Learning to read and spell words

http://jlr.sagepub.com/Journal of Literacy Research

http://jlr.sagepub.com/content/19/1/5The online version of this article can be found at:

DOI: 10.1080/10862968709547585

1987 19: 5Journal of Literacy ResearchLinnea C. Ehri

Learning to Read and Spell Words

Published by:

http://www.sagepublications.com

On behalf of:

Literary Research Association

can be found at:Journal of Literacy ResearchAdditional services and information for

http://jlr.sagepub.com/cgi/alertsEmail Alerts:

http://jlr.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://jlr.sagepub.com/content/19/1/5.refs.htmlCitations:

What is This?

- Mar 1, 1987Version of Record >>

at NATIONAL CHUNG HSING UNIV on April 10, 2014jlr.sagepub.comDownloaded from at NATIONAL CHUNG HSING UNIV on April 10, 2014jlr.sagepub.comDownloaded from

http://jlr.sagepub.com/


http://jlr.sagepub.com/content/19/1/5

http://jlr.sagepub.com/content/19/1/5



http://www.nrconline.org

http://www.nrconline.org

http://jlr.sagepub.com/cgi/alerts

http://jlr.sagepub.com/cgi/alerts

http://jlr.sagepub.com/subscriptions

http://jlr.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.sagepub.com/journalsPermissions.nav

http://jlr.sagepub.com/content/19/1/5.refs.html

http://jlr.sagepub.com/content/19/1/5.refs.html

http://jlr.sagepub.com/content/19/1/5.full.pdf

http://jlr.sagepub.com/content/19/1/5.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml

http://online.sagepub.com/site/sphelp/vorhelp.xhtml





Journal of Reading Behavior1987, Volume XIX, No. 1

LEARNING TO READ AND SPELL WORDS

Linnea C. EhriUniversity of California, School of Education, Davis, CA 95616

ABSTRACT

Learning to read and spell words is a central part of becoming literate. During textreading, most words are processed, and skilled readers are able to do thiseffortlessly. How they become skilled at processing graphic cues has been the focusof our research. Findings indicate that prereaders do not acquire graphic skill bylearning to read signs and labels in their environment. Rather, mastery of letters isrequired. Whereas prereaders use visual or context cues to identify words, as soonas children move into reading they shift to letter-sound cues. Initially, words areread by accessing remembered associations between a few letters in spellings andsounds in pronunciations. Later, when decoding skill matures, complete spellingsare analyzed as phonemic symbols for pronunciations and are stored in memory.Various studies indicate that having a visual picture of speech in memory is animportant part of a person's information-processing equipment. Spellings mayinfluence how words are pronounced, what sounds people think are in words, howquickly people judge spoken word rhymes, how rapidly pronunciations changeover time.

Although the great debate about methods of teaching reading may be over, itis not true that all the mysteries of beginning reading instruction have been solved.There are plenty inviting attention. We know that code emphasis programsproduce better beginning readers than meaning emphasis programs (Chall, 1967).However, we do not yet understand how or exactly why. There are many ways tostructure phonics instruction. Sounding out and blending may or may not betaught. Segmenting words into phonemes may or may not receive attention.1

1The terms phonetic, phonemic, and phonology are used to refer to that aspect of languageinvolving speech sounds (i.e., phones and phonemes) and their structure. Although the terms phoneticand phonemic have different meanings linguistically, psychologists often ignore the distinction andregard the terms as synonymous. This is done here. The term phonemic awareness refers to a

(Continued on page 6)

5 at NATIONAL CHUNG HSING UNIV on April 10, 2014jlr.sagepub.comDownloaded from



Journal of Reading Behavior

Children may be taught letter-sound rules but then they may not be given practicereading words that exhibit those rules. They may even practice reading many wordsthat violate the rules they have been taught (Juel & Roper/Schneider, 1985). Weneed to understand whether and how these experiences make a difference.

The case is not that meaning emphasis programs are totally ineffective. Somechildren appear to acquire decoding skills on their own without being formallytaught. We need to study how this happens.

We do not understand the relationship between reading instruction andspelling instruction. Some children are taught to spell right away in first grade assoon as reading begins. Others do not begin spelling until a year later. It may be thatspelling instruction operates behind the scenes to strengthen decoding skill (Ehri, inpress).

A point to note about English is that its spellings are not completely phonetic.Although the majority of letters in words correspond to sounds, there are enoughexceptions and variations to complicate the course of acquisition. How dobeginning readers and spellers handle phonetic anomalies in word spellings? Dothey ignore the letters, or give them special pronunciations, or mark them as silentin memory? At some point, readers and spellers must graduate from a phoneticunderstanding of spellings to a morphemic understanding where they recognizespelling patterns that recur in words to make them systematic. Knowing this helpslearners make sense of many letters that previously appeared to be phoneticallydeviant. We do not understand how or when this transition occurs in beginningreaders. These and many more questions need to be studied.

Our intent in this paper is to review the issues we have addressed in ourresearch on printed word learning and to speculate about what our findingssuggest. In a preface, we justify our focus on word learning by describing the centralrole of word identification in learning to read. Also, we provide a developmentalcontext for this work by presenting a very general outline of the stages of readingand spelling acquisition. In the core of the paper, we describe how children attaincompetence at processing alphabetic cues in learning to read and spell words. Wealso show how learning to read and spell are major events influencing the course ofspoken language development. To avoid misinterpretation, it is important to notethat our work focuses on processes that lie behind learning to read and spell, notupon methods of instruction. Although we study how beginners learn aboutprinted words, we are not suggesting that classroom instruction should consist of

metalinguistic skill involving the ability to focus on and manipulate sounds in speech, for example,segmenting words into sounds, or blending sounds to form words. Speakers, particularly youngchildren, may have full control over the production of phones and phonemes in their meaningful speechyet they may lack conscious awareness of these units of language and how they are sequenced to formwords.

6

at NATIONAL CHUNG HSING UNIV on April 10, 2014jlr.sagepub.comDownloaded from



Learning to Read and Spell

word learning and word recall. Our research is at least twice removed frominstruction. If readers are interested in details of the research, they should consultthe original studies.

Word Identification During Text Reading

From eye movement data (Carpenter & Just, 1981; McConkie& Zola, 1981),we know that when mature readers read text, they do not pass over words and lookonly at those that they cannot predict from context. Rather, their eyes fixate onmost of the words. Less mature readers fixate on every word, often more than once.Also, analyses of oral reading errors produced by children reveal that every word isprocessed (Weber, 1970). The most common error is to substitute other words forthose appearing in text. This happens much more often than omitting words orinserting extra words. Both lines of research suggest that the text reading processincludes a step for fixating and identifying each printed word as it comes up.

How are these words identified? Our view and that of others is that readersaccess various sources of information stored in memory. Readers possess semanticand syntactic information that enables them to form expectations about upcomingwords in text (Goodman, 1970). They may expect words from a particular semanticcategory, such as a kind of animal, or from a particular syntactic form class, such asa noun or verb. These expectations are derived from their knowledge of the world,from knowledge about the structure of their language, and from informationalready read in the text. In identifying a word, readers can use semantic andsyntactic expectations to predict what the word is or at least what features it has. Inaddition, these expectations can be used to confirm words identified by othersources.

Readers also possess two sources of graphic information in memory thatenable them to identify words from their printed forms. One source is the reader'sstore of printed words, often referred to as a print lexicon or a sight vocabulary.The word store contains the spellings of individual words associated with theirpronunciations and meanings. When words in this store are encountered in print,the printed form activates all this information in memory and results in immediaterecognition of the word (Ehri, 1978, 1980a, 1980b; Ehri & Roberts, 1979; Ehri &Wilce, 1980a). Also, information about specific word spellings can be used toidentify unknown words by application of an analogy strategy which involvesrecognizing shared letters in a common stem (e.g., reading "slough" by analogy to"rough") (Glushko, 1979, 1981; Kay & Marcel, 1981; Marcel, 1980; Marsh,Friedman, Welch, & Desberg, 1981).

The other source of graphic information is the reader's general knowledge ofhow the orthographic system maps speech. This information may be used toidentify unfamiliar words by transforming letters into blended sounds. In addition,it may be used to confirm words identified by other sources. For example, the

7




Journal of Reading Behavior

reader might use semantic expectations to guess that the word is "kitten" and thencheck the letter-sound correspondences to make sure. Some or all of these sourcesof information may be involved in identifying any particular word during textreading. When more than one source yields the same identification, this createsredundancy in the system. Such redundancy is thought to enhance the accuracyand fluency of the text reading process (Perfetti, 1984).

One of the issues which has divided researchers is the question of whetherindividual words are read or only sampled during text reading (Stanovich, 1980).The main objection to individual word reading appears to be that it consumes toomuch attention, effort and time to execute. However, if one examines the nature ofword reading skill and the course of its development, one finds that this is not true.According to LaBerge and Samuels (1974), as interpreted by Ehri and Wilce (1979b,1983), learning to read individual words passes through three successive phases.During Phase 1, an unfamiliar word is recognized with increasing accuracy asreaders attend to letter-sound relations each time they read it. In Phase 2, as a resultof more practice, a familiar word comes to be recognized automatically as a whole,without attention and without deliberate processing of component letter-sounds.In Phase 3, the word comes to be recognized with increasing speed as identificationprocesses are consolidated in memory. It is following Phase 2 when readers canrecognize words automatically that the words should be easy to recognize in text,because recognition requires little effort.

To experience what it means to recognize words automatically, try labeling thepictures in Figure 1. Name the pictures as rapidly as you can. Ignore the words

Buggy

Figure I. Pictures printed with semantically related words that create interferencein a task requiring subjects to name the pictures as rapidly as possible andto ignore the printed words.

8




Learning to Read and Spell

printed on the pictures. You will find that the distracting words are hard to resist,and they definitely slow you down (Rosinski, Golinkoff, & Kukish, 1975). Eventhough you do not say them aloud, you know what words are there and what theymean. Guttentag and Haith (1978) have used this task and found that children asyoung as the end of first grade can process known words automatically. The point isthat if readers can recognize words this easily when they are wishing to ignore them,then they certainly can recognize the same words easily when they encounter themin text. Thus, individual word reading is not an obtrusive or time-consumingprocess.

From our findings and those of others, it is clear that learning to processgraphic cues accurately, automatically, and rapidly is one of the hardest parts oflearning to read. It is a part that consumes substantial learning time. It is a skill thatclearly separates good readers from poor readers (Stanovich, 1980). Thus, it isespecially important to understand how beginners become skilled at graphic cueprocessing.

Stages of Reading and Spelling Acquisition

Several events occur during the first three stages in learning to read and spell(Chall, 1979). Stage 0 covers the prereading period from birth until childrenbecome able to read print. During this stage, children acquire oral language skills.They are exposed to and begin to participate in the activities of a literate culture.They acquire concepts about reading and writing, for example, what print lookslike, where it is found, how to write their names, and so on. They learn the shapesand names of alphabet letters and how letters differ from numbers. They learn toidentify signs in their environment such as McDonald's, Coca Cola, and Stop. Theypractice writing by scribbling or copying print or inventing their own spellings tolabel drawings or to send messages (Mason & Allen, in press).

Stage 1 is the initial reading and decoding stage. Children learn how tosegment meaningful language into words and phonemes. They learn how letterssymbolize phonemes in words. They begin to acquire a sight vocabulary and usethis to read simple text. Based on Biemiller's (1970) research, Chall (1979)distinguishes two phases of text reading during this stage. During the first phasewhen readers are reading text and encounter unfamiliar words, they use contextcues to guess what the words are. Although their misreadings preserve syntacticand semantic information in the text, what they say bears little resemblance to theprinted letters. However, once readers learn more about letter-sound relations, theybegin attempting to decode letters. Chall (1979) refers to this second phase as onewhen readers become glued to print rather than to meaning, and she suggests thatchildren must go through this phase in order to move into the next phase when theybecome unglued from print and able to attend both to graphic and to contextualcues in their reading. This ungluing happens during Stage 2 when readers' decoding

9




10 Journal of Reading Behavior

skills grow stronger. Also during Stage 2, readers become able to read many morewords accurately, automatically, and rapidly.

In the development of spelling skill, three stages corresponding to the readingstages can be distinguished (Ehri, in press; Gentry, 1982; Henderson, 1981; Morris,1981). Once prereaders have learned letter names during Stage 0, they can use thisinformation to invent semi-phonetic spellings of words. They distinguish one ortwo sounds, usually the first or last, and represent these with letters. The letters theychoose are ones whose names contain these sounds, for example, "giraffe" spelledJ-F. When children move into Stage 1 and learn more about letter-sound relationsand about phonemic segmentation and decoding, their spellings become morecomplete phonetically and the letters they choose are more conventional, forexample, "giraffe" spelled GERAF. During Stage 2, as children learn more aboutthe spelling patterns of English words, their commitment to the principle of oneletter for every sound is relaxed, and they adopt morphemic as well as phoneticpatterns to spell. For example, in writing endings on past tense verbs, they use theletters E-D consistently regardless of whether the final sound is / t / as in the word"stepped," or /d/ as in "jogged," or /sd/ as in "sprinted."

These spelling stages portray how children invent spellings when theconventional forms of words are unknown. Another aspect of spelling acquisitioninvolves storing the spellings of specific words in memory. This occurs whenlearners interpret letters as symbols for sounds in the pronunciations of words(Ehri, 1984,1985). Not only direct spelling instruction but also reading experiencescontribute to the development of word spelling skill (Ehri, in press). When thereading experience entails reading words in isolation on flash cards, beginnersremember the spellings of words better than when the experience involves readingwords in sentences (Ehri & Roberts, 1979; Ehri & Wilce, 1980a). The correlationsobserved between reading and spelling are high, ranging from .66 to .90 (Ehri, inpress; Shanahan, 1984). Poor readers (at least in English) are almost always poorspellers. However, poor spellers are not always poor readers. A small percentage ofolder poorer spellers (age 12 and beyond) display adequate reading skill (Frith,1980). They are thought to operate with partial memory for spellings. They knowenough letters in words to read them accurately. They can spell phonetically andcan recall some letters in words. However, because they lack memory for all theletters, they have difficulty producing perfect word spellings (Ehri, in press).

Movement into Reading

This general picture of the course of reading and spelling acquisition providesthe background for specific questions that we have raised about the learningprocess. One question is, how do prereaders move into Stage 1? What knowledgesources, skills, and experiences enable them to begin reading words by processinggraphic cues without any help from pictures or other context cues? One




Learning to Read and Spell 11

explanation proposed is that this ability evolves naturally and spontaneously out ofchildren's prereading experiences, the same way that their oral language develops.Harste, Burke, and Woodward (1982) and the Goodmans (Goodman & Goodman,1979; Goodman & Altwerger, 1981) have developed this explanation. During thepreschool years, children become able to identify print that they frequently see intheir environment, for example, stop and exit signs, labels on milk and cerealcartons, and names of fast-food restaurants. Acquiring these print-meaningassociations provides the foundation for learning about the graphic system. First,children become aware that print is distinctively different from nonprint cues.Although they are not able to read signs, they can point to the place where it says"McDonald's." Gradually as a result of repeated exposure to these labels and signs,the print itself becomes decontextualized and can be recognized from graphic cuesalone without its characteristic environment.

An alternative explanation is that in order to begin using graphic cues,children must acquire certain prerequisite skills such as alphabet letter knowledge(Ehri, 1983). When prereaders who lack these skills exhibit an ability to read signsin their environment, they are "reading" the environment, not the print (Mason,1980). In order for them to shift their attention from environmental cues to theprint itself, they need to know the alphabet. Because there are so many letter shapesto be learned and to be associated with meaningless, arbitrary sounds, it is likelythat acquisition requires explicit instruction and practice. It is not picked up simplythrough exposure to the letters.

We performed a study to see whether environmental print experiences enableyoung children to process graphic cues (Masonheimer, Drum, & Ehri, 1984). Ourapproach was to select preschoolers who were experts at reading signs and labels intheir environment. From 228 children, we selected 102 three to five year olds whocould identify correctly at least eight out of 10 signs and labels in photographs, forexample, a Pepsi label on a bottle, a McDonald's sign on a restaurant. Weexamined these experts' ability to read various kinds of print with and withoutcontext cues. They were given a word reading task in which the 10 familiar signsand labels were printed in manuscript type on cards along with some preprimerwords. They were shown upper- and lowercase letters to name. They were given asign/label reading task in which the stylized print appeared either with its contextor logo or in isolation.

We reasoned that if environmental print experiences lead to gradualdecontextualization of graphic cues, we ought to observe a normal distribution ofscores in our word reading task. Results showed just the opposite, a bimodaldistribution. Most of the children, 94%, read few if any words. The remaining 6%read most of the words. There were no children in the middle of the distribution. Innaming letters, the nonreaders and readers differed greatly. Whereas all of thereaders could name at least 98% of the letters, very few nonreaders could namethem this well. Nonreaders' mean was 62% correct. Results in the stylized print





reading task revealed that the children who could read had no trouble reading thelabels either with or without their logos and contexts. Nonreaders had no troublewhen the labels were accompanied by logos or context cues. However, when thesecues were removed and only the stylized print remained, identification dropped to23%. These results show that most of our experts were not processing graphic cuesbut rather were reading environmental cues.

The same children were called back for a second experiment. We wanted totake another look at graphic cue processing with a more sensitive measure. Thistime children were shown labels accompanied only by their logos, for example, aMcDonald's sign backed by golden arches. In each label one of the letters wasaltered by replacing it with another letter having a very different shape. An exampleof this is presented in Figure 2. The new letters were printed in the same letter style,

XEPSIFigure 2. Example of a label and its logo with altered print.

they appeared in first, middle, or final positions in the words, and none altered theoverall shape of the label. First, subjects were asked to tell what the labels said. If noletter changes were noticed, they were asked, "Is there anything strange or wrongabout this picture? Is there a mistake?" After all the altered labels had beenpresented, each was shown again, this time alongside the original unaltered label,and subjects were asked whether both were the same or whether there were anymistakes.

Results revealed that the nonreaders were oblivious to graphic cues. Whenasked what the print said, they responded with the contextual label. They noticedfew if any letter changes even when asked whether there was a mistake. When theycompared altered labels to original labels, only about one third of the changes were





detected. These findings reveal very clearly that most of our environmental printexperts were not processing graphic cues.

Results of these studies provide little support for the view that children movecloser to acquiring reading skill after they have accumulated substantial experiencewith environmental print. Why is this so? Why isn't reading acquisition like orallanguage acquisition? Our explanation is that during the prereading stage, there isno "press" on children to look beyond the cues that are easiest to discern and mostobvious. From a functional point of view, little purpose is served by attending toletters in the environment. All the information prereaders need can be obtainedfrom non-alphabetic sources. Moreover, if letters are not familiar forms, there iseven less reason to notice them.

Acquisition of written language may contrast with oral language developmentin this respect. The process of learning to speak is driven by semantic andpragmatic forces plus the necessity of processing linguistic information. Childrenhave a strong need to make sense of what other people say to them, and non-vocalsources of information do not do the job. As a result, children are pushed todiscover, tacitly of course, the grammatical structures underlying these utterances(Slobin, 1971). Also, their search may be assisted by a brain uniquely suited for thispurpose (N. Chomsky, 1968). In contrast, the task of extracting meaning fromenvironmental print is not nearly so complicated and does not depend solely upondiscriminating and interpreting graphic units correctly. As a result, preschoolersnever analyze written language sufficiently to discover how the alphabetic systemworks.

If environmental print experiences are not sufficient to move prereaders intoStage 1 reading, then how do they become able to process graphic cues? We havesuggested that one thing they need to have mastered is the alphabet (Ehri, 1983).Some evidence for this was uncovered in our print expert study. Whereas all of thereaders could name practically every upper- and lowercase letter, only 6% of thenonreaders performed this well. We have observed this relationship in otherstudies, too (Ehri & Wilce, 1985). In fact, it is common knowledge that one of thebest predictors of reading achievement at the end of first grade is the ability to nameletters upon entering first grade (Chall, 1967). It is a better predictor even than IQ(Share, Jorm, Maclean, & Matthews, 1984). Our explanation for the relationship(Ehri, 1983) is that knowledge of letters provides children with the foundation forbeginning to process graphic cues in printed words. Knowing shapes helps themdistinguish and remember the visual constituents of words. Knowing letter nameshelps them associate relevant sounds with letters because most of the names containthese sounds. Research on the spellings invented by young children reveals thatthey are very good at analyzing sounds in letter names (Read, 1971). Of course,knowing letter names is just a starting point since some of the names are misleading,such as H, Q, Y, and W, and since there are many more sounds to be learned thatare not found in the names.





How do beginning readers get started using graphic cues to read words? Wehave been involved in a dispute with Phil Gough over this question, and we haveconducted two studies to attempt to settle the issue. Gough and Hillinger (1980)propose that beginning readers pass through two stages in their use of graphic cuesto read words. We will call children in the first stage cue readers and children in thesecond stage cipher readers.2 During the first cue reading stage, the mechanism ofselective paired associate learning is used to read words. Readers select a visuallydistinctive cue located in or around the printed word and associate this with theword in memory, for example, the tail on the final g in the spelling of "dog," or athumbprint appearing next to the word. In most cases, the visual cues selected areunrelated to the meanings of the words, and they are never related to sounds.Gough and Hillinger (1980) speculate that children use visual cues in learning toread their first 40 or so words. In attempting to employ this technique with moreand more words, cue readers run into trouble. It becomes harder to find a uniquevisual cue in each printed word to form an association. Different words sharing thesame features are mistaken for each other (Otto & Pizillo, 1970). Because they arearbitrary, associations are often forgotten. As a result, learners are unable to readwords reliably over time (Mason, 1980). The mounting confusion and frustrationeventually results in a shift to cipher reading, the second stage of development.

Cipher reading becomes possible when readers master letter-sound mappingrelations and phonemic segmentation. Upon seeing a word, cipher readers canapply their general knowledge of how orthography maps speech to generate apronunciation of the word that takes account of all of its letters. Cipher readingenables readers to decode spellings they have never seen before and to read wordsaccurately and consistently over time.

Our dispute with Gough is over his two-stage conception of the word readingprocess. His view is that children use visual cues to read words until they becomeable to decipher words. Our view is that there is an intermediate stage betweenvisual cue reading and deciphering. We call this stage phonetic cue reading. It is likevisual cue reading in that only one or a few specific cues, not the entire spelling, areselected and associated with the word in memory. However, it is different fromvisual cue reading in that the cues are letters that link the spelling to thepronunciation of the word. Once learners know letter names or sounds, they canbegin to form phonetic associations between printed words and their pronuncia-tions in memory to read words. To illustrate, beginners might learn to read theword "jail" by selecting the letters J and L and associating the names of these letters

2Gough and Hillinger (1980) refer to these subjects as code readers and cipher readers. We havechanged the name of the former group to "cue" readers because the term "code" has proved confusing. Itis similar to the term "decoding" yet it refers to the group of readers who cannot decode. Those who candecode are the cipher readers.





with sounds heard in the word's pronunciation. These associations are stored inmemory and retrieved the next time the word is seen. Phonetic cue reading shouldbe more effective than visual cue reading because the associations between spellingsand pronunciations are systematic rather than arbitrary and thus are easier toremember.

Phonetic cue reading might be regarded as a primitive form of deciphering.Although phonetic cue learners use letter-sound relations to read words as cipherreaders do, they process only some of the letters, not all of them. They read wordsby forming and retrieving phonetic associations in memory. They do not havesufficient knowledge of the orthographic system to read unfamiliar words bydecoding them from scratch.

We conducted a study to find out when during development beginning readersmight become able to use phonetic cues rather than visual cues in learning to readwords (Ehri & Wilce, 1985). In this study, we taught prereaders and beginningreaders to read two kinds of word spellings. These are listed in Table 1. One set

Table 1

Visual and Phonetic Spellings Taught to Children in the Word Learning Task

List 1

List 2

TargetWords

elephantdiapercombarmpencilchicken

giraffeballoonmaskkneescissorsturtle

PhoneticSpellings

LFTDIPRKOMRMPNSLHKN

JRFBLUNMSKNESZRSTRDL

VisualSpellings

WBCXGST

UHE

FoQDJK

yuLp

WBCXGST

UHE

FoQDJK

YMp

consisted of simplified phonetic spellings of words. All the letters corresponded tosounds found in the names of the letters. For example, J-R-F spelled "giraffe. "Theother set consisted of visually distinctive spellings. These spellings were created byvarying the height and ascending or descending positions of letters to give eachword a unique contour. Also, each word had unique letters not appearing in otherwords. However, none of the letters in visual spellings corresponded to sounds in





the words. Children were told what word each spelling symbolized and were givenseveral trials to learn to read the words. Subjects learned two sets of spellings, eitherthe phonetic spellings from List 1 and the visual spellings from List 2, or vice versa.

In order to study word learning performance developmentally, we selectedkindergarteners and divided them into three groups according to their wordreading ability: prereaders who read 0-1 words from a preprimer/ primer word list,novices who read 1-11 words, and veterans who read 11-36 words. Of particularinterest were the novice readers who had just moved into word reading and couldread only a few words, certainly fewer than 40. We reasoned that if Gough andHillinger (1980) are right, these children should find it easier to learn to read wordshaving distinctive visual cues than words having phonetic cues, whereas if we areright, they should learn the phonetic spellings more easily than the visual spellings.

Results are displayed in Figure 3. Statistical tests revealed that the noviceslearned to read the phonetic spellings significantly faster than the visual spellings.

6 r

o 4LUo:o:o ,

• Phonetic

_ — - • Visual

Prereader Novice

GROUP

Veteran

Note. From "Movement into Reading: Is the First Stage of Printed Word LearningVisual or Phonetic?" by L. C. Ehri and L. S. Wilce, 1985, Reading ResearchQuarterly, 20. Reprinted by permission.

Figure 3. Mean number of phonetic and visual spellings identified correctly in theword learning task by prereaders, novices, and veteran beginningreaders.





Their performance learning visual and phonetic spellings was in fact similar to thatof the more advanced beginning readers, the veterans who also learned phoneticspellings significantly better than visual spellings. These findings support our claimthat children who have just moved into word reading can use phonetic cueseffectively to read words and better than they can use visual cues. Interestingly, thesubjects who found visual cues significantly easier to use than phonetic cues werethe prereaders. This suggests that Gough's portrayal of visual cue reading mayapply more to nonreaders than to children who have moved into initial wordreading.

In this same study, we examined children's knowledge of alphabet letters andfound that the novices had mastered letter names but the prereaders had not. Thesefindings echo those in our print expert study where we also found that mastery ofletters distinguished prereaders from readers. This is further support for our claimthat one prerequisite for being able to process graphic cues effectively is knowingshapes and names or sounds of alphabet letters.

From Cue Reading to Cipher Reading

From results of our study comparing the use of visual and phonetic cues, weconcluded that when children first become able to read words out of context, thekind of graphic cues they use to form associations and remember words are mainlythe more salient letter-sound cues, perhaps only one or two of these. The purpose ofour next study was to manipulate the difference between phonetic cue readers andcipher readers experimentally in order to obtain more information about how thetwo kinds of readers differ in learning to read and spell words (Ehri & Wilce, inpress, a). We expected that phonetic cue and cipher readers would both useletter-sound relations rather than visual cues in their reading and spelling but thatthey would differ in the extent that they were able to use these cues.

Pretests were given to a number of kindergarteners to select children whocould not decipher nonsense words but who had moved into word reading andknew letter-sound relations. Half of these subjects were assigned to become cipherreaders, half to become phonetic cue readers. Cipher readers were taught to decode12 sets of similarly spelled words, mostly nonsense (total of 99 words). The wordswere formed out of nine consonants and four short vowels. Subjects first learned toread sets of consonant-vowel-consonants (CVC) to criterion (e.g., LAS, RAS,S AS), then sets with consonant clusters to criterion (e.g., DRUB, SPUM, LUND).The cue readers rehearsed isolated letter-sound relations (e.g., L—/I/, S—/s/,A—/ a/). There were nine cipher and nine cue readers. After training, we comparedcipher and cue readers' skill in learning to read a set of 15 similarly spelled realwords and also their memory for the words' spellings. Some words were CVCs,some had consonant blends. The same initial and final letters recurred in severalwords (e.g., spin, stab, stamp, stand).





Our aim was to find out how much easier it would be for cipher readers tolearn to read and spell these words than for cue readers. We expected that cipherreaders would learn to read words faster than cue readers, they would read wordsmore consistently, and they would produce a greater proportion of graphicallysimilar intrusion errors when they mistook one word for another. We were alsointerested in whether cue readers would show any signs of using letter-soundrelations in their reading and spelling. Based on our earlier study, we expected thatthey would process phonetic cues but that their attempts would not be completelysuccessful since they would not be able to process enough letter-sound relations todistinguish among all the similarly spelled words.

Figure 4 depicts scores of the two groups on the various posttest measures. An

100

80

60

40

20

0

a 100

2 80

60

40

20

0

16%

C D# *

72%

35*

Cu Ci

Decoding

E

37%

Cu Ci

Words

5 2 %

Trials

Word Learning Task

ReadingConsistency

Cu CiLetters Pronouncedin Misreadings

Word Learning Task

In.s.

7 0 %

Jn.s.

90%79%

18% 20%

Cu Ci

AllLetters

Cu Ci

Vowels ConsonantClusters

Spelling Task

Cu Ci Cu Ci

Pre- Initiol sconsonontol FinalNasals Consonants

Note. From "Do Cipher Readers Read and Spell Better Than Cue Readers?" byL. C. Ehri and L. S. Wilce, in press, Journal of Educational Psychology. Reprintedby permission.

Figure 4. Mean percent correct in the word reading and spelling posttests forcipher readers (N = 9) and cue readers (N = 9). Letters identify differentmeasures. Asterisks beneath letters indicate differences that werestatistically significant (p < .05). n.s. = not significant.





asterisk appears atop those pairs of means that statistically differed from eachother. The letter's n.s. indicate no statistical difference. The upper left cornercontains nonsense word decoding scores, labeled Measure A. It is evident thatcipher and cue readers differed greatly as a result of the training they received.Cipher readers could decode most of the nonsense spellings, whereas cue readerscould decode very few, indicating that the groups differed qualitatively in theirdeciphering skill. , •

The next panel contains mean scores in the word learning task where subjectspracticed reading 15 similarly spelled words. Clearly, cipher readers outperformedcue readers. In fact, all but one cipher reader attained a perfect score on at least onetrial. No cue reader did this well. Moreover, scores of cipher readers' showedconsistent gains on successive trials, whereas scores of cue readers were somewhaterratic. Often, cue readers would read a word correctly on one trial and then fail toread it correctly on a later trial. To show this, the number of times that a subjectread a word and continued to read it correctly on later trials was counted anddivided by the number of times words were read correctly at least once to yield ameasure of subjects' word reading consistency. From Figure 4, Measure C, it isapparent that cipher readers read over twice as many words consistently as cuereaders, who often forgot words or mixed them up with other words.

It was not the case that cue readers were attempting to process and remembernon-alphabetic visual cues. As evident in Figure 4, Measure D, cue readers'misreadings of words included some reasonable pronunciation for 52% of theletters they saw in spellings. This indicates that cue readers were using letter-soundcues. However, they were not quite as skilled as cipher readers who pronounced66% of the letters in their misreadings. This indicates that cipher readers wereattending to more graphic cues than cue readers.

Most of the cue readers' misreadings (81%) were other words on the list thatshared letters with the printed words, particularly initial letters. The most commonerrors were misreading LAP as "lamp," and STAB as "stamp." This is evidence thatcue readers were mixing up words on the basis of partial letter cues.

Interestingly, cue readers made some semantic errors in contrast to cipherreaders. Six cue readers produced a total of 10 substitutions such as the following:"light" or "lantern" for LAMP, "bang" or "shot" for BLAST, "baby" for BIB.Some of these words were suggested by sentence contexts that had accompaniedthe words during the learning trials. No cipher reader produced any semanticmisreadings of this sort. One explanation is that cipher readers' misreadings werecontrolled more by graphic cues than cue readers' misreadings. It may be that whenreaders begin attending primarily to letter-sound cues in reading words,semantically related words are suppressed because readers realize that sounds in thewords do not match up well enough with spellings.

In recalling the spellings of words, cipher readers generally outperformed cuereaders, not surprisingly since cipher readers were superior decoders and since they





had learned to read more of the words. Results in the bottom half of Figure 4 revealthat cipher readers produced significantly more correct word and letter spellingsincluding vowels and consonant clusters in particular. Cue readers spelled very fewconsonant clusters correctly, indicating that decoding training makes a bigcontribution here. Apparently, when readers learn to decode consonant clusters,they also become able to spell these sounds, even without direct spelling instruction,perhaps because decoding training teaches how to analyze consonant blends intotwo sounds. Without training, only one sound is detected (Treiman, 1985).

There was one type of consonant cluster, however, that cipher readers learnedto decode but did not carry into their spellings. This was the preconsonantal nasalwhich occurs between a vowel and a final consonant, for example, the M in theword "lamp," and the N in "blond." Six of the words contained these pre-consonantal nasals. From Measure I in Figure 4, it is evident that very fewpreconsonantal nasals were spelled by either group, and the groups did not differsignificantly. This is despite the fact that cipher readers completed training on thesesounds and despite the fact that they learned to read the words they were spelling.Research on children's invented spellings has shown that beginners usually omit thepreconsonantal nasal, apparently because it is not articulated as a separate segment(Read, 1971). Rather, when you say "lamp," the nasal flap opens when the vowelsound begins and the vowel and nasal are articulated simultaneously. It may be thatdecoding training did not make learners sufficiently aware of this segment torepresent it in their spellings.

How did cue readers perform on the spelling task? Although performance wasnot strong on most measures, there was one exception. Cue readers were able tospell the initial and final consonants of most words, and their performance did notdiffer significantly from that of cipher readers. (See Figure 4, Measure J.) Onereason may be that initial and final letters served as the phonetic cues that cuereaders tried to use during the word reading task. The spellings of one cue readerprovided extra support for this. She spelled 97% of the first and final letterscorrectly but few middle letters correctly. In addition, when she wrote these words,she capitalized the initial and final letters as they had appeared in the words she hadseen, but she wrote most of the medial letters in lowercase, even though these toohad been seen in capital letters. These behaviors suggest that this cue readerattended to and remembered boundary letters during the word reading task butignored medial letters.

Evidence from our earlier study (Ehri & Wilce, 1985) suggests that cue readers'skill in remembering first and final letters was not a result of visual cue memory. Inthis earlier study where cue readers (the novices) learned to read visually distinctivespellings, a test of their memory for first and final letters revealed very poorperformance. Only 26% of the letters in visual spellings were recalled. This value ismuch lower than the 79% observed in this study, indicating that phonetic cueprocessing is a more likely explanation.





In sum, findings of this study show that acquiring the ability to decipher printoffers clear advantages. It enables readers to learn to read similarly spelled wordsaccurately and consistently. It also contributes to their ability to spell the wordsalthough spelling skill is not as strongly supported as reading skill.

Findings provide evidence for our claim that there is another way to processand remember graphic cues as symbols for words besides memorizing visualfeatures and decoding letters into sounds. This way was exhibited by our cuereaders who attempted to process a few salient letter-sound relations and storethese associations in memory. The cue readers we studied were novice beginnersand not very skilled at this process. It may be that with practice cue readers becomebetter at forming associations out of the more salient phonetic cues. Thismechanism of word reading may have much potential for explaining how poorreaders are able to read words. We know from other studies that poor readers arecue readers. They have weak decoding skill and spelling skill yet often they are ableto read many real words (Ehri & Wilce, 1983; Perfetti & Hogaboam, 1975). Theexplanation usually given for this is that poor readers notice and remember visualcues in the words (Gough & Hillinger, 1980). However, as we found in our earlierstudy (Ehri & Wilce, 1985), visual associations are harder to remember thanphonetic associations. Since poor readers usually know the alphabet, they shouldbe able to process phonetic cues. Because they have had more experience with print,poor readers ought to be able to do phonetic cue processing much better than ournovice readers. This possibility needs investigation.

Our findings carry implications for instruction. They reveal the importance ofmoving students beyond a phonetic cue reading strategy and teaching them cipherreading. Cipher reading entails learning not only letter-sound relations but alsohow the sequence of letters in spellings symbolizes phonemes in pronunciations andhow letter-sounds are blended to form pronunciations. However, our findings donot clarify how to teach deciphering skill. Although our training method workedwith some children, rehearsing nonsense syllables over and over is boring andtedious. We chose this method for experimental purposes. More interestingmethods need to be identified for classroom use.

Acquisition of a Print Lexicon

The question that initiated another series of studies on printed word learningwas how do beginning readers build up a lexicon of printed words (Ehri, 1978,1980a, 1980b). The focus here was on cipher readers, not on cue readers. The theoryproposed was that readers store visual spellings in memory by analyzing how theletters symbolize phonemic constituents in pronunciations. To do this requiresdeciphering skill. Readers must know the particular sounds that letters symbolize,and they must know how to segment pronunciations into phonemes. For example,in reading and remembering the word "rich," they must recognize how the





graphemic units, R, I, and the digraph CH each symbolizes a phoneme in thepronunciation. To the extent that letters are processed in this way, spellings ofindividual words are retained in memory as the word's visual representation.

To test this theory, we conducted several studies. In one set of studies, weshowed that spellings can perform a mnemonic function, that letters can preservenonsense sounds in memory when they provide an adequate printed symbol for thesounds (Ehri & Wilce, 1979a). A common example of this mnemonic functionoccurs when you hear an unfamiliar name and attempt to enhance your memoryfor it by inquiring how the name is spelled. In this study, we used a paired associatelearning task. Young readers were given several trials to learn four oral CVCnonsense syllables such as "mav," "rel," "kip" and "guz." Recall of these responseswas prompted by numbers. During study periods, children saw either correctspellings of the sounds, or misspellings, or they rehearsed the sounds. On test trials,only the numbers were presented and children recalled the sounds. We found thatchildren learned the sounds fastest when they viewed correct spellings during studyperiods. Looking at misspellings made it especially hard to learn the sounds. Ourinterpretation is that spellings improved memory for sounds because they wereretained as visual symbols preserving the sounds in memory. In this study, weobserved a very high correlation between young readers' ability to use lettersmnemonically and the size of their print lexicons. This encouraged us in our beliefthat letter symbolization underlies readers' ability to store printed words inmemory.

In another series of studies (Ehri & Wilce, 1982), we examined children'smemory for words containing silent letters, for example, the T in "listen." Wereasoned that if readers store spellings by analyzing how letters symbolize sounds,they should find silent letters harder to remember than pronounced letters.However, because silent letters are singled out as exceptions by the letter-soundsymbolization process, silent letters may become more salient in the memoryrepresentation. In this study, we gave children two tasks to compare their memoryfor silent and pronounced letters in words whose spellings they knew. First, theyimagined the spellings of the words and decided whether each contained a specifiedtarget letter. For example, they imagined the word "listen" and then decidedwhether it had a T. We measured reaction times to make these decisions. Then thetarget letters were presented again and subjects were asked to recall the word theyhad judged for that letter. In the imagining task, we found that subjects correctlyidentified pronounced letters more accurately than silent letters although scoreswere high in both cases. This indicates that silent letters are harder to remember.However, when we looked at subjects' reaction times to identify letters correctly, wefound that silent letters were identified faster than pronounced letters. Moreover, inthe recall task, silent letters prompted the retrieval of more words than pronouncedletters. These findings indicate that silent letters are salient in memory.





One way that readers might deal with silent letters in words is to mark them asexceptions in the memory representation. Another way is to phonemically encodespellings so that silent letters are assigned sounds, for example, pronouncing"lissen" as "lis-ten." Some spellers report using this strategy to enhance theirmemory for spellings. According to our theory, creating and storing a specialspelling pronunciation for words should boost memory for the spellings since this ishow letters are stored in memory, as sound symbols. We performed a study to testthis hypothesis (Drake & Ehri, 1984).

Fourth graders were shown spellings of words divided into syllables, and theyphonemically encoded each syllable as it was spelled. The words they studiedcontained some schwa vowels and silent letters. For example, the word "chocolate"was pronounced "choc-o-late." The medial silent O was pronounced as a longvowel, and the schwa vowel "uh"in"lut"was pronounced as it is spelled, with a longA sound, "late." The word "excellent" was pronounced "ex-cel-lent" with the twoschwa vowels transformed into short E sounds as symbolized in the spelling.Results of this study confirmed predictions. Subjects who produced phonemicencodings of the spellings remembered letters better than subjects in a controlgroup who pronounced spellings conventionally in the manner prescribed by thedictionary. Memory for silent letters was superior. Also, memory for schwaspellings was superior. However, memory for double letters which are not pickedup in a phonemic encoding was not superior. Findings are interpreted as furtherevidence that spellings are retained in memory when spellers construe letters assound symbols.

Results of this study carry implications for spelling instruction, particularlyfor teaching spellers about schwa vowels. To review, schwa is what happens tovowel sounds when they occur in unstressed syllables in multisyllabic words. Thedistinctive vowel sound is reduced to a nondistinctive schwa or "uh" sound. Forexample, take the words "horizon" and "horizontal." The stressed long I in"horizon" becomes reduced to schwa when the stress is shifted to another syllable,as in "horizontal." Not surprisingly, schwa spellings are among the hardest lettersto remember in words. The sound may be spelled with any of the five vowel lettersor combinations. What our findings suggest is that if you want to improve students'memory for the spellings of schwa, you should teach them that schwa is not the"true" vowel sound in the word. Schwa is simply the result of stress reduction.Rather, the "true" sound is that revealed in the spelling. To discover and rememberthe "true" sound, the spelling must be phonemically encoded. Our results suggestthat teaching students this bit of fiction should help them remember the spellings ofschwa vowels.

Of course, some spelling teachers have been teaching this strategy for years.The problem with the tactic is that although it may aid spelling memory, it treatsspellings ideosyncratically and ignores any historical or phonological justification





for letter patterns. Also, the idea that there are "true" sounds borders on beingprescriptive and on disregarding the complexities of sound variation in speech.

Spellings Influence Speech

One other line of research also provided support for our view of printed wordmemory. In addition, this research led us into controversy. We reasoned that ifspellings are retained in memory by being analyzed as sound symbols, then theyought to influence speech when they are learned. They may influence people'sbeliefs about the sounds that comprise words, and they may even alter the way thatpeople pronounce words. The chances of influence should be greatest for wordscontaining ambiguous sounds and for words whose spellings are discrepant withtheir pronunciations. Also, influence should be greatest in younger children who donot possess much awareness of phonemes when they begin learning to read.Learning to interpret spellings as symbols for sounds should clarify for them howthe words are structured phonemically and perhaps even how the words aresupposed to be pronounced. For example, spellings may teach readers to say"February" rather than "February."

In our first experiment, we examined the influence of spellings on children'sconception of phonemes in words (Ehri & Wilce, 1980b). We selected seven wordswhose spellings contained an extra, potentially pronounceable letter. For example,"pitch" has a T in its spelling. A sound for this letter can be found in the sequence ofarticulatory movements produced in saying the word. Without the letter, however,you probably would not distinguish a separate /t / segment. For each of theseextra-letter words, we selected a control word matched phonemically but lackingan extra letter in the spelling. For example, the control word for "pitch" was "rich."Other examples are (extra letter italicized) "cafch" versus "much," "bac/ge" versus"page," "can you" versus "menu," "new" versus "do." We gave fourth graders aphonemic segmentation task to find out what sounds they thought were in thesewords. Children spoke each word, then divided it into segments by pronouncingeach sound and marking it with a token. We counted the number of times extraphonemes were detected in extra-letter words and in control words. Resultsrevealed a huge difference. Extra phonemes were frequently detected in wordswhose spellings symbolized these phonemes, but they were rarely detected incontrol words. We also checked whether subjects knew the spellings of the words.Most did. However, subjects who could not spell the words also did not find extrasegments in pronunciations. It was not the case that subjects were merelyrepresenting letters in the segmentation task. Most children ignored truly silentletters and they did not put down two tokens for digraphs such as CH. Also, byrequiring subjects to say the sounds, we discouraged simple letter counting.

The "pitch-rich" study was followed up with a training study (Ehri & Wilce,1980b). Children were taught to read spellings of nonsense words structured like





the real words used in the earlier study. Half of the subjects practiced reading wordspellings that included extra letters. The other half read control spellings. Thewords were pronounced identically in both conditions. Then subjects were given aphonemic segmentation task. As expected, subjects who saw extra letters inspellings often included the extra phonemes in their segmentations whereascontrols did not.

We conducted another training study to see if learning spellings would alterchildren's ideas about speech. In this study, we examined the influence of spellingson alveolar flaps (Ehri & Wilce, in press, b). Alveolar flaps are consonants that arespelled with D or T. They occur in the middle of words between two vowels, such asin "ladder," "meral," "amc," and "hottest." Although this sound is symbolizedeither with D or T in spellings, most Americans do not preserve the distinction inspeech. Rather, they pronounced it as a flap, something close to /d / . Ourhypothesis was that readers who know spellings will regard the sounds as either /1 /or / d/ depending upon the spelling, and they will preserve these spellings in carefulspeech. For example, if you ask them to pronounce the first syllable in the word"glitter," they will say "glit," converting the flap to a / t / .

To test this hypothesis, we selected 16 words containing flaps, half spelled withT, half spelled with D. Second graders were divided randomly into two groups. Onegroup practiced reading the words. The other group practiced saying the words butnever saw spellings. Sentences were provided to clarify word meanings. On the nextday, both groups were taught to perform a rhyme judgment task. This was to findout whether they thought the medial sounds were / d/ or / 1 / . Subjects pronouncedthe first syllable of each word and then, from pictures provided, chose the picturename with which it rhymed. To illustrate, for the word "meteor," they might say thefirst syllable as "meet" or "meed." Then they had to decide whether it rhymed with"feet" or "seed" which were the names of two pictures they were shown.Incidentally, the experimenter never pronounced the syllables. They were alwayselicited from the subject. Also, spellings were never shown or mentioned during therhyme judgment task. Any influence came from subjects' memories for thespellings.

Results confirmed our hypothesis. Subjects who had studied the printedwords judged significantly more of the flap sounds according to their spellings thancontrol subjects who did not see printed words. Differences between the two groupsfavoring experimentals were greater on words spelled with T than with D becausecontrol subjects judged more of the sounds as /d/ than as / t / . Of course, controlsubjects' perception was actually more accurate since the sound is closer to /d/ inthe speech of most Americans. These findings provide evidence that spellings mayhave a distorting influence on readers' perception of sound in words. According toour view, this is because the spellings are interpreted as symbols for pronunciationsand are stored in memory this way.





In one other study, we observed spellings to influence fourth graders'pronunciations of syllables in multisyllabic words (Ehri, 1984). The words we chosewere ones whose spellings suggested that there might be an extra syllable in theword. For example, does the word "interesting" have three or four syllables? Is itpronounced "intresting" or "interesting"? We examined how many syllables oursubjects thought were in the spellings of 11 words like this one. We also examinedwhether they knew the spellings of the words. We found that subjects who knewspellings were more likely to segment words into the syllables represented in thespellings than subjects who did not know spellings. This is one more piece ofevidence for the formative influence of print on speech.

This research'on print-speech relations has led us into some controversies. Thefirst controversy is over the relationship between phonemic segmentation andlearning to read (Ehri, 1979). Phonemic segmentation refers to the ability to dividewords into their constituent sounds or phonemes. Several studies have shown thatphonemic segmentation is quite difficult for young children below the age of five orsix. The task is difficult because there are no boundaries marking phonemes inspeech. Rather, sounds overlap, they may be co-articulated, and neighboringsounds may distort or alter their phonetic features. Furthermore, sounds areephemeral. They last for a very brief time and so are hard to catch hold of andinspect. Liberman and Shankweiler (1979), Fox and Routh (1975, 1976, 1984),Gough and Hillinger (1980), and others have claimed that awareness of phonemesin speech is a prerequisite for learning to read and that children should be taught tosegment phonemes before they begin reading instruction.

Our research along with that of others (Morais, Cary, Alegria, & Bertelson,1979) has suggested another possibility—that phonemic awareness may be part oflearning to read rather than something that is developed beforehand. One functionof letters in spellings is to clarify what sounds are being heard in words (Ehri, 1984,1985). Because phoneme segments are difficult to detect in speech, having a visible

picture of the sounds should help tremendously. Many spellings in English can beregarded as phonemic transcriptions of speech. At least this is how beginningreaders are taught to interpret them and this is how they store the spellings of wordsin memory, as we have shown in our studies. What we suggest is that childrenshould be taught phonemic awareness as they learn to read, not beforehand.Learning segmentation with letters ought to be more effective since the ultimategoal is to prepare beginners to process spellings as phonemic symbols.

We performed a training study (Hohn & Ehri, 1983) and found that childrenwho were taught to segment words into phonemes using letters learned the skillbetter than children who were taught to segment using blank markers. Bradley andBryant (1983) performed a training study in which subjects were taught phonemicsegmentation over a two-year period while they were learning to read. They foundthat experimental subjects were ahead of controls both in reading and in spelling atthe end of two years. Training that included alphabet letters was the most effective.





Others have performed studies on this question as well. In fact, the December,1984, issue of the Journal of Educational Psychology contains three such studies,all providing evidence for the prerequisite claim. Of course, it may be thatphonemic segmentation is both a prerequisite and a consequence of learning toread and spell. That is, knowing how to segment may enable children to learn toread and spell more effectively. Likewise, as children learn about the alphabeticspelling system from their reading and spelling instruction, their phonemicsegmentation skill may improve.

The second controversy is over the issue of whether speech is primary andprint merely a passive, reflected form of speech, or whether print exerts an active,formative influence oh speech (Ehri, 1984,1985). Many linguists tend to believe thatspeech is primary and print is a reflection. For example, Bloomfield (1933)described print as "merely a way of recording language by means of visible marks."Further on he says, "In order to study writing, we must know something aboutlanguage, but the reverse is not true." (p. 21). More recently, Chomsky hasconsidered writing to reflect deep structure phonology (C. Chomsky, 1970). Incontrast, our research has led us to believe, not that print is passive and reflective,but rather that print shapes some aspects of speech, particularly during childhoodwhen reading and spelling skills are acquired. We looked into the literature foradditional support and found a variety of intriguing studies and arguments some ofwhich are mentioned below (Ehri, 1984, 1985).

There are psychological studies indicating that spellings sit in memory andinfluence performance in oral language tasks. For example, Seidenberg andTanenhaus (1979) had subjects listen to a succession of nouns such as "glue" and"shoe" and decide whether each rhymed with the target word "clue." They foundthat subjects responded faster to words having spellings that were similar to thetarget word, in this case, "glue," than to words with a different spelling, such as"shoe." Also, it was found that when words had similar spellings but did not rhyme,for example, "bomb" and "tomb," subjects took longer to make a negativedecision with these words than with words having different spellings andpronunciations, for example, "bomb" and "room." Spellings of words were neithershown nor mentioned so their effects had to come from subjects' memories.

The maverick linguist Householder (1971) has argued for the primacy ofwriting over speech. He points out that historically there are many more instanceswhere spellings have changed the pronunciations of words than where pronuncia-tions have altered spellings. Also, the law is on the side of spellings. It cares little ifpeople decide to change the way they pronounce their names, but court action isrequired to make an altered spelling legitimate.

There is evidence that printed language acts to constrain phonological drift, aterm referring to changes in the pronunciations of words that occur in communitiesof speakers over time. According to Gelb (1952), English pronunciations havechanged relatively little over the last four or five hundred years as a result of its





writing system. This contrasts with dramatic shifts that occurred prior to that time.Also, this contrasts with rapid linguistic changes that are evident in modernprimitive societies which lack a phonemic writing system. Gelb points out thatsome American Indian languages are changing so fast that people of the presentgeneration have difficulty conversing with people three or four generations older.Further support for Gelb's claim is presented by Bright (1960) and Bright andRamanujan (1962) who compared phonological drift in several South Asian Indiancommunities that spoke various dialects. They found less phonological drift incommunities that had a written form of language than in communities that hadonly a spoken form. According to our theory, the reason why written languagefreezes pronunciations and inhibits phonological change is that spellings castpronunciations into fixed phonemic forms. The letters specify which phonemes arethere, and, because speakers hold these in memory, they are reminded of the formsand adhere to them in their speech.

In conclusion, this and other evidence indicates that print exerts a formativeinfluence on speech. We regard these as important and exciting studies that elevatereading and spelling research to a central position in cognitive psychology andpsycholinguistics. They reveal that the study of how children learn to read and spellis important not only for applied reasons having to do with skill acquisition andschool achievement but also for clarifying our basic understanding of how humaninformation-processing equipment develops. During acquisition, print works itsway into the minds of learners and influences how they perceive and process speech.As such, learning to read and spell are major events during the course of languagedevelopment. If one is studying how children's language develops or how adultsprocess language, consideration must be given to the influence of knowing how toread and spell.

REFERENCES

Biemiller, A. (1970). The development of the use of graphic and contextual information as children learnto read. Reading Research Quarterly, 6, 75-96.

Bloomfield, L. (1933). Language. New York: Holt, Rinehart & Winston.Bradley, L., & Bryant, P. E. (1983). Categorizing sounds and learning to read—A causal connection.

Nature, 301, 419-421.Bright, W. (1960). Linguistic change in some Indian caste dialects. In C. A. Ferguson & J. J. Gumperz

(Eds.), Linguistic diversity in South Asia. International Journal of American Linguistics, 26,19-26.

Bright, W., & Ramanujan, A. K. (1962). Sociolinguistic variation and language change. In H. G. Lunt(Ed.), Proceedings of the Ninth International Congress of Linguists. Cambridge, MA: Mouton.

Carpenter, P. A., & Just, M. A. (1981). Cognitive processes in reading: Models based on readers' eyefixations. In A. M. Lesgold & C. A. Perfetti (Eds.), Interactive processes in reading. Hillsdale, NJ:Erlbaum.





Chall, J. (1967). Learning to read: The great debate. New York: McGraw-Hill.Chall, J. S. (1979). The great debate: Ten years later with a modest proposal for reading stages. In L. G.

Resnick & P. A. Weaver (Eds.), Theory and practice of early reading (Vol. 1, pp. 29-56). Hillsdale,NJ: Erlbaum.

Chomsky, C. (1970). Reading, writing and phonology. Harvard Educational Review, 40, 287-309.Chomsky, N. (1968). Language and mind. New York: Harcourt Brace Jovanovich.Drake, D. A., & Ehri, L. C. (1984). Spelling acquisition: Effects of pronouncing words on memory for

their spellings. Cognition and Instruction, 1, 297-320.Ehri, L. C. (1978). Beginning reading from a psycholinguistic perspective: Amalgamation of word

identities. In F. B. Murray (Ed.), The recognition of words (pp. 1-33), IRA Series on theDevelopment of the Reading Process. Newark, DE: International Reading Association.

Ehri, L. C. (1979). Linguistic insight: Threshold of reading acquisition. In T. G. Waller & G. E.Mackinnon (Eds.), Reading research: Advances in theory and practice (Vol. 1). New York:Academic Press.

Ehri, L. C. (1980a). The development of orthographic images. In U. Frith (Ed.), Cognitive processes inspelling. London: Academic Press.

Ehri, L. C. (1980b). The role of orthographic images in learning printed words. In J. Kavanagh & R.Venezky (Eds.), Orthography, reading and dyslexia. Baltimore, MD: University Park Press.

Ehri, L. C. (1983). Critique of five studies related to letter name knowledge and learning to read. In L.Gentile, M. Kamil, & J. Blanchard (Eds.), Reading research revisited (pp. 143-153). Columbus,OH: C. E. Merrill.

Ehri, L. C. (1984). How orthography alters spoken language competencies in children learning to readand spell. In J. Downing & R. Valtin (Eds.), Language awareness and learning to read (pp.119-147). New York: Springer Verlag.

Ehri, L. C. (1985). Effects of printed language acquisition on speech. In D. Olson, N. G. Torrance, & A.Hildyard (Eds.), Literacy, language and learning: The nature and consequence of reading andwriting. Cambridge, England: Cambridge University Press.

Ehri, L. C. (in press). Sources of difficulty in learning to spell and read. In M. L. Wolraich & D. Routh(Eds.), Advances in developmental and behavioral pediatrics. Greenwich, CT: Jai Press.

Ehri, L. C., & Roberts, K. (1979). Do beginners learn printed words better in context or isolation? ChildDevelopment, 50, 675-685.

Ehri, L. C., & Wilce, L. S. (1979a). The mnemonic value of orthography among beginning readers.Journal of Educational Psychology, 71, 26-40.

Ehri, L. C., & Wilce, L. S. (1979b). Does word training increase or decrease interference in a Strooptask? Journal of Experimental Child Psychology, 27, 352-364.

Ehri, L. C., & Wilce, L. S. (1980a). Do beginners learn to read function words better in sentences or inlists? Reading Research Quarterly, 15, 451-476.

Ehri, L. C., & Wilce, L. S. (1980b). The influence of orthography on readers'conceptualization of thephonemic structure of words. Applied Psycholinguistics, 1, 371-385.

Ehri, L. C., & Wilce, L. S. (1982). The salience of silent letters in children's memory for word spellings.Memory and Cognition, 104, 155-166.

Ehri, L. C., & Wilce, L. S. (1983). Development of word identification speed in skilled and less skilledbeginning readers. Journal of Educational Psychology, 75, 3-18.

Ehri, L. C., & Wilce, L. S. (1985). Movement into reading: Is the first stage of printed word learningvisual or phonetic? Reading Research Quarterly, 20, 163-179.

Ehri, L. C., & Wilce, L. S. (in press, a). Do cipher readers read and spell better than cue readers? Journalof Educational Psychology.

Ehri, L. C., & Wilce, L. S. (in press, b). The influence of spellings on speech: Are alveolar flaps /d/ or/t /? In D. Yaden & S. Templeton (Eds.), Metalinguistic awareness and beginning literacy. Exeter,NH: Heinemann Educational Books.





Fox, B., & Routh, D. K. (1975). Analyzing spoken language into words, syllables, and phonemes: Adevelopmental study. Journal of Psycholinguistic Research, 4, 332-342.

Fox, B., & Routh, D. K. (1976). Phonemic analysis and synthesis as word-attack skills. Journal ofEducational Psychology, 68, 70-74.

Fox, B., & Routh, D. K. (1984). Phonemic analysis and synthesis as word attack skills: Revisited.Journal of Educational Psychology, 76, 1059-1064.

Frith, U. (1980). Unexpected spelling problems. In U. Frith (Ed.), Cognitive processes in spelling (pp.495-515). London: Academic Press.

Gelb, I. J. (1952). A study of writing. Chicago, IL: University of Chicago Press.Gentry, J. R. (1982). An analysis of developmental spelling in GNYS AT WRK. The Reading Teacher,

36, 192-200.Glushko, R. J. (1979). The organization and activation of orthographic knowledge in reading aloud.

Journal of Experimental Psychology: Human Perception and Performance, 5, 674-691.Glushko, R. J. (1981). Principles for pronouncing print: The psychology of phonography. In A. M.

Lesgold & C. A. Perfetti (Eds.), Interactive processes in reading (pp. 61-84). Hillsdale, NJ:Erlbaum.

Goodman, K. S. (1970). Reading: A psycholinguistic guessing game. In H. Singer & R. B. Ruddell(Eds.), Theoretical models and processes of reading (pp. 497-508). Newark, DE: InternationalReading Association.

Goodman, K. S., & Goodman, Y. M. (1979). Learning to read is natural. In L. B. Resnick & P. A.Weaver (Eds.), Theory andpractice ofearly reading (Vol. l,pp. 137-154). Hillsdale, NJ: LawrenceErlbaum Associates.

Goodman, Y. M., & Altwerger, B. (1981). Print awareness in preschool children: A working paper. Astudy of the development of literacy in preschool children. Occasional Papers No. 4, Program inLanguage and Literacy, University of Arizona.

Gough, P. B., & Hillinger, M. (1980). Learning to read: An unnatural act. Bulletin of the Orton Society,30, 180-196.

Guttentag, R. E., & Haith, M. M. (1978). Automatic processing as a function of age and reading ability.Child Development, 49, 707-716.

Harste, J. C., Burke, C. L., & Woodward, V. A. (1982). Children's language and world: Initialencounters with print. In J. Langer & M. Smith-Burke (Eds.), Bridging the gap: Reader meetsauthor (pp. 105-131). Newark, DE: International Reading Association.

Henderson, E. H. (1981). Learning to read and spell. De Kalb, IL: Northern Illinois University Press.Hohn, W. E., & Ehri, L. C. (1983). Do alphabet letters help prereaders acquire phonemic segmentation

skill? Journal of Educational Psychology, 75, 752-762.Householder, F. (1971). Linguistic speculations. Cambridge, England: Cambridge University Press.Juel, C., & Roper/Schneider, D. (1985). The influence of basal readers on first grade reading. Reading

Research Quarterly, 20, 134-152.Kay, J., & Marcel, T. (1981). One process not two in reading aloud: Lexical analogies do the work of

nonlexical rules. Quarterly Journal of Experimental Psychology, 33, 397-413.LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading.

Cognitive Psychology, 6, 293-323.Liberman, I. Y., & Shankweiler, D. (1979). Speech, the alphabet and teaching to read. In L. B. Resnick

& P. A. Weaver (Eds.), Theory and practice of early reading (Vol. 2, pp. 109-132). Hillsdale, NJ:Erlbaum.

Marcel, T. (1980). Surface dyslexia and beginning reading: A revised hypothesis of the pronunciation ofprint and its impairments. In M. Coltheart, K. Patterson, & J. C. Marshall (Eds.), Deep dyslexia.London: Routledge.

Marsh, G., Friedman, M., Welch, V., & Desberg, P. (1981). A cognitive-developmental theory ofreading acquisition. In G. E. Mackinnon & T. G. Waller (Eds.), Reading research: Advances intheory and practice (Vol. 3, pp. 199-221). New York: Academic Press.





Mason, J. (1980). When do children begin to read: An exploration of four-year-old children's letter andword reading competencies. Reading Research Quarterly, 15, 203-227.

Mason, J. M., & Allen, J. (in press). A review of emergent literacy with implications for research andpractice in reading. In E. Z. Rothkopf (Ed.), Review of research in education (Vol. 13). WashingtonDC: American Educational Research Association.

Masonheimer, P. E., Drum, P. A., & Ehri, L. C. (1984). Does environmental print identification leadchildren into word reading? Journal of Reading Behavior, 16, 257-271.

McConkie, G. W., & Zola, D. (1981). Language constraints and the functional stimulus in reading. InA. M. Lesgold & C. A. Perfetti (Eds.), Interactive processes in reading (pp. 155-175). Hillsdale, NJ:Erlbaum.

Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of speech as a sequence ofphones arise spontaneously? Cognition, 7, 323-331.

Morris, D. (1981). Concept of word: A developmental phenomenon in the beginning reading andwriting processes. Language Arts, 58, 659-668.

Otto, H., & Pizillo, C. (1970). Effect of intralist similarity on kindergarten pupils' rate of wordacquisition and transfer. Journal of Reading Behavior, 3, 14-19.

Perfetti, C. A. (1984). Reading acquisition and beyond: Decoding includes cognition. American Journalof Education, 93, 40-60.

Perfetti, C. A., & Hogaboam, T. (1975). Relationship between single word decoding and readingcomprehension skill. Journal of Educational Psychology, 67, 461-469.

Read, C. (1971). Pre-school children's knowledge of English phonology. Harvard Educational Review,41, 1-34.

Rosinski, R. R., Golinkoff, R. M., & Kukish, K. S. (1975). Automatic semantic processing in apicture-word interference task. Child Development, 46, 247-253.

Seidenberg, M. S., & Tanenhaus, M. K. (1979). Orthographic effects on rhyme monitoring. Journal ofExperimental Psychology: Human Learning and Memory. 5, 546-554.

Shanahan, T. (1984). Nature of the reading-writing relation: An exploratory multivariate analysis.Journal of Educational Psychology, 76, 466-477.

Share, D. L., Jorm, A. F., Maclean, R., & Matthews, R. (1984). Sources of individual differences inreading acquisition. Journal of Educational Psychology, 76, 1309-1324.

Slobin, D. I. (1971). Psycholinguistics. Glenview, IL: Scott Foresman.Stanovich, K. E. (1980). Toward an interactive-compensatory model of individual differences in the

development of reading fluency. Reading Research Quarterly, 16, 32-71.Treiman, R. (1985). Phonemic analysis, spelling and reading. In T. Carr (Ed.), New directions in child

development: The development of reading skills (27, pp. 5-18). San Francisco: Jossey-Bass.Weber, R. (1970). A linguistic analysis of first-grade reading errors. Reading Research Quarterly, 5,

427-451.

AUTHOR NOTES

This was an invited address delivered at the annual meeting of the AmericanEducational Research Association in April, 1985. Our research has been supportedin part by grants from the National Institute of Education (Grant No. NIE-G-83-0039 and NIE-G-77-0009) and from the National Institute of Child Health andHuman Development (Grant No. HD-12903-01). Lee Wilce has assisted with muchof this research.




Date post:	23-Dec-2016
Category:	Documents
Upload:	linnea
View:	215 times
Download:	0 times

Learning to read and spell words

Documents