The Science of Learning to Read

Words: A Case for Systematic

Phonics Instruction

Linnea C. Ehri ABSTR ACT

Graduate Center of the City University The author reviews theory and research by Ehri and her colleagues to docu-
of New York, New York, USA ment how a scientific approach has been applied over the years to conduct
controlled studies whose findings reveal how beginners learn to read words in
and out of text. Words may be read by decoding letters into blended sounds
or by predicting words from context, but the way that contributes most to
reading and comprehending text is reading words automatically from mem-
ory by sight. The evidence shows that words are read from memory when
graphemes are connected to phonemes. This bonds spellings of individual
words to their pronunciations along with their meanings in memory. Readers
must know grapheme–phoneme relations and have decoding skill to form con-
nections, and must read words in text to associate spellings with meanings.
Readers move through four developmental phases as they acquire knowledge
about the alphabetic writing system and apply it to read and write words and
build their sight vocabularies. Grapheme–phoneme knowledge and phonemic
segmentation are key foundational skills that launch development followed
subsequently by knowledge of syllabic and morphemic spelling–sound units.
Findings show that when spellings attach to pronunciations and meanings
in memory, they enhance memory for vocabulary words. This research
underscores the importance of systematic phonics instruction that teach-
es students the knowledge and skills that are essential in acquiring word-
reading skill.

F
or many years, my collaborators and I have been applying scientific
procedures to carry out experiments in order to understand how
beginners learn to read (Ehri, 2017). Science works by testing
hypotheses, conducting controlled studies to rule out alternative expla-
nations, and drawing conclusions based on the evidence. Multiple stud-
ies yield mounting evidence either supporting or refuting hypotheses. In
the case of scientific research on learning to read, hypotheses have been
derived from theories about how learning occurs. Theories have been
supported or modified to accommodate the evidence. Over time, a
clearer picture of how students learn to read has emerged. This article is
intended to recount and illustrate with specific studies how the science
of learning to read words has evolved in our laboratory, what findings of
studies have shown, and how results support the importance of begin-
ners receiving systematic phonics instruction when they learn to read.
Although most of our studies have been conducted in English, we
­present some evidence for their relevance in other alphabetic languages.
Reading Research Quarterly, 55(S1) In order to carry out studies that support inferences about cause–
pp. S45–S60 | doi:10.1002/rrq.334 effect relations uncontaminated by erroneous factors, we have employed
© 2020 International Literacy Association.

S45
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
controlled experiments. Pretests are given to assess entry- Goodman (1970) drew his evidence from an analysis
level literacy skills (e.g., letter knowledge, word decoding) of readers’ oral word-reading errors while reading text,
that students need to participate in a study or that disqual- referred to as miscues. He and others observed that the
ify students as too advanced to participate. Also, pretests majority of errors preserved semantic and syntactic infor-
verify that treatment and control groups do not differ prior mation (Clay, 1968; Weber, 1970). Fewer errors reflected
to training. Random assignment is used to place students use of graphic or phonological cues. These findings were
in training and control groups. Standardized procedures interpreted to support his theory and to explain how all
are administered uniformly by research assistants working words are read in text, by sampling cues to guess words.
with individual students. Instruction is focused on teach- My background in psycholinguistics made me sympa-
ing specific knowledge or skills. Posttests measure effects thetic to Goodman’s (1970) theory. Certainly, readers hold
of the instruction. Statistical tests are conducted to assess semantic and syntactic expectations that are activated by
effects of training on outcome measures. These steps serve prior text when reading words. However, I was not con-
to increase the likelihood that results of studies support vinced that these govern all forms of word recognition.
hypotheses and rule out alternative explanations involving Readers read many more words correctly than incorrectly
factors that have been controlled. in any text that they can comprehend. Anything below
90% accuracy is considered frustration-level reading, so
miscues constitute only a small proportion of the words
Importance of Learning that are read. Cue sampling and guessing should cause
many more errors than actually occur and would hinder
to Read Words reading speed. A different process must explain how words
Our first attempt to study reading processes from a psy- are read correctly and quickly during text reading. This led
cholinguistic perspective involved an experiment exam- me to propose an alternative psycholinguistic theory sug-
ining whether embedding visual intonation cues in text gesting that readers read words accurately not by guessing
would improve third and fourth graders’ reading speed but by storing written words in memory and then reading
and comprehension (Ehri & Wilce, 1974). Written words them from memory by sight (Ehri, 1978). My collabora-
were assigned three different sizes to reflect levels of pitch tors and I sought evidence for this theory by conducting
and stress in spoken sentences. The intoned text was many studies over the years.
compared with a text with word sizes varied randomly
and a text with uniform word size. Results revealed that
third graders read the intoned text more rapidly than the Amalgamation of
other texts. Several follow-up studies were conducted, but
replications failed to show any benefit of the intoned text.
Word Identities Theory
The stumbling block became evident. Measuring text Following the institute, I wrote a paper proposing that be­­
reading speed was consistently foiled by word-reading ginners learn to read words from memory by amalgamat-
difficulties, so attention was redirected to a study of word- ing or bonding their various identities together to form
reading processes. single lexical units in memory. These identities include
orthographic (spellings), phonological (pronunciations),
morphological (word roots and affixes), syntactic (gram-
matical function in sentences), and semantic (meanings;
Psycholinguistic Guessing Game Ehri, 1978). Readers have already bonded some of these
I was introduced to a psycholinguistic theory of learning identities in memory from their competence with spoken
to read at an institute in 1974 where Ken Goodman spoke language. In order for written words to be added to the
about his research. His explanation of how students read amalgams in memory, readers must bond spellings to pro-
words was of special interest. According to his view, learn- nunciations by applying their knowledge of letter–sound
ing to read involves learning to gain meaning from print relations to connect letter units to sound units within spe-
(Goodman, 1970). Students become good readers by cific words. The letter–sound units might be grapheme–
improving their ability to predict words in text by attend- phoneme units, onset-rimes, syllables, or morphemes
ing to semantic, syntactic, and graphic cues. They do not depending on a reader’s knowledge of the writing system.
improve by reading words precisely by learning to decode In order to bond spellings to syntactic and semantic iden-
words, as this only causes them to bark at print and tities, readers have to read words in contexts where syn-
impedes the activation of meaning. Reading is a psycho- tactic and semantic identities are activated when the
linguistic guessing game that involves sampling cues. It is spellings are seen. The first few times a student reads a
more important that readers’ predictions are consistent word, these connections are formed and stored in mem-
with semantic and syntactic information than graphic ory. Subsequently, when the word is seen, these connec-
cues in the text. tions are activated in memory to read the word.

S46 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Once words’ identities are amalgamated in memory, Subsequent studies verified that readers store the spe-
readers can read them as whole units quickly and auto- cific spellings they read rather than phonologically equiva-
matically with all of their identities activated. This applies lent alternative spellings (Ehri & Saltmarsh, 1995; Reitsma,
to all words, not just high-frequency or irregularly spelled 1983). In these studies, students were taught to read unfa-
words. When practiced in this way, words become recog- miliar words. Subsequently, students’ latencies to read
nized from memory by sight. This supplants the need for originally seen and phonologically equivalent spellings of
guessing or decoding words. Acquisition of sight word these words were tested. Results showed that students read
learning ability requires that students learn regularities of the original spellings more quickly, indicating that they
the writing system beginning with grapheme–phoneme were reading these words from memory. Also, Share
relations, phoneme segmentation, and decoding skill, so (2008) showed better memory for original than phono-
graphemes become connected to phonemes within spell- logically equivalent unseen spellings when readers read
ings of specific words in memory. I sent the paper to new vocabulary words in text. Moreover, he showed that
Goodman, who returned it with comments. Not surpris- very few exposures were needed to remember the
ingly, he rejected the idea that reading is a process of rec- spellings.
ognizing individual words (Ehri, 1998). The next question we addressed was how spellings get
At the time, our theory differed from other theories as into memory. Spellings of words consist of graphemes
well (Ehri, 1980, 1992). The dual-route view identified two (i.e., single letters or letter combinations; e.g., sh) that
ways to read words: by phonological decoding to sound symbolize the smallest sounds or phonemes in pronunci-
out and blend words and by accessing memorized visual ations. For example, ship contains three grapheme–
forms (J. Baron, 1977; Coltheart, Davelaar, Jonasson, & phoneme relations: sh-i-p. According to the theory, spell-
Besner, 1977). In the latter case, readers bypassed phonol- ings of words are remembered when readers apply their
ogy to read words. Visual representations included letters knowledge of grapheme–phoneme correspondences to
or letter sequences forming orthographic patterns, shape connect the specific graphemes in spellings to phonemes
or length information, and were associated with meanings in their pronunciations and secure them in memory.
and stored in memory. No phonological information was Knowledge of the alphabetic system provides the glue
included. Learning written words thus required repeated that bonds orthographic to phonological identities and
exposure and practice to secure these arbitrary associa- establishes spellings of words in memory.
tions in memory. In contrast, our view suggested that We knew from a previous study that young students
visual representations were formed when letter–sound have difficulty in remembering words lacking much mean-
connections bonded spellings to pronunciations and ing (Ehri, 1976). In the next study, first and second graders
meanings in memory. Systematic knowledge of the writing were taught four spoken consonant-vowel-consonant
system as it mapped speech provided the glue that secured (CVC) nonsense syllables in a paired-associate learning
written words in memory, not arbitrary associations. task (e.g., jad, wek, sim, lut; Ehri & Wilce, 1979b). Students
were shown four stimulus prompts and tried to recall the
nonwords associated with each over several test trials with
corrective feedback. The prompts were either meaningless
Orthographic Identities line drawings or single letters. During study and feedback
These disagreements challenged us to conduct studies to periods but not when recall was tested, students were
provide evidence for our theory. One claim was that when shown correct spellings of the nonwords, misspellings, or
readers practice reading specific words, they store the no spellings, but no attention was drawn to them. Results
spellings in memory. Readers do not sound out and blend revealed that students recalled the spoken nonwords on
letters to decode words anew every time they are seen. test trials much better when they had been exposed to
Also, readers do not guess words. If decoding or guessing correct spellings during learning. This provided evidence
were used, then readers would not remember much about that grapheme–phoneme relations in the spellings helped
the spellings of the words. Spellings would be left out on secure their pronunciations in memory. Note that stu-
the page rather than retained in memory. In one study, sec- dents did not explicitly decode the spellings to benefit
ond graders practiced reading one of two sets of nonwords from their presence during learning because the experi-
with phonologically equivalent spellings and pronuncia- menter pronounced the words when they were shown.
tions, such as bistion or bischun pronounced identically This suggests that when the nonwords were seen and
(Ehri, 1980). Then, students wrote the words from mem- heard, grapheme–phoneme connections were activated
ory. Their spellings showed that students remembered the spontaneously to secure pronunciations in memory.
specific letters they had read rather than different but pho- We submitted this experiment to a journal for publica-
nologically equivalent spellings. For example, when stu- tion, but reviewers were skeptical. They pointed to the
dents misspelled bistion, they included st, never ch, whereas extra practice that seeing spellings provided over no spell-
those who saw bischun remembered ch. ings. So, we repeated the experiment to rule out alternative

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S47
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
explanations. Students in the comparison conditions were grapheme–phoneme connections between spellings and
not shown spellings, but they received extra practice dur- pronunciations explains how beginners learn to read
ing study and feedback periods by listening to orally spelled words from memory.
nonwords, by hearing the separate phonemes in the non- To obtain additional evidence, we performed another
words pronounced, or by repeating the nonwords extra study to examine whether teaching beginning readers to
times. Results showed that in all cases, students who saw use grapheme–phoneme connections to spell words
written spellings recalled the nonwords much better than would improve their word-reading ability (Ehri & Wilce,
students not exposed to written spellings. 1987b). We selected kindergartners who knew 10 target
To clinch the case, we added a final experiment. In letters used to spell words in the study but had limited
one condition, second graders heard how each nonword spelling ability. One group was taught to segment spoken
was spelled by listening to names of the letters, and then words into phonemes and spell them with target graph-
were told to imagine what it looked like. In the control emes. The control group was taught to select the same
condition, students simply pronounced the nonwords target graphemes corresponding to individual phonemes
extra times. Results were the same and showed that even they heard, but they did not use graphemes to spell words.
creating spellings in their minds facilitated students’ Following training, both groups were taught to read 12
memory for pronunciations of the words. Findings across similarly spelled words over several test trials with feed-
experiments convinced reviewers of the mnemonic value back. The words were spelled with the target letters prac-
of orthography for remembering pronunciations. In addi- ticed, and all began with s. Students who had received
tion, we observed a very strong correlation between stu- spelling instruction learned to read more words over tri-
dents’ recall of pronunciations when spellings had been als than controls learned. These findings provide addi-
seen and their word-reading ability (r = .75). This sup- tional support for the contribution of grapheme–phoneme
ported our claim that forming grapheme–phoneme con- connections in learning to read words from memory.
nections between spellings and pronunciations is the Also, findings show that spelling instruction benefits
mechanism explaining how students learn to read words beginners’ word reading. In fact, several studies have
from memory. reported high correlations between reading words and
To obtain more direct causal evidence, we conducted spelling words, as high as .86, indicating that reading
a training experiment (Ehri & Wilce, 1987a). Novice and spelling words rely on the same knowledge sources
beginning readers in kindergarten were taught the and skills (Ehri, 2000).
graphophonemic connection-forming process to see We applied our graphophonemic connection-forming
whether it improved their ability to learn to read words theory to examine how students’ memory for spellings
from memory. Students knew relevant letter sounds but might be improved. Because spellings of many English
had little word-decoding ability. One group practiced words are variable and less predictable, they can be hard
reading 12 successive sets of six to 10 similarly spelled to remember. We reasoned that one way to enhance
words and nonwords (total of 99 CVCs, CCVCs, and memory might be to have students create special spelling
CVCCs). The spellings were formed out of nine conso- pronunciations that regularize the connections between
nants and four short vowels (e.g., bap, dit, lob, pum, ras, letters and sounds in words, such as pronouncing “choc-
sun). Kindergartners practiced each set until they could lut” (chocolate) as “choc-o-late.” This should create more
read it perfectly. This required them to pay attention to complete connections between letters and sounds within
and process all the grapheme–phoneme connections words and hence should improve students’ recall of the
within the words. A comparison group practiced saying spellings. This possibility was tested in studies with fourth
sounds of the same letters and remembering a spoken graders (Drake & Ehri, 1984) and with adults (Ocal &
word beginning with each letter sound, but they did not Ehri, 2017). Results supported the hypothesis. Having
use letters to read words. At the end of training, students students create spelling-based pronunciations when they
were given several test trials with corrective feedback to studied a set of words improved their memory for the
learn to read 15 similarly spelled real words that had not spellings as compared with having students practice stan-
been taught but were composed of the trained letters (e.g., dard pronunciations.
spin, stab, stamp, stand). The benefit of having practiced As beginning readers build their English sight vocab-
the graphophonemic connection-forming process to ularies, they encounter spellings containing irregular or
attend to all the letters in words was clearly evident. This unexpected letters, such as the silent letters in talk, listen,
group learned to read 90% of the words accurately by the sword, and sign. This may interfere with decoding accu-
third test trial, whereas the letter sound group never read racy but not with sight word learning. Most of the pho-
more than 40% of the words correctly across seven test nemes are regularly spelled, so grapheme–phoneme
trials with corrective feedback. The latter group’s main connections can be formed to bond spellings to pronun-
problem was mixing up words sharing the same letters. ciations in memory, just as with regularly spelled words.
These findings provide causal evidence that forming In one study, we found that pronounced letters were

S48 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
remembered better than silent letters in words, but silent sentence-reading group learned more about semantic
letters were detected more rapidly (Ehri & Wilce, 1982). It identities. This group wrote the words in semantically cor-
may be that when silent letters are exceptions, they stand rect sentences, whereas the isolation-trained group wrote
out visually in spellings stored in memory. incorrect sentences that mismatched spellings and mean-
ings of the homonyms. This study provided further evi-
dence for word identity amalgamation theory by showing
Syntactic and Semantic Identities that the type of word-reading practice makes a difference.
Additional evidence was provided in a more recent study
Early in our research, we examined word consciousness (Miles & Ehri, 2017). We note that contrary to claims that
in beginning readers and prereaders (Ehri, 1975, 1976). young students should only read words in meaningful
Results indicated that prereaders had more difficulty in contexts (Goodman, 1970), there is value in reading words
recognizing words pronounced in isolation as units in both in isolation and in context.
their language, particularly function words (e.g., and, the,
should, was), as compared with students who could read
these words. Based on amalgamation theory, we reasoned Unitization and Automaticity
that reading the written forms of function words in mean-
ingful contexts might teach students to recognize the syn- in Reading Words
tactic and semantic identities of these words when heard According to amalgamation theory, when written words
in isolation. have become familiar and their identities have bonded
To study this learning process, we conducted an together in memory, the words are no longer decoded by
experiment with first graders who were taught to read 10 sounding out and blending letters sequentially. The words
context-dependent words, that is, words whose meanings are read as single visual spelling units. All of the identities
were activated by contexts but remained relatively mean- are activated automatically as soon as the words are seen.
ingless in isolation (e.g., while, which, must, from; Ehri & This reasoning led to a study comparing skilled and
Wilce, 1980a). One group practiced reading the words in less skilled first-, second-, and fourth-grade readers’
isolation and hearing meaningful sentences containing speed in reading familiar object words (e.g., book, man,
the words. The other group practiced reading the words cat, ball), number words (e.g., five, six, ten), and non-
embedded in written sentences that the other group words (e.g., nel, kiv, tuk) and in naming pictures of the
heard. Posttests were administered to assess readers’ familiar objects and single numbers (Ehri & Wilce,
memory for orthographic and syntactic/semantic identi- 1983). We reasoned that if readers read familiar words
ties of the words. Students who had read the words in iso- from memory, then they should read the object and
lation showed better memory for their orthographic number words faster than the unfamiliar nonwords that
identities. This group read the words faster on a timed have to be decoded by sounding out letters. If readers
reading test and remembered spellings of the words bet- read the familiar words as unitized wholes, then readers
ter. However, students who had read the words in con- should read them as fast as they can name single digits,
texts learned syntactic/semantic identities of the words which have already been unitized in memory. Com­
better. Upon hearing the words, this group embedded parison of the times taken by skilled readers in all
them in more complete and meaningful spoken sen- three grades supported expectations. These students
tences. Also, they detected the presence of more of these read words faster than nonwords and read them as
words when they listened to sentences containing the quickly as they named single digits. Also, these readers
words. These results were interpreted as providing evi- took less time to read object words than to name pic-
dence for word identity amalgamation theory. The par- tures of the objects, indicating that spellings activated
ticular identities that young readers learn are influenced object names faster than pictures activated their names.
by their experiences in reading the words. In contrast, less skilled readers did not show unitization
We conducted another similar experiment (Ehri & until fourth grade, and they were less accurate and took
Roberts, 1979). In this case, first graders were taught to much longer to read nonwords than real words. These
read one of two spellings of eight pairs of homonyms (e.g., results provide evidence that familiar words are read
bald/bawled, rows/rose, choose/chews). One group prac- from memory rather than decoded letter by letter, and
ticed reading the words embedded in written sentences are read as whole units by skilled readers as young as
that clarified their meanings. The other group read the first grade but not by less skilled readers until after sec-
words on flashcards and listened to the same sentences ond or third grade. According to amalgamation theory,
spoken. As in the previous study, posttests revealed that unitization occurs when grapheme–phoneme connec-
the isolated-word, flashcard group learned orthographic tions fully bond spellings to pronunciations in memory,
identities better as evidenced by faster reading times and thus creating immediate access from written to spoken
better memory for spellings. However, the context-trained, words when they are seen.

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S49
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Another characteristic of skilled word reading is auto- and sounds in pronunciations to read and spell. The
maticity (LaBerge & Samuels, 1974). One task showing decoding stage emerges later.
that readers recognize words even when they try to ignore This disagreement led to a study distinguishing be­­
them is the picture–word interference task patterned after tween the visual cue reading stage and the partial alpha-
the Stroop (1935) color-naming task. Readers are shown betic stage (Ehri & Wilce, 1985). Kindergartners were
drawings of objects. Spellings that name different objects screened for reading ability. Prereaders were students
are printed on the pictures (e.g., a drawing of a horse with who knew very few letter sounds and read few, if any,
cow printed on it). Even though readers are told to name preprimer words. Partial alphabetic readers were stu-
the pictures and ignore the words, they have difficulty. dents who knew most letter sounds. These students
The presence of competing words creates interference could read a few words but could not decode new words.
and slows readers down, as compared with the presence Students were taught to read two types of words over
of nonwords or no words (Rosinski, Golinkoff, & Kukish, several learning trials. One type was spelled phonetically
1975). with letters mapping some of the sounds in words (e.g.,
We conducted a study to examine whether teaching JRF for giraffe). The other type was spelled with visually
first graders to read unfamiliar words would increase the distinctive but nonphonetic letters (e.g., WBC for giraffe).
amount of interference that the words caused in a picture- We reasoned that prereaders should rely more heavily on
naming task (Ehri & Wilce, 1979a). We reasoned that if distinctive visual cues in reading words, whereas stu-
these words caused more interference after than before stu- dents with letter–sound knowledge should be able to use
dents learned to read the words, this would show that stu- this knowledge to learn phonetically spelled words.
dents had learned to read the words automatically. First Results were supportive. Prereaders learned to read the
graders were taught to read 20 nouns that they could not visually distinctive spellings better than the phonetic
read before training (e.g., flag, horse, wagon, apple). Both spellings, indicating that they were using visual nonal-
before and after training, students completed a picture– phabetic cues to remember how to read the words. In
word interference task. The taught nouns naming objects contrast, partial alphabetic readers learned to read the
were imposed on pictures of different objects. Results phonetic spellings better than the visual spellings, indi-
showed that students took longer to name the pictures on cating that they were using alphabetic cues. These results
the posttest than on the pretest. A follow-up experiment provided evidence for a middle stage between a visual
showed that without any word training, picture-naming cue reading stage and a decoding stage. These findings
time did not change from pretest to posttest. It is notewor- have since been replicated in other studies (Bowman &
thy that taught words slowed students down in naming Treiman, 2002; de Abreu & Cardoso-Martins, 1998;
pictures, even though they were not pronouncing the Roberts, 2003; Scott & Ehri, 1990).
words but rather were ignoring them. Findings support the Subsequently, we proposed a theory of word-reading
claim that when learning to read words from memory, stu- development consisting of four phases rather than two
dents become able to read them automatically. stages (Ehri, 1987, 1992, 2005a). We labeled the phases to
reflect the predominant type of knowledge that readers
apply to read and spell words. Readers in the pre-alphabetic
Development in Learning phase rely primarily on visually salient cues and context
cues but not letter–sound cues to read and write words.
to Read Words: Phase Theory Readers move to the partial alphabetic phase when they
The next focus of our research program was on how can use their knowledge of letter names or sounds to read
word-reading skill emerges in beginners. Gough, Juel and write but cannot decode unfamiliar words. These stu-
and Roper/Schneider (1983) proposed a two-stage the- dents read and write the majority of words using partial
ory to explain the development of word-reading ability. letter–sound connections. Readers move to the full alpha-
According to the theory, beginners start out reading betic phase when they have acquired decoding skill and
words by using visual or contextual cues associated with can fully analyze and form grapheme–phoneme connec-
written words, such as the tail on the end of dog and the tions within words to read and spell them from memory.
humps in the middle of camel. Once students learn Readers move into the consolidated alphabetic phase
grapheme–phoneme relations, they shift and use these when they have accumulated fully analyzed spellings of
relations to decode words. We disagreed and argued that many words in lexical memory and, as a result, have
a middle, partial stage had been overlooked (Ehri & acquired knowledge of larger consolidated spelling pat-
Wilce, 1985). Once students learn some basic grapheme– terns representing spoken syllables and morphemes.
phoneme relations but before they can decode new These readers can use these larger units to decode multi-
words, they are able to use their letter–sound knowledge syllabic words and to form connections to read and spell
to form partial connections between letters in spellings multisyllabic words from memory.

S50 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
The type of knowledge that young learners use to and 6 years. The researchers examined whether growth in
read and spell words overlaps across phases. At any point spelling conformed to alphabetic phase theory or syllabic
in development, students will exhibit use of more than stage theory. Students’ responses were classified into one
one type of knowledge. However, their phase is deter- of three phases and stages based on whether more than
mined by the most commonly used type: pre-, partial, half of their spellings conformed to that phase or stage at
full, or consolidated alphabetic. We have conducted sev- each test point. Results showed that movement across
eral studies to identify the literacy skills of students test points was more consistent with growth from pre-
in each phase and their movement from one phase to alphabetic to partial alphabetic phases than from pre-
the next. syllabic to syllabic stages. Cardoso-Martins and colleagues
concluded that the partial alphabetic phase offers a more
accurate description of Brazilian students’ development in
understanding how print maps speech. Also, the findings
Pre- to Partial Alphabetic Phase show the relevance of phase theory for an alphabetic lan-
Movement from the pre-alphabetic to the partial alpha- guage other than English.
betic phase was revealed in several studies. In one study One issue that has divided researchers involves the
described earlier, pre-alphabetic phase readers learned to optimal spelling–sound unit to teach beginners to read.
read visually distinctive but nonphonetic spellings, such Those advocating larger units argue that it is easier for
as WBC for giraffe, more easily, whereas partial phase readers to detect syllables or onset-rimes in speech than to
readers learned to read phonetic spellings, such as JRF for detect phonemes (Ferreiro & Teberosky, 1986; Goswami
giraffe, more easily (Ehri & Wilce, 1985). In another study, & Bryant, 1990). Those advocating small units argue that
preschoolers’ ability to identify environmental print in because alphabetic writing systems represent speech at the
and out of contexts was examined (Masonheimer, Drum, level of phonemes, grapheme–phoneme knowledge is key
& Ehri, 1984). Young readers who could read words, such at the outset. We examined whether Portuguese-speaking
as Pepsi, in context failed when the print was displayed in first graders in the pre-alphabetic phase would benefit
isolation. Also, when shown Xepsi printed on Pepsi’s red more from instruction in grapheme–phoneme units or
and blue logo, preschoolers still read it as Pepsi and failed syllabic spelling units in learning to read and write words
to detect the error even when asked whether there was a (Sargiani, Ehri, & Maluf, 2019). Students knew letters but
mistake, revealing that these pre-alphabetic readers were could not read or write words. They were taught to read
not attending to letters. There were a few exceptions. Five eight sets of five CVs (40 total) composed of 10 conso-
out of six preschoolers who showed some word-reading nants and five vowels. One group was taught to sound out
ability detected the misspelling, whereas none of the 96 and blend graphemes in the CVs. The other group was
prereaders did. taught to read the CVs as whole syllables. Both groups
Similar results were found in a study with Israeli pre- were taught to a mastery criterion. Results on posttest
schoolers (Levin & Ehri, 2009). Most preschoolers could transfer tasks showed that the grapheme–phoneme group
read several personal names printed on cubbies in their read and spelled new words more accurately than the syl-
classrooms. However, when the names were shown in iso- lable group. Findings indicate that despite the greater
lation, only those preschoolers who knew many letters accessibility of syllables than phonemes in spoken
were able to recognize the names out of context. Results Portuguese, teaching grapheme–phoneme relations better
of these studies point to alphabetic knowledge as the abil- prepares pre-alphabetic phase readers to move into the
ity that enables readers in the pre-alphabetic phase to partial phase to read and spell words than teaching syl-
move to the partial alphabetic phase in reading words. labic units.
The relevance of phase theory for young learners
learning to read in Portuguese was addressed by Cardoso-
Martins, Corrêa, Lemos, and Napoleão (2006). An alter-
native to phase theory is Ferreiro and Teberosky’s (1986) Partial to Full Alphabetic Phase
syllabic stage theory, which dominates educators’ views Movement from the partial to the full alphabetic phase of
about beginning reading instruction in Brazil. The theory development was examined in two of our studies already
postulates three stages: presyllabic, syllabic, and alpha- described. Students were in the partial phase. They knew
betic. The middle stage corresponds to the partial phase letter–sound relations but could not decode new words.
and postulates that young learners detect syllables in spo- Treatment groups were taught to use grapheme–phoneme
ken words and spell them by writing one letter for each connections either to read many similarly spelled words
syllable before they spell alphabetically in the next stage. (Ehri & Wilce, 1987a) or to spell words (Ehri & Wilce,
Cardoso-Martins and colleagues examined Brazilian 1987b). Control groups practiced grapheme–phoneme rela-
Portuguese-speaking students’ spelling development in a tions in isolation. Training was expected to improve stu-
longitudinal study with periodic testing between ages 4 dents’ ability to form more complete grapheme–phoneme

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S51
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
connections between spellings and pronunciations of words reading speed to assess unitization described earlier
and hence move them closer to the full alphabetic phase. (Ehri & Wilce, 1983), poorer readers in first and second
Following training, students were given several trials to grades were less accurate and took much more time to
learn to read 12 to 15 similarly spelled words that required decode nonwords than skilled readers did. Also, poorer
paying attention to all the letters to read the words accu- readers took longer to read familiar words than to name
rately (e.g., blast, blond, stab, stamp, stand, lamp, lap, seal, single digits, suggesting that they had not formed com-
seats). The words were spelled with the same letters prac- plete grapheme–phoneme connections to read the words
ticed during training. Results showed that both types of as whole units. In another study (Ehri & Saltmarsh,
connection-forming instruction enabled treatment groups 1995), students with a reading disability showed evi-
to outperform control groups in reading these words. dence of partial cue reading. They were taught to read a
Movement from the partial to the full phase was especially set of words. On a test afterward, their latencies to read
impressive in the word-training study. On the word- original and altered spellings of the words indicated that
learning posttest, students taught to be full alphabetic phase students with a reading disability recognized when ini-
readers learned to read 90% of the similarly spelled words tial and final letters had been changed but not medial let-
on average within three trials, whereas the partial alphabetic ters. In contrast, typically developing readers recognized
phase readers never learned to read more than 40% of the letter changes in all positions in words.
words on average after seven trials. One reason was that the
latter students were confusing words sharing the same let-
ters (e.g., drip, drum, dump), hence revealing the limitation
of partial cues to read words. Consolidated Alphabetic Phase
The ability to decode new words marks entry into the Movement into the consolidated phase of development
full alphabetic phase. A synthetic procedure for decoding occurs when students acquire knowledge of multiletter
words is to say the phoneme corresponding to each graph- spelling–sound units and apply them to read words. This
eme and then blend them to pronounce the word. Learning knowledge may be acquired implicitly from extensive
this procedure is hindered when schwa vowels are added to word-reading experience. However, acquisition is more
stop consonants and have to be deleted during blending likely facilitated by explicit instruction in reading words
(e.g., /tǝ/ /a/ /pǝ/ for top). We conducted a study to see using onset-rime units, syllabic units, or morphemic
whether this hindrance could be overcome (Gonzalez-Frey spelling–meaning units. We compared explicit and implicit
& Ehri, in press). We compared two methods of teaching instruction with students in grades 6–10 who exhibited
decoding to kindergartners in the partial phase who knew below-average word-reading skill, scoring between the
letter sounds but could not decode nonwords. Students third- and fifth-grade reading levels (Bhattacharya & Ehri,
were taught to decode CVC nonwords containing continu- 2004). We examined whether teaching students to segment
ant consonants, which allowed phonemes to be stretched and blend syllabic units in 100 multisyllabic words and
and connected without interruption from schwa vowels providing extensive practice would improve their word
(“sssaaannn”). Students in the connected condition were reading. Applying this routine was expected to connect
taught to stretch and pronounce phonemes without break- written and spoken syllabic units and bond spellings of
ing the speech stream before blending. Students in the seg- words to their pronunciations in memory, and also to teach
mented condition were taught to stretch and say each students syllabic spelling units for use in reading other
phoneme but to break the speech stream between pho- words. This training was compared with two alternative
nemes (“sss-aaa-nnn”) before blending. Following learning conditions: an implicit learning condition in which stu-
to criterion, students completed a transfer test to decode 20 dents practiced reading the same 100 words as whole units
CVCs with stop consonants that are harder to blend because repeatedly and a no-treatment control condition.
of intrusion from schwa vowels when stops are pronounced On posttests following instruction, the biggest differ-
in isolation. Results showed that during training, kinder- ences were detected among students reading at a third-
gartners who received connected practice learned to decode grade-equivalent level. Syllable-trained students decoded
the nonwords more quickly, and on the transfer test, they words and pseudowords better and remembered more
read nonwords with stops more accurately than the seg- spellings of words than the other two groups did. Whole-
mented group. An error analysis revealed that breaking word instruction yielded little benefit on posttests.
between phonemes caused students in the segmented con- Students reading at the fourth- and fifth-grade levels were
dition to forget initial phonemes during blending. These already able to read our multisyllabic words, so fewer dif-
findings suggest how to teach decoding more effectively to ference were detected on posttests. These findings indi-
help students move into the full alphabetic phase. cate that teaching readers to move into the consolidated
Two of our studies have indicated that older strug- alphabetic phase by analyzing syllabic spelling units in
gling readers behave more like partial alphabetic phase multisyllabic words benefits their word reading and spell-
than full alphabetic phase readers. In the study of ing and generalizes to new words not taught.

S52 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Teaching students to analyze morphemic spelling– with letter shape and sound pictures than with pictures in
sound units in words is another approach to instruction which letters were not shaped like the objects (e.g., the let-
at the consolidated alphabetic phase. In one study, we ter s taught with a snake stretched out).
compared the effects of two kinds of vocabulary instruc- In a more recent study (Shmidman & Ehri, 2010), we
tion on reading skills of adult struggling readers who created letter shape–sound picture mnemonics to teach
were seeking alternative high school diplomas (Gray, 10 unfamiliar Hebrew letter–sound relations to English-
Ehri, & Locke, 2018). They received eight hours of speaking preschoolers who were in the pre-alphabetic
scripted tutoring to learn 40 academic vocabulary words phase. They knew no Hebrew letters and could read few,
embedded within a civics curriculum. One group was if any, English preprimer words. In the shape–sound
taught to analyze morpheme and syllable structures of mnemonic condition, students were taught letter shapes
words and morphophonemic origins of words. The con- that resembled drawings of objects whose English names
trol group received traditional whole-word instruction began with the letters’ sounds (e.g., ‫ד‬, desk, /d/; ‫ש‬, ship,
that taught words in multiple-sentence contexts, mean- /š/). In the control condition, students were taught letters
ingful connections among words, and spellings. Both that were associated with the same objects and names but
groups made comparable gains in learning the target drawn in a different shape from the letters. Results
words, but the morphophonemic group showed greater showed that preschoolers mastered letter–sound associa-
pre- to posttest gains on transfer tasks of reading words tions in fewer trials when taught with letters resembling
and pseudowords. Findings suggest the value of explicit object shapes. On transfer posttests, preschoolers were
instruction in word analysis to increase readers’ linguistic better able to use the Hebrew letters to read and write par-
awareness of morphological, phonological, and ortho- tial consonant spellings of English spoken words (e.g., ‫ש ד‬
graphic structures within words. to read or spell dish). These findings suggest that letter–
sound instruction can be improved by teaching students
letter sounds with shape–sound mnemonics.
Letter Knowledge and Phonemic We also examined ways to teach phonemic segmenta-
tion and its impact on learning to read words at the par-
Awareness as Foundational Skills tial alphabetic phase. Following up on an earlier study
We were especially interested in studying the foundational (Castiglioni-Spalten & Ehri, 2003), two ways to teach
knowledge and skills needed for young students to move phonemic segmentation were compared (Boyer & Ehri,
from the pre-alphabetic phase to the partial and full alpha- 2011). In one condition, beginning readers were taught to
betic phases in learning to read words from memory (Ehri use mouth pictures and letters to segment words and
& Roberts, 2006). Two foundational skills were thought to nonwords into phonemes and to spell words. To illustrate,
be central: letter knowledge and phonemic segmentation. students learned to segment the nonword /pof/ into pho-
We reasoned that in order to remember how to read words nemes by selecting three pictures, first showing the lips
using grapheme–phoneme connections, beginners need to closed for /p/, then the lips rounded and open for /o/, and
know letter shapes, names, and sounds. Also, beginners then the upper teeth resting on the lower lip for /f/. Then,
need phonemic segmentation skill so they can detect in students spelled the word, p-o-f. In the other condition,
pronunciations the separate phonemes to be connected to students were taught to segment and spell the same words
graphemes. We conducted studies to examine acquisition using just the letters. Both groups were taught to crite-
of both letter knowledge and phonemic segmentation and rion. A control group received no instruction. Teaching
their contribution to reading words. beginners to segment using articulation along with letters
The task of learning letter–sound relations requires was expected to strengthen the connections between
learning shapes and sounds and forming associations graphemes and phonemes. This was based on the motor
between the two. In classrooms, alphabet posters typically theory of speech perception suggesting that articulation is
display each letter accompanied by a picture whose name more central to the representation of phonemes in the
begins with the letter’s sound but whose shape is unre- brain than acoustic cues are (Liberman, 1999).
lated to the letter. Picture mnemonics that incorporate Results on posttests showed that both of the trained
both letter shapes and sounds were expected to be more groups segmented untaught words into phonemes and
effective in helping prereaders learn letter–sound associa- spelled words better than controls did. Importantly, fol-
tions, such as the letter s drawn as a snake and taught as lowing training, students were taught to read a set of
representing its initial sound, /s/ (Ehri, Deffner, & Wilce, words spelled with letters that they had used during seg-
1984). We expected training to benefit memory by mentation training. Students practiced reading six words
enabling students to see the bare letter, be reminded of with feedback over eight trials. Results showed that the
the shape of the snake and its initial sound, and then letter/mouth group learned to read the words from mem-
recall /s/. Results showed that students learned the sounds ory more easily than the letter-only group and that both
of letters more readily when the associations were taught groups far surpassed the control group. The favored

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S53
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
explanation is that teaching beginners to monitor mouth To test this, one group of second graders was taught to
positions served to activate the articulatory features of read the words and another group to repeat the spoken
phonemes in words as students practiced reading them. words without seeing spellings. A rhyming task was then
This strengthened phonemes’ connection to graphemes given to examine how students perceived the flaps in the
and better secured spellings in memory for reading the spoken forms of these words (e.g., “Does the first syllable
words. Findings suggest the value of teaching beginners in meteor rhyme with feet or seed?”). Results showed that
to monitor mouth positions and sounds during phone- students who had decoded spellings almost uniformly
mic segmentation instruction. identified the spoken phonemes as /t/ or /d/ according to
Results also showed that both forms of phonemic their spellings, whereas students who had not seen spell-
segmentation training enabled students to function at the ings were more likely to perceive the flaps as /d/. This
partial alphabetic phase in their word reading. In con- study provided more evidence that spellings are retained
trast, students who received no training showed little in memory when graphemes are connected to phonemes
ability to read words on posttests and, hence, remained at and that graphemes influence how readers perceive the
the pre-alphabetic phase. These results support the claim identities of spoken phonemes when there is ambiguity.
that letter knowledge and phoneme segmentation skill Other studies have also shown that alphabetic orthog-
are central in enabling readers to move from the pre- raphy influences how people process spoken language
alphabetic phase to the partial alphabetic phase of word- (Ehri, 1984). In studies of adults who have not learned
reading development. to read or who read in a nonalphabetic language such
as Chinese, tests showed that their phonemic awareness
was limited or nonexistent (Morais, Cary, Alegria, &
Impact of Orthography Bertelson, 1979; Read, Zhang, Nie, & Ding, 1986). Among
adults who are literate, knowing the spellings of words has
on Phonological Processing been found to influence their detection of rhyming words
According to amalgamation theory, when students learn in a spoken judgment task. Seidenberg and Tanenhaus
to read and spell words, a visual alphabetic representa- (1979) had adults listen to several target words and decide
tional system for speech is acquired and used to store whether each word rhymed with a cue word. Some targets
words in memory. Letters in spellings come to penetrate shared spellings with cue words (e.g., clue/glue), and other
and represent phonemes in pronunciations in the brain. targets rhymed but had different spellings (e.g., clue/shoe).
Various lines of research have shown that learning spell- Some targets did not rhyme but shared spellings (e.g.,
ings impacts phonological processes and memory for bomb/tomb), and other targets did not share spellings or
spoken words. rhyme (e.g., bomb/room). Results showed that “yes”
In a phoneme segmentation task, we showed that responses were faster to rhyming words when they shared
fourth graders’ conception of phonemes in words was spellings than when they did not. “No” responses were
influenced by graphemes in the spellings of the words slower to nonrhyming words when they shared spellings
(Ehri & Wilce, 1980b). For example, students segmented than when they did not. These results show the impact of
pitch into four phonemes (/p/, /I/, /t/, and /č/), whereas orthography on phonology even when words are only
they segmented rich into three phonemes (/r/, /I/, and /č/) spoken without any spellings present. In our view, the
despite both words containing the same spoken VC rime. impact occurs because spellings are bonded to pronunci-
This was interpreted to be a consequence of readers form- ations in memory and are activated even when words are
ing connections between graphemes and phonemes to spoken.
retain written words in memory. The extra letter t caused
readers to detect the presence of /t/ in pitch but not in
rich, which lacks t in its spelling. Impact of Orthography
To show that spellings were causal in their influence
on phonemic analysis, we conducted a training study on Vocabulary Learning
(Ehri & Wilce, 1986). Words containing an ambiguous Results of a study reviewed earlier showed that beginning
phoneme, a medial alveolar flap, were selected. These readers recalled the pronunciations of nonwords better
flaps may be spelled either with a t or a d (e.g., meteor, when they had been exposed to spellings of the words dur-
glitter, attic; huddle, modify, pedigree) but are typically ing learning than when they had not seen spellings (Ehri
perceived and pronounced as the phoneme /d/ in words & Wilce, 1979b). We extended this research to explore
spoken in American English. We reasoned that teaching whether showing spellings helps students learn new
students to read these words would activate grapheme– vocabulary words (Ehri, 2005b; Rosenthal & Ehri, 2008).
phoneme connections and would lead students to con- Second and fifth graders were taught two sets of unfamil-
ceptualize the flap as /t/ or /d/ according to the spelling. iar nouns and their meanings over several learning trials.

S54 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
The words were defined, depicted, and embedded in sen- passive exposure to spellings and whether both treatments
tences. During study periods, students were shown spell- would boost word memory as compared with no spelling
ings of one set but not the other set. Spellings were not exposure (Chambrè, Ehri, & Ness, 2020). In the decoding
present when recall of pronunciations and meanings was condition, students sounded out and blended spellings
tested. Results showed that spellings enhanced students’ during study and feedback periods but not when memory
memory for pronunciations and meanings as compared was tested. In the exposure-only condition, spellings were
with no spellings. The explanation is that spellings acti- shown, but no attention was drawn to them. In the no-
vated grapheme–phoneme connections to better secure exposure condition, words were learned without spellings
pronunciations and meanings in memory and, hence, but spoken extra times. Students practiced recalling words
facilitated vocabulary learning. over several test trials with feedback.
These findings have been replicated by others under a Results revealed that students who decoded spellings
variety of conditions. Orthography has been found to facili- learned pronunciations and meanings better than students
tate vocabulary learning in several distinct populations. who were only exposed to spellings. Seeing spellings
These include students with autism spectrum disorders enhanced learning more than not seeing them. A spelling
(Lucas & Norbury, 2014; Ricketts, Dockrell, Patel, Charman, recall posttest showed that students more accurately wrote
& Lindsay, 2015), Down syndrome (Mengoni, Nash, & words that they had seen than words not seen, with decod-
Hulme, 2013), English learners and bilingual students ing producing better spelling recall than exposure only.
(Jubenville, Sénéchal, & Malette, 2014), college students This verifies that spellings were retained in memory. These
(Miles, Ehri, & Lauterbach, 2016; Rastle, McCormick, findings support the theory that exposure to spellings acti-
Bayliss, & Davis, 2011), and students with specific language vates grapheme–phoneme connections to better secure
impairments or reading disabilities (L.S. Baron et al., 2018; spellings to pronunciations along with meanings in
Ricketts et al., 2015). However, orthographic facilitation memory. These connections are activated implicitly when
was less apparent in students reading Chinese characters spellings are simply exposed, but the connections are
(Li et al., 2016) and was absent in adolescents with visual strengthened when spellings are explicitly decoded. Re­­
impairments reading braille (Savaiano, Compton, Hatton & sults carry implications for vocabulary instruction, suggest-
Lloyd, 2016). ing that when pronunciations and meanings of vocabulary
Orthographic facilitation has been detected with stu- words are taught, students should be shown spellings and
dents who have learned to read. We examined whether should decode them.
prereaders who know letter names but are not yet reading
might use their letter knowledge to show orthographic
facilitation (O’Leary & Ehri, 2020). Four- and 5-year-olds
were given a proper name–learning task. They were Systematic Phonics Instruction
taught pronunciations of 10 made-up CV words that The course of development portraying how students learn
named drawings of distinctive characters (e.g., a pig with to read words evidenced in our theory and research is
wings named Fee). During study and feedback but not best aligned with the structure and goals of systematic
during the test trials, students were exposed either to phonics instruction, particularly in the primary grades.
phonetic spellings of the names (e.g., FE) or to unrelated This instruction provides the foundational knowledge
numbers (e.g., 62). No attention was drawn to print. that launches students’ development as alphabetic readers
Students learned the names better when they had seen and enables them to move through the phases. Scope and
spellings than numbers. These findings reveal that even sequence charts specify the major grapheme–phoneme
prereaders with letter knowledge can spontaneously use relations that must be mastered and their order of presen-
the sound values in letters to connect spellings to pronun- tation. Phonemic awareness instruction teaches students
ciations and enhance their memory for proper names. to segment and blend phonemes in spoken words. A rou-
Studies of orthographic facilitation have differed in tine for decoding words enables students to read unfamil-
whether the effect resulted from explicit or implicit learn- iar words and to store spellings of these words in memory.
ing. In several studies, no attention was drawn to spellings Spelling instruction helps students remember complete
when shown during learning, yet orthographic facilitation spellings of words. Decodable books provide beginners
was observed, revealing that the boost to word memory with practice in applying the grapheme–phoneme rela-
resulted from automatic activation of grapheme–phoneme tions that they have learned to decode words and to build
correspondences when spellings were seen and pro- their sight vocabularies. Reading words in meaningful
nounced. In other studies, the effect occurred when stu- contexts ensures that syntactic and semantic identities of
dents’ attention was directed at spellings during learning. words become bonded to spellings and pronunciations to
We wondered whether having first graders explicitly form amalgamated units in memory. Building a store of
decode the spellings of vocabulary words would improve sight words that can be read as single units from mem-
their memory for pronunciations and meanings over ory automatically is essential for students to read and

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S55
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
comprehend text. This allows readers to focus their atten- Another form of phonics instruction is teaching stu-
tion on the meaning of the text while words are recog- dents to decode unfamiliar words by analogy to known
nized automatically out of awareness. words. We worked with teachers at a school for struggling
Not only beginning reading but also more advanced readers to advise them in applying phase theory to mod-
reading benefits from systematic phonics instruction ify a reading-by-analogy phonics program (Gaskins, Ehri,
focused on teaching multiletter units to decode words. To Cress, O’Hara, & Donnelly, 1996). In the original pro-
move into the consolidated alphabetic phase, students gram, students were taught to read 120 key words con-
need to be taught spellings units that include onset-rimes, taining the most common spelling patterns during their
syllables, and morphemes. Knowledge of these units first year of reading instruction. Students were taught to
enables students to decode unfamiliar multisyllabic words segment these words into onsets and rimes and used the
and to store these words in memory for sight word read- rimes to read new words. However, some students had
ing and spelling. difficulty in storing the key words in memory. They
We have conducted research beyond small-scale lab- behaved like partial alphabetic phase readers in misread-
oratory experiments to examine the value of systematic ing similarly spelled words and misspelling key words.
phonics instruction. Service on the National Reading The program was revised to help students analyze
Panel led us to conduct two meta-analyses examining the grapheme–phoneme connections as they learned to read
effectiveness of phonemic awareness instruction across and spell each key word. This was expected to help stu-
many studies (Ehri, Nunes, Willows, et al., 2001) and of dents retain complete spellings of the key words in mem-
systematic phonics instruction across many studies (Ehri, ory so they could use them to read new words. We
Nunes, Stahl, & Willows, 2001). Results showed that both compared the effectiveness of the new and old programs
forms of instruction were more effective than alternative and found that students receiving the new program read
forms lacking this instruction, such as whole language or and spelled words better during the first two years of
whole-word approaches in helping students learn to read instruction, but the differences between programs were
words. Mean effect sizes on word- and nonword-reading reduced during years 3 and 4 (Ehri, Satlow, & Gaskins,
tasks were moderate, with Cohen’s d = 0.53 for phonemic 2009). Results suggest the foundational importance of
awareness instruction and ds ranging from 0.40 to 0.67 graphophonemic analysis when teaching students to read
for systematic phonics instruction. More recent meta- words in systematic phonics programs.
analyses have supported the effectiveness of phonics Developing a systematic phonics program that could
instruction (Jeynes, 2008; National Early Literacy Panel, be provided to teachers online without cost was the goal
2008; Wanzek et al., 2018). of the EL (Expeditionary Learning) Education organiza-
Systematic phonics programs come in various forms. tion. In 2015, they sought our advice in designing the
The hallmark of traditional synthetic phonics programs is reading foundations skills block of their K–2 curriculum.
to teach students to decode words synthetically by saying Phase theory was applied in developing lessons to address
the phonemes corresponding to graphemes and blending objectives of the Common Core State Standards. The
them to pronounce the words. We evaluated a yearlong designers elaborated phase theory to create microphases
synthetic phonics, teacher-mentoring program (Ehri & portraying a more fine-grained course of development
Flugman, 2018). Teachers of grades K–3 in urban, lower from the pre-alphabetic phase to the consolidated alpha-
socioeconomic schools were coached in how to teach one betic phase. Each phase was divided into early, middle,
of two phonics programs (Gillingham & Stillman, 1997; and late microphases, with assessment and instruction
Spalding, 2003). Mentors with expertise each worked specified throughout. “The Skills Block is meant to ensure
with the same teacher twice a week throughout the school that, by the end of grade 2, students acquire the depth of
year. They helped teachers plan lessons, they modeled skills they need in the Reading Foundations standards to
how to teach phonics in the teachers’ classrooms, and navigate grade-level text independently” (EL Education,
they provided feedback as teachers taught phonics les- n.d., para. 1). This provides an example of the application
sons. Monthly ratings showed that teachers improved of our theory and research to the development of a sys-
their phonics teaching skills. Students’ reading and spell- tematic phonics program.
ing performance showed large gains by year’s end and far
exceeded effect sizes from comparable data sources on
both word-reading and comprehension measures. Stu­
dents met grade-level expectations at the end of kinder- Concluding Comments
garten and first grade but fell short in second and third Our theory and research add to the science of reading
grades. Findings revealed the effectiveness of an intensive debate in several ways. We provide an example of how an
teacher-mentoring model in how to teach phonics sys- extensive program of scientific research has clarified
tematically and its positive impact on students known for important ingredients and milestones that need to be
lower reading achievement. incorporated into beginning reading instruction to make

S56 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
it more effective. Our findings challenge instructional include syllabic and morphemic regularities and statisti-
approaches claiming that beginners can learn to read cal regularities. Seymour, Aro, and Erskine (2003) showed
whole words before they have acquired knowledge of that students learning to read in English take much
grapheme–phoneme relations. Without this knowledge, longer to become proficient than students reading in
students would remain in the pre-alphabetic phase. Our more transparent writing systems, such as Spanish,
findings challenge the view that prereaders will move into Finnish, or Greek.
reading through exposure to and practice in reading We suggest that phase theory is relevant across all
authentically written, meaningful storybooks without alphabetic writing systems when students move into
much attention paid to teaching them foundational skills. reading. The partial and full alphabetic phases describe
Without this, progress will be halting and limited. the beginning period when students learn and apply
Students may not function beyond the partial alphabetic grapheme–phoneme relations to read regularly spelled
phase. Our findings challenge the strategy of teaching words. Evidence cited earlier in Cardoso-Martins et al.’s
students that guessing words using syntactic and seman- (2006) study indicated that phase theory more accurately
tic cues is better than decoding words using graphic cues. portrayed Portuguese students’ development from the
Guessing does not build students’ lexical memory to sup- pre-alphabetic phase to the partial alphabetic phase than
port word-reading accuracy and automaticity. Ferreiro and Teberosky’s (1986) syllabic theory. Although
Systematic phonics instruction has been mischarac- Portuguese spoken words are syllabic, we found that
terized as only skill and drill, with little attention to mean- beginners learned to read and spell better when they were
ing. This is false. Phonics programs may use engaging taught grapheme–phoneme units than syllabic spelling–
games or interesting materials to teach letter–sound asso- sound units (Sargiani et al., 2019). Whereas the early
ciations, for example, letter shape–sound picture mne- period in learning to read is similar across alphabetic
monics such as Sammy Snake in the Letterland program orthographies, the later period during the consolidated
that Lyn Wendon created (see https://us.lette​rland.com/). alphabetic phase may diverge. The need to learn more
Students apply their letter–sound knowledge to decode complex spelling patterns as part of the English writing
words in meaningful texts from the outset. This was true system makes acquisition more complex and protracted
in the phonics programs described previously. Teaching than in transparent systems.
letter sounds and decoding necessarily occupies a larger Over the years, many other researchers have published
portion of instructional time until students master foun- influential theories and findings on reading processes and
dational skills. This enables students to function at the their development that have advanced our knowledge and
full and consolidated alphabetic phases and benefit fully improved instruction. Of special note are researchers who
from more advanced forms of text reading and writing. have proposed and studied theories resembling amalgama-
Our developmental theory is consistent with the tion theory to explain how people read words. Those theo-
approach to reading instruction studied by Connor et al. ries also posit the formation of connections among
(2009) and Juel and Minden-Cupp (2000). Their work orthographic, phonological, and semantic identities (ingre-
suggests that students initially benefit most from joint dients of triangle models) to read words from memory,
teacher/student-managed, code-focused phonics instruc- including Perfetti’s lexical quality hypothesis (Perfetti,
tion to learn the major grapheme–phoneme associations 1992; Perfetti & Hart, 2002), Seidenberg and McClelland’s
and how to decode and spell words. This applies to read- (1989) computational triangle model, and subsequent
ing acquisition during the partial and full alphabetic derivatives (Plaut, 2005). Whereas we and Perfetti view
phases. Once learned, students are ready to move into written words as single lexical units bonded to their vari-
more child-managed, meaning-focused instruction that ous identities and represented in memory, the computa-
includes more extensive text reading and writing. This tional models view written words as having distributed
occurs as students move into the consolidated alphabetic representations and resulting from the activation of con-
phase. Implementing this approach requires that teachers nections among many units in memory.
assess students’ skills to determine which type of instruc- In sum, the theory and research presented in this
tion is appropriate. This approach offers a way to resolve article show that teaching students to decode unfamiliar
the reading wars, by providing both structured phonics- words and enabling students to store spellings of familiar
and meaning-based instruction tailored to individual stu- words bonded to their other identities in memory should
dents’ phase of development. be central goals of beginning reading instruction.
Most of our studies have been conducted in English. Decoding is a means of getting spellings of words into
One issue is whether our theory and findings apply to stu- memory so they can be read by sight. Being able to con-
dents learning to read in other writing systems. English is nect letters in spellings to sounds in pronunciations
unique among alphabetic systems in that spellings of spontaneously when spellings of words are seen and
words are more variable and opaque. The sources of regu- heard also serves to retain words in memory. Both decod-
larity extend beyond grapheme–phoneme relations to ing and letter–sound mapping skills require knowledge

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S57
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
of the alphabetic writing system. Gradual acquisition of Ehri, L.C. (1976). Word learning in beginning readers and prereaders:
this knowledge propels students through the alphabetic Effects of form class and defining contexts. Journal of Educational
Psychology, 68(6), 832–842. https://doi.org/10.1037/0022-0663.68.
phases to become skilled readers.
6.832
Ehri, L.C. (1978). Beginning reading from a psycholinguistic perspec-
NOTE tive: Amalgamation of word identities. In F. Murray (Ed.), The devel-
I express extreme gratitude to my doctoral students and research col- opment of the reading process (Monograph No. 3, pp. 1–33). Newark,
leagues, especially Lee Wilce, for their contributions to the studies con- DE: International Reading Association.
ducted in my lab and reported here. Ehri, L.C. (1980). The development of orthographic images. In U. Frith
(Ed.), Cognitive processes in spelling (pp. 311–338). London, UK:
Academic.
REFERENCES
Ehri, L.C. (1984). How orthography alters spoken language competen-
Baron, J. (1977). Mechanisms for pronouncing printed words: Use and cies in children learning to read and spell. In J. Downing & R. Valtin
acquisition. In D.L. Laberge & S.J. Samuels (Eds.), Basic processes in
(Eds.), Language awareness and learning to read (pp. 119–147). New
reading: Perception and comprehension (pp. 175–216). Hillsdale, NJ:
York, NY: Springer-Verlag.
Erlbaum.
Ehri, L.C. (1987). Learning to read and spell words. Journal of Reading
Baron, L.S., Hogan, T.P., Alt, M., Gray, S., Cabbage, K.L., Green, S., &
Behavior, 19(1), 5–31. https://doi.org/10.1080/10862​96870​9547585
Cowan, N. (2018). Children with dyslexia benefit from orthographic
Ehri, L.C. (1992). Reconceptualizing the development of sight word
facilitation during spoken word learning. Journal of Speech,
Language, and Hearing Research, 61(8), 2002–2014. https://doi. reading and its relationship to recoding. In P.B. Gough, L.C. Ehri, &
org/10.1044/2018_JSLHR-L-17-0336 R. Treiman (Eds.), Reading acquisition (pp. 107–143). Hillsdale, NJ:
Bhattacharya, A., & Ehri, L.C. (2004). Graphosyllabic analysis helps Erlbaum.
adolescent struggling readers read and spell words. Journal of Ehri, L.C. (1998). Research on learning to read and spell: A personal-
Learning Disabilities, 37(4), 331–348. https://doi.org/10.1177/00222​ historical perspective. Scientific Studies of Reading, 2(2), 97–114.
19404​03700​40501 https://doi.org/10.1207/s1532​799xs​sr0202_1
Bowman, M., & Treiman, R. (2002). Relating print and speech: The Ehri, L.C. (2000). Learning to read and learning to spell: Two sides of a
effects of letter names and word position on reading and spelling coin. Topics in Language Disorders, 20(3), 19–36. https://doi.org/10.
performance. Journal of Experimental Child Psychology, 82(4), 305– 1097/00011​363-20002​0030-00005
340. https://doi.org/10.1016/S0022-0965(02)00101-7 Ehri, L.C. (2005a). Development of sight word reading: Phases and
Boyer, N., & Ehri, L.C. (2011). Contribution of phonemic segmentation findings. In M.J. Snowling & C. Hulme (Eds.), The science of reading:
instruction with letters and articulation pictures to word reading and A handbook (pp. 135–154). Malden, MA: Blackwell.
spelling in beginners. Scientific Studies of Reading, 15(5), 440–470. Ehri, L.C. (2005b). Learning to read words: Theory, findings, and
https://doi.org/10.1080/10888​438.2010.520778 issues. Scientific Studies of Reading, 9(2), 167–188. https://doi.org/10.
Cardoso-Martins, C., Corrêa, M.F., Lemos, L.S., & Napoleão, R.F. (2006). 1207/s1532​799xs​sr0902_4
Is there a syllabic stage in spelling development? Evidence from Ehri, L.C. (2017). Orthographic mapping and literacy development
Portuguese-speaking children. Journal of Educational Psychology, revisited. In K. Cain, D.L. Compton, & R.K. Parrila (Eds.), Theories
98(3), 628–641. https://doi.org/10.1037/0022-0663.98.3.628 of reading development (pp. 127–146). Amsterdam, The Netherlands:
Castiglioni-Spalten, M.L., & Ehri, L.C. (2003). Phonemic awareness John Benjamins.
instruction: Contribution of articulatory segmentation to novice Ehri, L.C., Deffner, N.D., & Wilce, L.S. (1984). Pictorial mnemonics for
beginners’ reading and spelling. Scientific Studies of Reading, 7(1), phonics. Journal of Educational Psychology, 76(5), 880–893. https://
25–52. https://doi.org/10.1207/S1532​799XS​SR0701_03 doi.org/10.1037/0022-0663.76.5.880
Chambrè, S.J., Ehri, L.C., & Ness, M. (2020). Phonological decoding Ehri, L.C., & Flugman, B. (2018). Mentoring teachers in systematic
enhances orthographic facilitation of vocabulary learning in first phonics instruction: Effectiveness of an intensive year-long program
graders. Reading and Writing, 33(5), 1133–1162. https://doi.org/10. for kindergarten through 3rd grade teachers and their students.
1007/s11145-019-09997-w Reading and Writing, 31(2), 425–456. https://doi.org/10.1007/
Clay, M.M. (1968). A syntactic analysis of reading errors. Journal of s11145-017-9792-7
Verbal Learning and Verbal Behavior, 7(2), 434–438. https://doi.
Ehri, L.C., Nunes, S.R., Stahl, S.A., & Willows, D.M. (2001). Systematic
org/10.1016/S0022-5371(68)80029-5
phonics instruction helps students learn to read: Evidence from the
Coltheart, M., Davelaar, E., Jonasson, J.T., & Besner, D. (1977). Access
National Reading Panel’s meta-analysis. Review of Educational
to the internal lexicon. In S. Dornič (Ed.), Attention and perfor-
Research, 71(3), 393–447. https://doi.org/10.3102/00346​54307​1003393
mance VI (pp. 535–555). Hillsdale, NJ: Erlbaum.
Ehri, L.C., Nunes, S.R., Willows, D.M., Schuster, B.V., Yaghoub-Zadeh,
Connor, C.M., Piasta, S.B., Fishman, B., Glasney, S., Schatschneider, C.,
Crowe, E., … Morrison, F.J. (2009). Individualizing student instruc- Z., & Shanahan, T. (2001). Phonemic awareness instruction helps
tion precisely: Effects of child × instruction interactions on first children learn to read: Evidence from the National Reading Panel’s
graders’ literacy development. Child Development, 80(1), 77–100. meta-analysis. Reading Research Quarterly, 36(3), 250–287. https://
https://doi.org/10.1111/j.1467-8624.2008.01247.x doi.org/10.1598/RRQ.36.3.2
de Abreu, M.D., & Cardoso-Martins, C. (1998). Alphabetic access route Ehri, L.C., & Roberts, K.T. (1979). Do beginners learn printed words
in beginning reading acquisition in Portuguese: The role of letter- better in contexts or in isolation? Child Development, 50(3), 675–
name knowledge. Reading and Writing, 10(2), 85–104. https://doi. 685. https://doi.org/10.2307/1128932
org/10.1023/A:10079​39610145 Ehri, L.C., & Roberts, T. (2006). The roots of learning to read and write:
Drake, D.A., & Ehri, L.C. (1984). Spelling acquisition: Effects of pro- Acquisition of letters and phonemic awareness. In D.K. Dickinson &
nouncing words on memory for their spellings. Cognition and S.B. Neuman (Eds.), Handbook of early literacy research (Vol. 2, pp.
Instruction, 1(3), 297–320. https://doi.org/10.1207/s1532​690xc​i0103_2 113–131). New York, NY: Guilford.
Ehri, L.C. (1975). Word consciousness in readers and prereaders. Ehri, L.C., & Saltmarsh, J. (1995). Beginning readers outperform older
Journal of Educational Psychology, 67(2), 204–212. https://doi.org/10. disabled readers in learning to read words by sight. Reading and
1037/h0076942 Writing, 7(3), 295–326. https://doi.org/10.1007/BF031​62082

S58 | Reading Research Quarterly, 55(S1)

 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Ehri, L.C., Satlow, E., & Gaskins, I. (2009). Grapho-phonemic enrich- processing and instruction (pp. 207–211). Rochester, NY: National
ment strengthens keyword analogy instruction for struggling read- Reading Conference.
ers. Reading & Writing Quarterly, 25(2/3), 162–191. https://doi.org/ Gray, S.H., Ehri, L.C., & Locke, J.L. (2018). Morpho-phonemic analysis
10.1080/10573​56080​2683549 boosts word reading for adult struggling readers. Reading and
Ehri, L.C., & Wilce, L.S. (1974). Research in brief: Printed intonation Writing, 31(1), 75–98. https://doi.org/10.1007/s11145-017-9774-9
cues and reading in children. Visible Language, 8(3), 265–274. Jeynes, W.H. (2008). A meta-analysis of the relationship between pho-
Ehri, L.C., & Wilce, L.S. (1979a). Does word training increase or nics instruction and minority elementary school student academic
decrease interference in a Stroop task? Journal of Experimental Child achievement. Education and Urban Society, 40(2), 151–166. https://
Psychology, 27(2), 352–364. https://doi.org/10.1016/0022-0965(79) doi.org/10.1177/00131​24507​304128
90055-9 Jubenville, K., Sénéchal, M., & Malette, M. (2014). The moderating
Ehri, L.C., & Wilce, L.S. (1979b). The mnemonic value of orthography effect of orthographic consistency on oral vocabulary learning in
among beginning readers. Journal of Educational Psychology, 71(1), monolingual and bilingual children. Journal of Experimental Child
26–40. https://doi.org/10.1037/0022-0663.71.1.26 Psychology, 126, 245–263. https://doi.org/10.1016/j.jecp.2014.05.002
Ehri, L.C., & Wilce, L.S. (1980a). Do beginners learn to read function Juel, C., & Minden-Cupp, C. (2000). Learning to read words: Linguistic
words better in sentences or in lists? Reading Research Quarterly, units and instructional strategies. Reading Research Quarterly, 35(4),
15(4), 451–476. https://doi.org/10.2307/747274 458–492. https://doi.org/10.1598/RRQ.35.4.2
Ehri, L.C., & Wilce, L.S. (1980b). The influence of orthography on LaBerge, D., & Samuels, S.J. (1974). Toward a theory of automatic infor-
readers’ conceptualization of the phonemic structure of words. mation processing in reading. Cognitive Psychology, 6(2), 293–323.
Applied Psycholinguistics, 1(4), 371–385. https://doi.org/10.1017/S01 https://doi.org/10.1016/0010-0285(74)90015-2
42​71640​0009802 Levin, I., & Ehri, L.C. (2009). Young children’s ability to read and spell
Ehri, L.C., & Wilce, L.S. (1982). The salience of silent letters in chil- their own and classmates’ names: The role of letter knowledge.
dren’s memory for word spellings. Memory & Cognition, 10(2), 155– Scientific Studies of Reading, 13(3), 249–273. https://doi.org/10.1080/
166. https://doi.org/10.3758/BF032​09217 10888​43090​2851422
Ehri, L.C., & Wilce, L.S. (1983). Development of word identification Li, H., Zhang, J., Ehri, L., Chen, Y., Ruan, X., & Dong, Q. (2016). The
speed in skilled and less skilled beginning readers. Journal of role of orthography in oral vocabulary learning in Chinese children.
Educational Psychology, 75(1), 3–18. https://doi.org/10.1037/0022- Reading and Writing, 29(7), 1363–1381. https://doi.org/10.1007/
0663.75.1.3 s11145-016-9641-0
Ehri, L.C., & Wilce, L.S. (1985). Movement into reading: Is the first Liberman, A. (1999). The reading researcher and the reading teacher
stage of printed word learning visual or phonetic? Reading Research need the right theory of speech. Scientific Studies of Reading, 3(2),
Quarterly, 20(2), 163–179. https://doi.org/10.2307/747753 95–111. https://doi.org/10.1207/s1532​799xs​sr0302_1
Ehri, L.C., & Wilce, L.S. (1986). The influence of spellings on speech: Lucas, R., & Norbury, C.F. (2014). Orthography facilitates vocabulary
Are alveolar flaps /d/ or /t/? In D.B. Yaden & S. Templeton (Eds.), learning for children with autism spectrum disorders (ASD).
Metalinguistic awareness and beginning literacy: Conceptualizing Quarterly Journal of Experimental Psychology, 67(7), 1317–1334.
what it means to read and write (pp. 101–114). Portsmouth, NH: https://doi.org/10.1080/17470​218.2013.859714
Heinemann. Masonheimer, P.E., Drum, P.A., & Ehri, L.C. (1984). Does environmen-
Ehri, L.C., & Wilce, L.S. (1987a). Cipher versus cue reading: An experi- tal print identification lead children into word reading? Journal of
ment in decoding acquisition. Journal of Educational Psychology, Reading Behavior, 16(4), 257–271. https://doi.org/10.1080/10862​
79(1), 3–13. https://doi.org/10.1037/0022-0663.79.1.3 96840​9547520
Ehri, L.C., & Wilce, L.S. (1987b). Does learning to spell help beginners Mengoni, S.E., Nash, H., & Hulme, C. (2013). The benefit of ortho-
learn to read words? Reading Research Quarterly, 22(1), 47–65. graphic support for oral vocabulary learning in children with Down
https://doi.org/10.2307/747720 syndrome. Journal of Child Language, 40(1), 221–243. https://doi.
EL Education. (n.d.). About Skills Block. Retrieved from https://curri​ org/10.1017/S0305​00091​2000396
culum.eledu​cation.org/skill​sbloc​k/about Miles, K.P., & Ehri, L.C. (2017). Learning to read words on flashcards:
Ferreiro, E., & Teberosky, A. (1986). Psicogênese da língua escrita Effects of sentence contexts and word class in native and non­
[Psychogenesis of the written language]. São Paulo, Brazil: Artes native English-speaking kindergartners. Early Childhood Research
Médicas. Quarterly, 41(4), 103–113. https://doi.org/10.1016/j.ecresq.2017.06.
Gaskins, I.W., Ehri, L.C., Cress, C., O’Hara, C., & Donnelly, K. (1996). 001
Procedures for word learning: Making discoveries about words. The Miles, K.P., Ehri, L.C., & Lauterbach, M.D. (2016). Mnemonic value of
Reading Teacher, 50(4), 312–327. orthography for vocabulary learning in monolinguals and language
Gillingham, A., & Stillman, B.W. (1997). The Gillingham manual: minority English-speaking college students. Journal of College
Remedial training for children with specific disability in reading, Reading and Learning, 46(2), 99–112. https://doi.org/10.1080/10790​
spelling, and penmanship (8th ed.). Cambridge, MA: Educators 195.2015.1125818
Publishing Service. Morais, J., Cary, L., Alegria, J., & Bertelson, P. (1979). Does awareness of
Gonzalez-Frey, S.M., & Ehri, L.C. (in press). Connected phonation is speech as a sequence of phones arise spontaneously? Cognition, 7(4),
more effective than segmented phonation for teaching beginning 323–331. https://doi.org/10.1016/0010-0277(79)90020-9
readers to decode unfamiliar words. Scientific Studies of Reading. National Early Literacy Panel. (2008). Developing early literacy: Report
Goodman, K.S. (1970). Reading: A psycholinguistic guessing game. In of the National Early Literacy Panel: A scientific synthesis of early lit-
H. Singer & R.B. Ruddell (Eds.), Theoretical models and processes of eracy development and implications for intervention. Jessup, MD:
reading (pp. 497–508). Newark, DE: International Reading National Institute for Literacy.
Association. Ocal, T., & Ehri, L.C. (2017). Spelling pronunciations help college stu-
Goswami, U., & Bryant, P. (1990). Phonological skills and learning to dents remember how to spell difficult words. Reading and Writing,
read. Hove, UK: Psychology. 30(5), 947–967. https://doi.org/10.1007/s11145-016-9707-z
Gough, P., Juel, C., & Roper/Schneider, D. (1983). Code and cipher: A O’Leary, R., & Ehri, L.C. (2020). Orthography facilitates memory for
two-stage conception of initial reading acquisition. In J.A. Niles & proper names in emergent readers. Reading Research Quarterly,
L.A. Harris (Eds.), Searches for meaning in reading/language 55(1), 75–93. https://doi.org/10.1002/rrq.255

The Science of Learning to Read Words: A Case for Systematic Phonics Instruction | S59
 19362722, 2020, S1, Downloaded from https://ila.onlinelibrary.wiley.com/doi/10.1002/rrq.334 by National Institutes Of Health Malaysia, Wiley Online Library on [23/10/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Perfetti, C.A. (1992). The representation problem in reading acquisi- Scott, J.A., & Ehri, L.C. (1990). Sight word reading in prereaders: Use of
tion. In P.B. Gough, L.C. Ehri, & R. Treiman (Eds.), Reading acquisi- logographic vs. alphabetic access routes. Journal of Reading Behavior,
tion (pp. 145–174). Hillsdale, NJ: Erlbaum. 22(2), 149–166. https://doi.org/10.1080/10862​96900​9547701
Perfetti, C.A., & Hart, L. (2002). The lexical quality hypothesis. In L. Seidenberg, M.S., & McClelland, J.L. (1989). A distributed, develop-
Verhoeven, C. Elbro, & P. Reitsma (Eds.), Precursors of functional mental model of word recognition and naming. Psychological
literacy (pp. 189–213). Amsterdam, The Netherlands: John Review, 96(4), 523–568. https://doi.org/10.1037/0033-295X.96.4.523
Benjamins. Seidenberg, M.S., & Tanenhaus, M.K. (1979). Orthographic effects on
Plaut, D.C. (2005). Connectionist approaches to reading. In M.J. rhyming monitoring. Journal of Experimental Psychology: Human
Snowling & C. Hulme (Eds.), The science of reading: A handbook Learning and Memory, 5(6), 546–554. https://doi.org/10.1037/0278-
(pp. 24–38). Malden, MA: Blackwell. 7393.5.6.546
Rastle, K., McCormick, S.F., Bayliss, L., & Davis, C.J. (2011). Seymour, P.H.K., Aro, M., & Erskine, J.M. (with COST Action A8 net-
Orthography influences the perception and production of speech. work). (2003). Foundation literacy acquisition in European orthog-
Journal of Experimental Psychology: Learning, Memory, and raphies. British Journal of Psychology, 94(2), 143–174. https://doi.
Cognition, 37(6), 1588–1594. https://doi.org/10.1037/a0024833 org/10.1348/00071​26033​21661859
Read, C., Zhang, Y., Nie, H., & Ding, B. (1986). The ability to manipu- Share, D.L. (2008). Orthographic learning, phonology and self-teach-
late speech sounds depends on knowing alphabetic reading. ing. In R.V. Kail (Ed.), Advances in child development and behavior
Cognition, 24(1/2), 31–44. https://doi.org/10.1016/0010-0277(86) (pp. 31–82). Amsterdam, The Netherlands: Elsevier.
90003-X Shmidman, A., & Ehri, L.C. (2010). Embedded picture mnemonics to
Reitsma, P. (1983). Printed word learning in beginning readers. Journal learn letters. Scientific Studies of Reading, 14(2), 159–182. https://
of Experimental Child Psychology, 36(2), 321–339. https://doi. doi.org/10.1080/10888​43090​3117492
org/10.1016/0022-0965(83)90036-X Spalding, R.B. (2003). The writing road to reading. New York, NY:
Ricketts, J., Dockrell, J.E., Patel, N., Charman, T., & Lindsay, G. (2015). HarperCollins.
Do children with specific language impairment and autism spec- Stroop, J. (1935). Studies of interference in serial verbal reactions.
trum disorders benefit from the presence of orthography when Journal of Experimental Psychology, 18(6), 643–662. https://doi.
learning new spoken words? Journal of Experimental Child org/10.1037/h0054651
Psychology, 134, 43–61. https://doi.org/10.1016/j.jecp.2015.01.015 Wanzek, J., Stevens, E.A., Williams, K.J., Scammacca, N., Vaughn, S., &
Roberts, T. (2003). Effects of alphabet letter instruction on young chil- Sargent, K. (2018). Current evidence on the effects of intensive early
dren’s word recognition. Journal of Educational Psychology, 95(1), reading interventions. Journal of Learning Disabilities, 51(6), 612–
41–51. https://doi.org/10.1037/0022-0663.95.1.41 624. https://doi.org/10.1177/00222​19418​775110
Rosenthal, J., & Ehri, L.C. (2008). The mnemonic value of orthography Weber, R. (1970). A linguistic analysis of first-grade reading errors.
for vocabulary learning. Journal of Educational Psychology, 100(1), Reading Research Quarterly, 5(3), 427–451. https://doi.org/10.2307/
175–191. https://doi.org/10.1037/0022-0663.100.1.175 747079
Rosinski, R.R., Golinkoff, R.M., & Kukish, K.S. (1975). Automatic
semantic processing in a picture-word interference task. Child Submitted April 30, 2020
Development, 46(1), 247–253. https://doi.org/10.2307/1128859 Final revision received June 10, 2020
Sargiani, R.A., Ehri, L.C., & Maluf, M.R. (2019, July). Brazilian Accepted June 11, 2020
Portuguese beginning readers benefit more from orthographic map-
ping of grapheme-phonemes than from syllables in learning to read
LINNEA C. EHRI is a distinguished professor in the Ph.D.
and spell words. Paper presented at the annual meeting of the Society
for the Scientific Study of Reading, Toronto, ON, Canada. Program in Educational Psychology at the Graduate Center of
Savaiano, M.E., Compton, D.L., Hatton, D.D., & Lloyd, B.P. (2016). the City University of New York, New York, USA; email lehri@
Vocabulary word instruction for students who read braille. gc.cuny.edu. Her research interests include how young students
Exceptional Children, 82(3), 337–353. https://doi.org/10.1177/00144​ learn to read and spell words, the course of development of
02915​598774 these abilities, and how to make instruction more effective.

S60 | Reading Research Quarterly, 55(S1)