolation (i.e., the circles and squares
non-printcues such as the pictures, and the ab- shown in Figure 3 for subjectswho read at least
sence of pressureto process print cues because 2 words correctly) also displayedsome phone-
the text had been memorized. Because 78 per- mic segmentationskill. In addition, there were
cent of the subjects received a word reading 14 subjectswho displayedsome phonemic seg-
score of zero, a regressionanalysis was not con- mentationskill but read no words (i.e., the cir-
ducted. cles shownfor subjectswho read0 wordsin text
Subjects read more words when the words but segmentedat least 1 blend). However,there
appearedin text. Of the 17 subjects who read were no subjects who read words but lacked
any words, 15 read more words in text than in phonemic segmentation skill. This indicates
isolation, indicating the facilitating effect of that having some phonemic segmentationskill
context (Goodman, 1965; Stanovich, 1986). In was necessary but not sufficient for learning to
the regression analysis, two factors explained readwords in text.
unique variance in the numberof words read in Observations of subjects' strategies for
text: the number of preprimer words subjects reading the words in text suggested that only
had read on the pretest, and the number of one subject identified words by reciting and
blends correctly segmented into phonemes (see pointing to all the preceding words in the line.
Table3). Thus, the relationship between word reading
A scatter plot of scores was examined to and phonemic segmentationis not explainedby
clarify this relationship.In Figure 3, datapoints skill in fingerpoint-readingthe text to deduce
are distinguishedaccordingto whether subjects words. A more common strategywhen subjects
Figure 3
Scatterplot of scores on the phonemic segmentationtask and word readingin text (with datapoints
distinguishedby whethersubjectsread any preprimerwords on pretest)
12
x PreprimerWords
m Read (N=12)
10
er O Not Read (N=18)
cis 88
" 6
O
l 0
4 - O
0 0 0
"2 O0
.0
0 2 4 6 8 10 12 14 16 18 20 22 24
Number of Blends Segmented into Phonemes
were uncertain about words was to recite the Success overall was greater when subjects
poem to themselves in an attempt to pluck a examined lines of text to locate single words
likely candidatefromthe set of lexical possibili- spoken by the experimenter than when they
ties. Two subjects were observed doing this read individualwords pointed to by the experi-
with a couple of words. However, most words menter.Fewersubjects(19 percent)failed com-
were read directly without any observable de- pletely in this task than had failed at reading
ductive processing, indicating that subjects words in text (58 percent). In the regression
were using printcues to retrievethe wordsfrom analysis, two factorsaccountedfor uniquevari-
memory. ance in word location scores: preprimerword
Phonemic segmentation skill may have reading ability and letter-name knowledge.
provedimportantfor rememberinghow to read Wordreadingability may have enabledsubjects
words for two reasons. It may have helped sub- to rememberhow to read the words or how to
jects store words in memory to read them, as a readneighboringwords, which helped them de-
process of forming access routes into memory duce the location of targetwords. Knowinglet-
by linking letters in spellings to phonemes in ter names may have enabled subjects to locate
pronunciations(Ehri, 1987, in press). Alterna- words using these cues. One child verbalized
tively, phonemic segmentation may have en- this strategy,saying that 'paddled was easy to
abled subjectsto pluck words covertly from the find because of the p." Six subjects used the
memorized lines by helping them analyze first method of reciting the entire line, pointing to
soundsin words. The fact thatword readingwas each word in turnuntil they arrivedat the target
better in context than in isolation indicatesthat word.
subjectswere using contextin some way to read A scatterplot of scores was constructedto
the words. Decoding skill does not explain the clarify these relationships.In Figure4, the data
relationshipbecause all of the subjectswere un- points distinguish subjects according to their
skilled decoders, and the words were not easily preprimerword readingskill. It is apparentthat
decoded. subjects who were most successful in locating
Inspection of word misreadings revealed words in text (i.e., those who located more than
that many subjects respondedwith whole lines half the wordscorrectly)were mostly preprimer
when asked to read single words. For example, word readerswho knew almost all the letters. In
subjects misread the word blue as where the contrast,subjectswho were partiallysuccessful
river runs blue. The weaker readerswere more at locating words showed a range of letter
likely to respondwith whole lines thanwere the knowledge.
other subjects. Of the subjectswho had readno
preprimerwords and knew fewer than 15 letter
names (n = 12), 92 percent responded with
lines rather than words at least once. In con-
trast, of those who had read some preprimer
Discussion
wordsand knew most of the letters, only 33 per-
cent gave entire lines ratherthan words. Lines Resultsof the presentstudy indicatethat, in
may have been given for several reasons: Sub- orderfor emergentreadersto achieve some suc-
jects may have been "reading" the pictures cess learning to fingerpoint-read memorized
rather than the print, or they may have asso- text and to remember information about the
ciated lines rather than words with the letter print from this activity, they need to possess
sequences, or they may have had difficulty dis- some print-relatedknowledge. Phonemic seg-
tinguishing word boundaries in speech. Line mentation skill enables beginners to learn to
reading appeared to be a default response fingerpoint-readmemorizedtext and to remem-
among subjects who could read some words, ber how to read individual words in the text.
for subjects did not recite lines when they were Also, being able to read even a few preprimer
able to read single wordscorrectly. words enables beginners to rememberhow to
Figure 4
Scatter plot of scores on the letter-namingtask and on locating words in text (with data points
distinguishedby whethersubjectsreadany preprimerwords on pretest)
10
x PreprimerWords
9
Read (N=12)
.• 8 0 Not Read (N=18)
a 7
o 6 O-
".J 5 -
4-
0
3- 0 000
0 2 0 00 0O
I- OU0
8 0 o8
:0 0 0 0 0 00 0
0 2 4 6 8 10 12 14 16 18 20 22 24 26
Number of Letters Named
read individualwords in the text. Knowing let- had expected this factor to have a greater im-
ter names enables beginners to locate words in pact.
lines of text they have read and to recognize Perhapsif we had measuredsubjects'abil-
when letters in the text have been altered. ity to use letters for spelling words, as Morris
One limitation of this study should be did (1983, 1989), or subjects' letter naming
noted. Because results are correlational,causal speed (Adams, 1990), or their knowledge of
interpretationsof the data must remain tenta- letter-sound correspondences, then letter
tive. Furtherexperimentalresearchis needed to knowledge might have accounted for unique
show that trainingin these capabilitiesbenefits variancein other dependentmeasures. Alterna-
fingerpoint-readingin the ways described. tively, perhapsletterknowledgewas less impor-
One purpose of the study was to examine tant for fingerpoint-readingbecause subjects
the importance of letter-name knowledge for did not have to rely exclusively on letter cues in
fingerpoint-readingand word reading. Results print to read correctly. Instead they could use
indicatedthat letter knowledgeenabled subjects their memory for the text, the pictures accom-
to locate individual words spoken by the ex- panying the text, and white spaces separating
perimenterin text, but it did not enable them letter sequences in the lines of text. Phonemic
to rememberhow to read individualwords, nor segmentation skill, by contrast, may have
to fingerpoint-read memorized text. Because provedof uniqueimportancebecause therewere
letter-nameknowledge is a powerful predictor no other means available for analyzing units in
of readingachievement(Shareet al., 1984), we speech correctly.
Phonemic segmentation skill contributed conclusions differ somewhat from those drawn
not only to subjects'ability to rememberhow to by Morris (1989). Morris interpretedhis longi-
read individual words in text, but also to their tudinal data as indicatingcausal links between
abilityto matchprintwith speech duringfinger- successively developing capabilities. He con-
point-reading. It proved even more important cluded that knowledge of initial consonantlet-
than preprimer word reading skill for finger- ters/sounds in words enables mastery of the
point-readingperformance,indicatingthatpho- ability to matchwrittenwords to spokenwords,
nemic units may be more centralin learningthis which in turnenables masteryof phonemicseg-
operationthan lexical units. Phonemicsegmen- mentation,which enables masteryof word read-
tation skill has been found to be an important ing. Our data suggest that letter knowledge,
contributorto reading skill by many other re- phonemic segmentationskill, and word reading
searchers(Ehri, 1979; Juel, Griffith, & Gough, ability each enable differentaspects of the abil-
1986; Liberman, Shankweiler, Liberman, ity to matchwrittenwordsand spoken words.
Fowler, & Fischer, 1977; Lundberg, Frost, & Differences in the difficulty levels of the
Petersen, 1988; Morris, 1983; Perfetti, Beck, tasks may accountfor some of the disparitybe-
Bell, & Hughs, 1987; Share,Jorm, Maclean, & tween Morris's(1989) conclusions and our own.
Matthews, 1984; Stanovich, Cunningham, & Morris measured more advanced phonemic
Feeman, 1984; Treiman& Baron, 1983; Tun- segmentation,whereaswe looked at a rudimen-
mer & Nesdale, 1982, 1985; Vellutino & tary form, possibly more comparable to his
Scanlon, 1987). The measureof phonemic seg- word-initialconsonanttask. Also, it may be that
mentation used here differed from measures causal relationshipsbetween these variablesare
used by others. Also, the subjects tested were reciprocal rather than unidirectional (Ehri,
younger and less literate than those in other 1979; Stanovich, 1986): Entrylevels of phone-
studies. mic segmentation and word reading abilities
In our phonemic task, preschoolers were may enable various aspects of fingerpoint-
taughtto associate four picturesof mouthposi- reading, which in turn may promote further
tions with isolated sounds and then to segment developmentof segmentationand word reading.
the soundsin wordsby pronouncingeach sound The importance of letter knowledge for
separatelyand selecting the appropriatemouth processingprintduringfingerpoint-readingwas
picture. Lindamoodand Lindamood(1975) in- also measured. We found that subjects who
cluded such a task in their programto teach au- were able to detectletter alterationsin text prior
ditory conceptualizationskill. We thought that to seeing the original and altered texts side-
this activity wouldprovidechildrenwith a more by-side were those who knew most of the letters
concrete method of analyzing sounds than the (see Figure 2). Our explanation is that letter
method of counting or tapping out phonemes knowledge providedthe basis for remembering
used by others, and hence might be more sensi- elements of the print configuration. It enabled
tive to individualdifferences between nonread- subjects to store aspects of the original text in
ers who lack spelling knowledgeto help them in memory, such as the first letter or the first and
these tasks (Ehri & Wilce, 1980; Tunmer & final letters in salient words. However,an alter-
Nesdale, 1982). From Table 1 and Figure 3, it native interpretationneeds to be considered.
is apparentthat we did create a task that could Subjectsmay have recognizedthe orthographic
be grasped by the least mature readers. Most abnormality or illegality of the altered text
(86 percent) were able to learn the picture- (e.g., WSWSAL WOHOEWNIH).Three facts
sound associations to criterion and to segment make such an explanation unlikely. First, the
some blends into phoneme segments. print was all in capital letters, so there were no
One purpose of the present study was to deviant word shapes signaling that something
discover the relations between beginners' was amiss. Second, most if not all of our sub-
knowledge of print and their success in various jects had insufficient experience with conven-
aspects of the fingerpoint-reading task. Our tional print to be sensitive to orthographic
structure. Third, research findings have indi- jects in Masonheimeret al.'s(1984) studyand in
cated that the ability to distinguish legal from the presentstudy did not notice letter alterations
illegal orthographicsequences does not emerge is that they did not know letters well enough to
typically until second grade (Allington, 1978; store the print in memory in order to detect the
Gibson & Levin, 1975; Juola, Schadler, Cha- changes.
bot, & McCaughey, 1978; Leslie & Shannon, Findings in the present study also indicate
1981; Leslie & Thimke, 1986; Massaro& Hes- that beginnersdo not rememberhow to read in-
tand, 1983). Our subjects were in preschool or dividual words from letter cues as a result of
kindergartenand had not even begun formal simply being exposed to the written words and
reading instruction.We conclude that sensitiv- their spoken matches in the course of finger-
ity to orthographicstructure does not explain point-reading memorized text. In the present
why subjects with letter knowledge were better study, it was primarilythose subjectswho dem-
able to detect letter changes in the text. The onstratedword reading ability prior to the ex-
more likely explanationis that letter knowledge perimentwho rememberedhow to read any of
enabled subjectsto rememberwhat the original the targetwords when the wordswere presented
text had looked like. in isolation. It is noteworthy,in fact, that these
The presentfindings suggest resolutionof a subjects did learn how to read some individual
discrepancy arising in two previous studies. words, even though there were plenty of other
Masonheimeret al. (1984) found that children non-printcues available, such as picturesand a
who were experts at reading environmental well-rehearsedtext. These results suggest that
print did not detect letter alterationsin familiar word reading memory may require a special
signs (e.g., Xepsi for Pepsi) and concluded that mechanism. The unprepared mind does not
these children paid little attention to letters. pick up words and rememberhow to read them
However,McGee, Lomax, and Head (1988) ob- spontaneously.On the other hand, the prepared
served environmentalprint readerspoint to and mind does not avoid using printcues to remem-
comment on letters in signs, even when they ber how to read words, even when there are
could not readthe signs, indicatingthatthey did other non-printcues available. Subjects in the
pay attention to letters. The present findings present study were novice readerslacking well-
suggest a possible distinction between paying developed word memory mechanisms, so the
attentionto letters and being able to remember strong showing of this variableis noteworthy.
letters. Further research may be needed to estab-
In the letter discriminationtask, most sub- lish the generalizability of our findings. Our
jects were able to attendto lettersto distinguish text was printed in uppercase letters, because
similarly spelled words. Yet, in the alteredtext beginners are known to be more familiar with
task, only those with substantialletter knowl- uppercasethan lowercaseletters (Smythe, Sten-
edge had sufficient memory of the original text nett, Hardy, & Wilson, 1970-1971), and be-
to recognize letter alterations when the texts cause uppercase letters are easier to
were not side-by-side. The correlationbetween discriminate. However, most big books are
these two tasks was low and was not signifi- printed in a mix of upper and lower case. We
cantly different from zero. These results show believe that the processes and relationshipsob-
that attendingto letters in print is not the same served in this studywould be apparentalso with
capability as being able to remember letters. text printed conventionally. However, this re-
Immaturereaders may notice letters and even mains to be verified.
name them, but unless they know letters well, In considering the implications of the
they will not be able to store letter information presentfindings for early readinginstruction,it
in memory. It is letter memory, not letter dis- is importantto recognize thatwe examinedonly
crimination,that is critical for developingread- the effects of short-termpracticeon fingerpoint-
ing skill (Ehri, 1987; Ehri & Wilce, 1985, reading, in a study conductedin the laboratory.
1987a, 1987b). Our explanationfor why sub- The practicethat Holdaway(1979) advocatesis
long-term, extending over 3 months, during exception: The teacher-instructed group did
which time "sixty books, twenty poems, ten better on the concepts-about-printtest. These
songs, and a chant or two"might be processed. resultsindicatethatimmersionin printis suffic-
Moreover,advocatesof memorizedtext reading ient for moving childreninto reading.However,
do not claim that practice alone moves child- it shouldbe notedthat immersionin this experi-
ren closer to independentreading. Holdaway's ment went beyond the simple practice effects
(1979) shared book experience includes more examined in the present study. Although the
thanjust practice; in addition, teachers are di- teacher-ledinstructionwas designed as part of
rected to model fingerpoint-reading,to explain only one of the treatments, it is likely that
and discuss the structureof alphabeticprint, to teacher explanation and modeling were also
draw attentionto printedwords, empty spaces, presentin the other immersioncondition, given
and letter-sound associations, and to provide the kinds of activities that studentsperformed.
corrective feedback and reinforcementfor stu- In conclusion, our results should not be in-
dents'efforts. terpretedas challenging the use of memorized
What the results of this study and the one text reading in classrooms. On the contrary,
by Morris (1989) indicateis thatbeginnersmay other researchindicatesthatthis is a very effec-
be unableto make much progressin various as- tive means of moving children into reading.
pects of memorized text reading without also Clay (1979) has incorporatedthe method into
acquiring specific print-related knowledge. her Reading Recovery program. It may be the
Holdaway(1979) and othersare less clear about most effective way to bring at-risk youngsters
the importanceof these skills for making pro- with limited home literacy experiences to the
gress, and when they are needed. To validate brinkof reading(Masonet al., 1990). Whatour
our conclusions, trainingstudies should be per- research does is to point out how children's
formedin which memorizedtext readingis con- success with such proceduresis influenced by
tinued over a longer period of time. Not only what they already know about print. It may be
practice but also instruction in print-related important to ensure that these skills are in
skills might be manipulatedexperimentallyto place before children are expected to display
assess their respective contributionsto reading fingerpoint-readingon their own. If childrendo
acquisition. not know lettersand do not have a graspof rudi-
Although the design of their study was mentaryphonemic segmentation,they will have
quasi-experimental,Reutzel, Oda, and Moore trouble performing the necessary operations.
(1989) have reportedone trainingstudy lasting However,this is not to say that childrenlacking
6 months. They compareda traditional"readi- these skills should not be exposed to
ness"kindergartenclassroomto a classroomus- fingerpoint-reading experiences. There are
ing an "immersion in print" approach and a many other benefits to be gained from teacher-
classroom using immersionplus teacher-ledin- modeled memorized text reading, such as
struction. The immersion approach included left-to-rightorientation,familiaritywith written
shared book experience following Holdaway's languagestructures,and realizationthat spoken
(1979) recommendations,plus language exper- language corresponds to written language.
ience and writing using invented spelling. Moreover, such advance exposure may enable
Teacher-ledinstructionconsisted of "short,ex- beginners to catch on very quickly to
plicit lessons on selected print concepts during fingerpoint-readingonce the enabling skills are
the daily rereadingof a big book."This included acquired.
instruction on "concept of word" and how to
identify words in text. Both immersion groups REFERENCES
ADAMS, M.J. (1990). to read: Thinking and
were found to surpass the readiness group on Beginning
learning aboutprint. Cambridge,MA: MIT Press.
print awarenessmeasures, readinesstests, and ALLINGTON, R.L.(1978).Sensitivityto orthographic
struc-
word readingtests. The two immersion groups tureas a functionof gradeandreadingability.Journal
performedequally well on these tests, with one of ReadingBehavior,10, 437-439.
CLAY,M.M.(1972). Reading: Thepatterningof complexbe- LUNDBERG, I., FROST, J., & PETERSEN, O. (1988).Effects of
haviour.Auckland, New Zealand:Heinemann. an extensive program for stimulating phonological
CLAY,M.M.(1979). The early detection of reading difficul- awareness in preschool children. Reading Research
ties. Auckland, New Zealand:Heinemann. Quarterly,23, 263-284.
DYSON, A.H. (1984). Learning to write/learning to do MARTIN, B., & BROGAN, P. (1972). Teacher's guide, instant
school: Emergent writers'interpretationsof school lit- readers. New York:Holt, Rinehart& Winston.
eracy tasks. Research in the Teachingof English, 18, MASON,J., & ALLEN,J. (1986). A review of emergent liter-
233-264. acy with implicationsfor researchand practice in read-
EHRI,L.C.(1979). Linguistic insight: Thresholdof reading ing. In E.Z. Rothkopf (Ed.), Review of research in
acquisition. In T.G. Waller& G.E. MacKinnon(Eds.), education(Vol. 13, pp. 3-47). Washington,DC: Ameri-
Reading research: Advances in theory and practice can EducationalResearchAssociation.
(Vol. 1, pp. 63-114). New York:AcademicPress. MASON, J., KERR, B., SINHA, S., & MCCORMICK, C. (1990).
EHRI,L.C. (1987). Learningto read and spell words. Jour- Shared book reading in an Early Start program for
nal of ReadingBehavior, 19, 5-31. at-risk children. In J. Zutell, S. McCormick, M.
EHRI,L.C.(in press). Reconceptualizingthe developmentof Connolly, & P. O'Keefe(Eds.), Literacytheoryand re-
sight word reading and its relationshipto recoding. In search: Analysesfrom multipleparadigms. Thirty-ninth
P.B. Gough, L.C. Ehri, & R. Treiman(Eds.), Reading yearbookof the National Reading Conference(pp. 189-
acquisition. Hillsdale, NJ: Erlbaum. 198). Chicago, IL: National ReadingConference.
EHRI, L.C., & WILCE, L.S. (1980). The influence of orthog- MASONHEIMER, P.E., DRUM, P.A., & EHRI, L.C. (1984).
raphy on readers' conceptualization of the phonemic Does environmentalprint identification lead children
structureof words. Applied Psycholinguistics, 1, 371- into word reading? Journal of Reading Behavior, 16,
385. 257-272.
EHRI, L.C., & WILCE, L.S. (1985). Movementinto reading: MASSARO, D., & HESTAND, J. (1983). Developmentalrela-
Is the first stage of printedword learningvisual or pho- tions between reading ability and knowledge of ortho-
netic? ReadingResearch Quarterly,20, 163-179. graphic structure. Contemporary Educational
EHRI, L.C., & WILCE, L.S. (1987a). Cipher versus cue read- Psychology, 8, 174-180.
ing: An experimentin decoding acquisition. Journal of MCGEE, L.M., LOMAX, R.G., & HEAD, M.H. (1988). Young
EducationalPsychology, 79, 3-13. children'swritten language knowledge: What environ-
EHRI, L.C., & WILCE, L.S. (1987b). Does learning to spell mental and functionalprint reading reveals. Journal of
help beginners learn to read words? Reading Research ReadingBehavior,20, 99-118.
Quarterly,22, 47-65. MORRIS,D. (1983). Concept of word and phoneme aware-
GIBSON, E., & LEVIN, H. (1975). Thepsychology of reading. ness in the beginning reader. Research in the Teaching
Cambridge,MA: MIT Press. of English, 17, 359-373.
GOODMAN,K.S. (1965). A linguistic study of cues and mis- MORRIS, D. (1989). The relationshipbetween word aware-
cues in reading.ElementaryEnglish, 42, 639-643. ness and phonemeawareness in learningto read:A lon-
HOLDAWAY, D. (1979). The foundations of literacy. Sidney, gitudinal study in kindergarten.Manuscriptsubmitted
Australia:Ashton Scholastic. for publication.
JUEL, C., GRIFFITH, P.L., & GOUGH, P.B. (1986). Acquisi- MORRIS, D. (in press). Concept of word: A pivotal under-
tion of literacy: A longitudinalstudy of childrenin first standingin the learningto read process. In E. Hender-
and second grade. Journal of EducationalPsychology, son, S. Templeton, & D. Bear (Eds.), Developmentof
78, 243-255. orthographic knowledge: The foundations of literacy.
JUOLA, J., SCHADLER, M., CHABOT, R., & MCCAUGHEY, M. Hillsdale, NJ: Erlbaum.
(1978). The development of visual information- MORRIS, D., & HENDERSON, E.H. (1981). Assessing the be-
processing skills related to reading. Journal of ginning reader's'concept of word' Reading World,20,
ExperimentalChildPsychology, 25, 459-476. 279-285.
LESLIE, L., & SHANNON, A. (1981). Recognition of ortho- PERFETTI, C.A., BECK, I., BELL, L., & HUGHS, C. (1987).
graphic structureduring beginning reading. Journal of Phonemic knowledge and learning to read are recip-
ReadingBehavior, 13, 313-324. rocal: A longitudinal study of first grade children.
LESLIE, L., & THIMKE, B. (1986). The use of orthographic Merrill-PalmerQuarterly,33, 283-319.
knowledge in beginning reading. Journal of Reading REUTZEL, D.R., ODA, L.K., & MOORE, B.H. (1989). Devel-
Behavior, 18, 229-241. oping print awareness:The effect of three instructional
LIBERMAN, I.Y., SHANKWEILER, D., LIBERMAN, A.M., approacheson kindergarteners' printawareness,reading
FOWLER, C., & FISCHER, EW. (1977). Phonetic segmen- readiness, and wordreading.Journalof ReadingBehav-
tation and recoding in the beginning reader. In A.S. ior, 21, 197-217.
Reber & D.L. Scarborough(Eds.), Towarda psychol- SHARE, D.L., JORM, A.E, MACLEAN, R., & MATTHEWS, R.
ogy of reading (pp. 207-225). Hillsdale, NJ: Erlbaum. (1984). Sources of individualdifferencesin readingac-
LINDAMOOD, C.H., & LINDAMOOD, P.C. (1975). The audi- quisition.Journal of EducationalPsychology, 76, 1309-
tory discrimination in-depth program. Boston, MA: 1324.
TeachingResourcesCorporation. SLOSSON ORAL READING TEST. (1963). East Aurora, NY:
Slosson EducationalPublications.
SMYTHE, P.C., STENNETT, R.G., HARDY, M., & WILSON, H.R. TUNMER, W., & NESDALE, A.R. (1985). Phonemicsegmenta-
(1970-1971). Developmental patterns in elemental tion skill and beginningreading.Journal of Educational
skills: Knowledge of upper-case and lower-case letter Psychology, 77, 417-427.
names. Journalof ReadingBehavior,3, 24-33. VELLUTINO, .ER., & SCANLON, D.M. (1987). Phonological
K.E. (1986). Matthew effects in reading: Some
STANOVICH, coding, phonological awareness, and reading ability:
consequences of individual differences in the acqui- Evidence from a longitudinaland experimentalstudy.
sition of literacy. Reading Research Quarterly, 21, Merrill-PalmerQuarterly,33, 321-363.
360-406. YADEN, D.B., & TEMPLETON,S. (Eds.). (1986). Metalin-
STANOVICH, K.E., CUNNINGHAM, A.E., & FEEMAN, D.J. guistic awareness and beginning literacy. Portsmouth,
(1984). Intelligence, cognitive skills, and early reading NH: Heinemann.
progress. ReadingResearchQuarterly,19, 178-303.
SULZBY,E. (1985). Children'semergent readingof favorite
storybooks: A developmentalstudy. Reading Research Footnotes
Quarterly,20, 458-481. This research was supported by Grant No. HD-23719
TAYLOR,N.E. (1986). Developing beginning literacy con- awardedby the National Instituteof Child Health and Hu-
cepts: Contentand context. In D.B. Yaden, & S. Tem- man Development.
pleton (Eds.), Metalinguisticawareness and beginning We express our gratitude to the Davis Parent Nursery
literacy (pp. 173-184). Portsmouth,NH: Heinemann. School, the Redbug Montessori School, The Children's
TEALE, W.H., & SULZBY, E. (1987). The culturalpracticeof Center at North Davis ElementarySchool, and the Davis
storybook reading:Its effects on young children'sliter- Joint Unified School District for their cooperationand as-
acy development.In D.A. Wagner(Ed.), Thefuture of sistance in the conductof this study.
literacy in a changing world (pp. 111-130). New York: Requests for reprintsshould be sent to Linnea C. Ehri,
Pergamon. PhD Programin EducationalPsychology, CUNY Graduate
TREIMAN, R., & BARON, J. (1983). Phonemicanalysis train- School, 33 West 42nd St., New York, NY 10036-8099
ing helps children benefit from spelling-sound rules. (USA).
Memoryand Cognition, 11, 382-389.
TUNMER, W., & NESDALE, A.R. (1982). The effects of di- Received October30, 1989
graphs and pseudowordson phonemic segmentationin Revisionreceived August 23, 1990
young children.AppliedPsycholinguistics,3, 299-311. Final revision acceptedApril 9, 1991