"Automaticity Theory and EFL in Japan: With Some Specific
Applications for Reading"
by Fumiko Yoshimura
Kyushu Institute of Information Sciences
Japan

Introduction

Currently the overall emphasis of much language instruction has been
shifting from knowledge acquisition to that of communicative usage.
One of the most typical dilemmas that EFL instructors in Japan (and
elsewhere) face, however, is that even if we give our students a
significant amount of time in the classroom to use English in
communicative activities, they still do not acquire the proficiency
levels we might expect to justify such activities. What might be
missing in the process of our instruction and practice activities? In
trying to answer this question, in this article, I will introduce the
concept of Automaticity Theory (AT) and explain how we can apply the
theory to the actual EFL curriculum. This application, I contend, can
help to ensure that our language learners become able to coordinate
individual skills in more complex tasks leading to independence in
communicative ability.

Automaticity Theory

Automaticity Theory (AT) attempts to explain how people acquire
skills as a function of the automaticity of operating processes.
Schneider and Fisk (1983) explain the mechanisms of skill acquisition
in terms of contrasting automatic and controlled types of cognitive
and memory processing:

Automatic processing is a fast, parallel, fairly effortless
process which is not limited by short-term memory capacity, is not
under direct subject control and performs well-developed skilled
behaviors. Automatic processing typically develops when subjects deal
with the stimulus consistently over many trials....Controlled
processing is characterized as a slow, generally serial, effortful,
capacity limited, subject controlled processing mode that must be
used to deal with novel or inconsistent information. (p.120)

Schneider and Fisk (1983) also show how practice changes controlled
processing into automatic processing. According to them, auto-matic
productions are modular and will develop when the component processes
are consistent. This modular processing system can be hierar-chical,
with the same module being one part of many different skills. The
assumption is that there is an upper limit to human attention span.
Practice, however, can make automatic productions relatively free of
limited memory resources; thus there is no necessary limit to the
number of automatic processes which can be active at the same time.
Moreover, practice makes productions autonomous, reducing direct
conscious control of the subject. This is a crucial stage on which it
can be said that good and poor learners divide. Automaticity which
has been acquired through effective, repeated practice makes it
possible to process different stimuli at different stages
simultaneously, as in something like a psychological and pyscho-motor
version of a complex production line. Schneider and Fisk (1983)
illustrate this change of behavior in practicing a motor skill by
describing the change in learning how to play the piano. At the
novice level, performance is very slow, serial, and capacity limited.
Controlled processing is in effect at this stage and the learner must
allot much of finite attention capacity to each motor task. After
substantial practice, however, the learner builds up a vocabulary of
playable notes by consistently repeating each note in a given phrase
thousands of times. As the automatic productions develop, the
performer can speed up the responses, incor-porate more complicated
rhythm information, and begin to have sufficient capacity freed up
and made available to attend to the patterns of notes, familiar
scales and chords, and then finally onto entire sections in the
music. Figure 1 shows the continuum of automatization adapted from
Whitaker (1983, p.199):

< http://www.egroups.com/files/
fllsig-list/LAC+6+Graphics/yoshi1%7E1ver2.jpg >

(Note: you must fill out your egroup profile in order
to access this link to the graphic; also, if it doesn't work as a hot
link, try cutting and pasting it into a browser.)

According to Whitaker (1983), the stages of behavior acquisition are
best expressed as a continuum, not a dichotomy. Starting from the
left end of this continuum, we gradually acquire the automaticity of
a behavior with repeated practice. In learning a musical instrument,
people start from the novel (or novice) stage; with sufficient
practice and improvement, they acquire the skills necessary to play a
piece of music beautifully and fluently.

By way of comparison and contrast, Anderson (1995) perceives the
development of skill acquisition as the development of problem-
solving operators. He divides the processes into three stages; the
cognitive stage, the associative stage, and the autonomous stage. He
describes the general characteristics of each stage as follows: In
the cognitive stage, learners commit to memory a set of facts
relevant to the skill. Typically they rehearse these facts as they
first perform the skill. The process is slow. The information they
have learned amounts to a set of problem-solving operators for the
skill. In the associative stage, the connections among the various
elements required for successful performance are strengthened. Errors
are detected and eliminated as well. Learners, by this time, have
converted the verbal knowledge once memorized into procedural
knowledge. In the autonomous stage, the procedure becomes more
skilled, more automated, and more rapid. In becoming so, it requires
fewer and fewer attentional resources. Learners also develop more
complex skills in the direction of becoming more automated and
requiring fewer processing resources. Anderson says, "it is the
procedural, not the declarative, knowledge that governs the skilled
performance" (p.274).

Applying AT to a FL Curriculum

The models of skill acquisition described above show how people
develop automaticity with practice, and they break down a complex
process over time into understandable stages. In learning a foreign
language, just as with other skill acquisition processes, we must
start from an absolute beginning stage at which we have no language
and must progress over time until we have acquired language
proficiency. Ultimately, we hope to attain the stage where we can
exert control over language well enough to allocate our attention to
understanding and responding to the content of the messages -- to
actual communication. In real mainstream classroom instruction,
however, it is hard to see how the process of acquiring functional
proficiency levels over stages is actually acknowledged and dealt
with. In terms of input and what is the object of study, discourse is
typically broken down into smaller, discrete items for analysis and
manipulation. The items are typically grammar points, key vocabulary,
typical expressions, and the language associated with communicative
situations and functions, etc. For each isolated item, explanation
and opportunities for practice activities are often given.

Instructors assume that it is the learners' responsibility to
practice what has been covered in class until they have acquired the
target proficiency. On the other hand, many learners seem to think
they have practiced enough after only a few times, even if the
learners remain well short of being fluent and proficient in the
objective of study. Keeping in mind what instructors and learners
think about the matter, if we compare it to the Schneider and Fisk
(1983) model, we can see that these learners typically stay at the
faltering, controlled processing stage. In terms of the Anderson
(1995) model, it might be said that in much instruction the
associative stage is neglected; but it is at this stage where
learners come to coordinate many individual elements as a bridge to
the autonomy stage. Too many of our language learners never develop
skills to the point where they can perform more integrative and
complex tasks of language use, communication, and literacy.

They need to free up their cognitive and memory resources by becoming
fluent, automatic, and efficient at certain elements of processing in
order to devote their mental resources to more involved, complex
tasks of real communication and interaction. In short, they need to
stick it out with some practice tasks until stages of automaticity
have been reached. After practicing distinct skills until a fluency
with them has been reached, learners then need to practice them in
more integrative, less framed tasks. In so doing, they will also
learn how to balance their attention span; their cognitive and memory
resources can be more efficiently shared out to the various
integrated parts of increasingly complex tasks. Taking this into
consideration, we need to restructure the whole curriculum to
incorporate language training adapted to the associative stage of
Anderson's (1995) model. Figure 2 demonstrates a model which could be
applied to the structuring of the formal language curriculum around
the concepts of AT.

< http://www.egroups.com/files/
fllsig-list/LAC+6+Graphics/YOSHI2%7E1ver2.GIF >

(Note: you must fill out your egroup profile in order
to access this link to the graphic; also, if it doesn't work as a hot
link, try cutting and pasting it into a browser.)

In creating a curriculum, first the goals of our instruction need to
be defined. This means that we need to define the characteristics of
true communication which is at the autonomous stage of language
acquisition. Secondly, what is involved in a true communicative
situation needs to be analyzed and broken into distinct elements so
that teachers can incorporate them into their syllabus. Each element
should be taught so that learners can understand it and have
opportunities to practice it until they can use it without allotting
too much conscious effort. Employing communicative games and
activities are good because they are fun and create situations where
meaning is negotiated and exchanged.

The true challenge for language instructors, however, is how to
orchestrate the needs at different levels and come up with a
curriculum which helps learners to develop automaticity gradually and
systematically. This includes filling the discrepancy between the
overall goal of a language course and the goals of individual
lessons. In developing a motor skill such as driving or playing
sports, the most emphasized stage is not during but after individual
items or activities are practiced. After learning basic skills,
practice is given for larger units composed of the smaller isolated
skills already learned. The units of practice get larger and larger
until learners attain the goal of proficiency. In language learning,
incorporating integrative tasks is important because it gives
learners opportunities to use distinct skills in less framed, more
complex tasks. Making explicit clear goals is important as well in
order to motivate learners to practice the same underlying skills
over and over.

Rationale for Applying AT to EFL Instruction in Japan

AT is particularly applicable to EFL instruction in Japan and
elsewhere because of the following two reasons: (1) the age of the
language learners and (2) the lack of input and incidental learning
in the total FL learning environment. It is plausible to say that
many L1 development and cognition researchers (e.g. Fodor, 1985;
Jackendoff, 1996) contend that there is a special faculty for
language acquisition, and therefore language acquisition should be
treated as something significantly if not totally different from
other skills acquisition. But it is questionable if the contention is
applicable to FL learning, which often only starts to take place
after puberty. One explanation of this is that the Critical Period
Hypothesis suggests that there is a time in human development when
the brain is predisposed for success in language learning.
Developmental change in the brain, it is argued, changes the nature
of second language acquisition. According to this view, language
learning which occurs after the end of the critical period may not be
based on the innate structures believed to contribute to first
language or second language acquisition in early childhood. Rather,
older learners depend on more general learning abilities the same
ones they might use to learn other kinds of skills or information
(Lightbown and Spada, 1997, p.42).

However, the development of such general skills is often described,
accounted for and explained in terms of AT (e.g. Bloom, 1986;
Schneider and Fisk, 1983; Anderson, 1995). Many elements of literacy
in a language, too, also can be said to be a general type of
learning, though language development obviously interacts and limits
it.

Compared with many L2 learners in SL situations, FL learners are even
more disadvantaged in terms of an environment to reinforce FL
learning, with both input and output often being limited to formal
classroom settings. L2 (such as ESL) and FL (such as EFL) learners
are often categorized into the same group. However, they represent
two distinct groups, as Swaffer and Bacon (1993) point out, "L2
learners operate in the culture of the language they are learning and
can access input outside the classroom with relative ease, whereas FL
students cannot" (p.125). Thus, it is very difficult to expect and
wrong to assume that incidental learning will automatically occur in
the FL setting. To achieve a target proficiency in FL learning, a
systematic and efficient learning environment should be created
intentionally in the language classrooms. AT can suggest ways for
teachers to achieve such an environment. Therefore, from the combined
perspectives of the learners' age and the learning environment, the
application of automaticity theory is justified in EFL instruction in
environments such as Japan.

Applying AT to EFL Reading Instruction in Japan: An Analysis of the
Situation

Using many of the insights gained from actual application of AT to a
JSL program at the university level in the US (see Yoshimura, 1999),
I would like to address here the theory's possible use for EFL
reading in Japan. Concentrating on EFL reading is particularly
relevant here because written texts are often the major source of
input for students living in a country where the target language is
neither a significant native nor second one. The failure of
developing learners' reading proficiencies in most Japanese EFL
classes can be attributed to the lack of emphasis on the training at
what has been called the associative stage. In this section, what is
involved in EFL reading will be analyzed and skills which require
systematic training toward automaticity will be pointed out. Then,
the following section demonstrates how we can apply automaticity
theory to the actual EFL reading instruction and evaluation in Japan.

According to Bernhardt (1996), in L2 reading both text-driven
operations (e.g. word recognition, phonemic/graphemic decoding, and
syntactic feature recognition) and knowledge-driven operations (e.g.
intratextual perception, metacognition, and prior knowledge) work
simultaneously with varying degrees of success. Bernhardt cites
Johnston's (1983) comments on the risks involved in this simultaneous
operation: "The qualitative mismatch between text and reader may pose
a far more insidious problem -- quite subtly causing the reader to
build a completely inappropriate model of the text meaning without
becoming aware of the problem. It is not that inferences would not be
made, but that inappropriate ones would be made" (p.31). And the
success of creating an appropriate model of text meaning in L2 and FL
reading depends on the accuracy and efficiency of text-driven
operations, which are subskills for L2 and FL reading comprehension.

Researchers agree that reading skills can be automated with repeated
practice (e.g. Schneider and Fisk, 1983; Bloom, 1986; Samuels and
Flor, 1997). However, as Samuels and Flor (1997) warn, not all skills
or knowledge bases can be so automatized. According to them, "In
general, tasks with a high degree of regularity and sameness, such as
word recognition, learning to use a typewriter, or memorizing
multiplication tables, can be automatized, whereas tasks that are
constantly changing, such as text comprehension, continue to require
attention and effort" (p.112).

Therefore, to talk about automaticity in L2 or FL reading, we need to
divide elements involved in the reading act into what can be
automated and what cannot. Knowledge-driven operations such as
intratextual perception, metacognition, and prior knowledge may work
mainly as individual differences in learners' general reading skills
developed in their L1. More importantly, these operations cannot be
automated because they are constantly changing depending on the
context and continue to require attention and effort. However,
automaticity in text-driven skills may well free up memory and
cognition for the type of fluent reading that requires constant
attention and effort, and breakdowns in such skills can prove to be
the "weakest link" in the entire reading process.

What can be automatized are tasks such as word recognition,
phonemic/graphemic decoding, and syntactic feature recognition.
These "tasks with a high degree of regularity and sameness" are the
very tasks in which systematic training should be given to learners.
What task should be central in a particular reading instruction
depends on the reader's proficiency level. Bernhardt (1996)
illustrates the distribution of reading errors which appear along
with second language literacy development (p.169). According to the
illustration, word recognition errors and the phonemic/graphemic
confusions appear in early stages of proficiency. Syntactic errors
have a normal curve shape in account of their development: they
develop as a function of greater exposure and growth in the language
and then decline gradually. Two of the higher order aspects of
reading, background knowledge usage and intratextual perceptions, are
described by exponential curves, indicating that a reader begins to
rely more on the language and less on what he/she thinks the language
contains as his/her proficiency develops. As a reader's proficiency
develops, the central errors and the most important reading factors
change as well. Therefore, the focus of each period of instruction
should be modified depending on the learner's proficiency.

Taking into Account Language Background Differences

On top of the factors which influence general L2 and FL reading,
language-specific factors need to be examined. The following is a
brief comparison and contrast of the linguistic features and
differences across Japanese (L1) and English (FL).

These two unrelated languages differ considerably in various aspects.
While written Japanese mainly uses a combination of hiragana,
katakana (both syllabic symbols), and kanji (symbols which represent
Japanese at the morpheme and word-level of meaning), written English
utilizes a roman alphabet of 26 letters. While the basic Japanese
sentence structure is SOV with postpositions signalling many of the
grammatical relationships, the predominant English word order is SVO
with particles functioning as prepositions. While the Japanese
language largely depends on its post-positional particles to indicate
the parts of speech, English depends much on the word order for the
same function. Because particles are largely responsible for
indicating the relations among words in Japanese, Japanese word order
is relatively more flexible. On the other hand, because English
mainly relies on word order to indicate the intended relations among
words, the overall sentence structures are more regular. However,
mature, written English is characterized by some complexity in
embedding and subordination.

All English sentences can be categorized into the following five
types: SV, SVC, SVO, SVOO, SVOC. There is only one predicate verb (V)
in each sentence except in a compound sentence. If we can find the
predicate verb in each sentence, we can find the subject (S), which
is always placed before the verb, and the object (O) or the
complement (C), which is always placed after the verb. The parts of
speech can be expressed in a word, a phrase, or a clause. Because of
this regularity, internalizing English word order could be automated
with repeated practice. Systematic practice should be given to
Japanese EFL learners, whose L1 utilizes totally different ways of
signaling the intended relations among words. Synthesizing all the
above, systematic training should be given to text-driven operations
such as word recognition, phonemic/graphemic decoding, and syntactic
feature recognition. The selection of the central skills in each
class should be made in light of the learners' reading proficiencies
and the learning contexts, including the linguistic differences
between the L1 and the target language.

Recommendations for Actual EFL Reading Instruction

In this section, a curriculum to develop learners' EFL reading
proficiencies will be recommended with the above analysis in mind. To
avoid inappropriate knowledge-driven operations, text-driven
operations should be practiced repeatedly until students gain
automaticity. At the initial stage, the focus of practice should be
smaller units such as a letter or a word. Learners should be given
enough training to recognize English letters and words. Because the
number of letters in an alphabet is limited, each encounter with
words may give the learners practice in letter recognition. To
enhance word recognition, on the other hand, conscious efforts should
be made on the part of instructors to make this aspect a part of
regular vocabulary instruction. Many L2 learners may complain that
memorizing words are boring and that they will soon forget words even
though they memorize them once. Recycling the same words learned
before in numerous different contexts may reduce their study load and
will also help them recognize words in future reading encounters.

After the initial stage, the most important task may be to help
students internalize and schematize the most typical English sentence
structures and to use them to analyze the intended relations among
words. Though the unit of processing may expand into a paragraph or
the whole text later on, at or below the intermediate level where
most Japanese EFL learners belong, understanding the meanings of each
sentence will remain important. In order to internalize the basic
English structures, instructors may want to provide explicit
explanation of the typical English sentence structures and give
enough training in using them to read English texts. The material
should be short and simple using typical structures. Gradually the
material may become difficult and complicated with the addition of
more grammatical elements and the combination of various structures.

In each stage of proficiency, the learners should be given sufficient
training to reach a stage where they can read the material rapidly
and accurately. The content of the material should be easy and
concrete at first, gradually becoming more difficult and abstract. In
general, reading material that uses typical, mature written English
may contain more complicated sentence structures when compared with
spoken material. Exposing students with graded, spoken materials in
the early stage may give them opportunities to listen to texts with
the basic sentence structures and help them to internalize and
schematize the basic English sentence structures. Though the ultimate
goal of most FL reading practices may be to comprehend the text,
syntactic feature recognition should be emphasized particularly to
the L2 or FL readers whose native language structures are very
different from those of the target language. Otherwise, learners may
depend too much on their background knowledge and so not pay enough
attention to what is actually written in a text.

This over dependence on top-down comprehension--though natural enough-
- may lead them to run the risk of creating inappropriate models of
text meanings. Instructors can ask learners to find the subject and
the predicate verb of each sentence which appear regularly. For
example, instructors can keep asking learners "Who?" and "Did what?"
in reading sentences. Because what follows each verb can be predicted
from the nature of the verb, the questions can be created on the
spot, if necessary. Soon the learners will internalize the questions
and use them to analyze encountered sentences. Even if they encounter
complex sentences, they can keep using the same questions to analyze
the sentence structures and also to find which chunk of words serves
what part of speech. Though this kind of practice may seem boring,
enough emphasis and practice should be given to learners until they
have acquired the sort of automaticity that frees up the working
memory required to read mature prose fluently .

The Value of Evaluation

Evaluation is another factor which plays a crucial role in ensuring
that learners have internalized the important subskills for L2 and FL
reading comprehension. Therefore, evaluation needs to incorporate the
concept of automaticity as well. Most reading tests measure the
learners' comprehension, which is the ultimate goal of most standard
reading practice. The assumption is that if learners can answer the
questions asking their text comprehension, they can use in a
coordinated manner the sets of subskills necessary for text
comprehension. However, this assumption does not always hold true.
Poor L2 or FL learners may construct and continually depend on their
incomplete and possibly erroneous background knowledge and create
inappropriate models of text meanings. Even worse, they may not be
aware of their problems. Evaluation should play a role of diagnosing
the cause of the learners' mistakes and provide feedback to the
learners. This is the reason the usage of these subskills in reading
should be measured separately. The following paragraphs demonstrate
how to measure these subskills.

Word Recognition Automaticity

Samuels and Flor (1997) suggest a way of assessing automatic
performance in word recognition. According to their suggestion,
readers can be asked to perform two tasks. First, they could be asked
to listen to the instructor reading a passage and later they would be
asked to tell everything they remember about the passage. Second,
they could be asked to read a passage out loud and later they would
be asked to tell everything they remember about the passage. For
students who have attained automatic word recognition for reading,
the listening and oral reading scores should be comparable. For
students who are not automatic, the listening score should be better
because the oral reading test demands a level of simultaneous visual
decoding and comprehension not yet attained (pp. 113-114). The proper
way to apply such a method to EFL students still needs to be explored
and worked into a recommended procedure. Students who are automatic
at visual decoding of words can generally auditorally attend to the
oral texts with accuracy, speed, and expression, and do so with good
comprehension. However, there would seem to be possible complications
with direct application to EFL students. For example, the reading
task would have to be strictly timed so as not to make it any easier
than the listening task; in other words the time requirements of the
reading task would have to duplicate the real time limitations of
real listening.

Practicing word recognition is a task that clearly belongs within the
context of direct instruction in the EFL reading classroom, where the
teacher has to focus on both reading and language development and
practice. It is beyond the scope of this paper to cover in details
tasks for practicing word recognition for automaticity, but it should
be pointed out and emphasized that real reading tasks at linguistic
levels appropriate to the learner might ultimately prove the single
best way to develop word recognition skills. However, those
instructors wishing to isolate the practice for some types of limited
practice, there are exercises that can be constructed for this
purpose (see Paran, 1996).

Automaticity in Syntactic Feature Recognition

To measure automatic performance in syntactic feature recognition in
reading, how well learners can apply the internalized English
structures to actual reading processes should be measured "on line"
with real reading tasks. The following is a way of measuring the
online syntactic feature recognition. Ask the learners to draw lines
between meaningful chunks of words and to underline the predicate
verb in each sentence while they read a text. This does not
considerably alter the actual reading processes in use, but it does
help make them more apparent to the instructor. Teachers can then
confirm whether or not the learners draw lines at the appropriate
divisions and if the predicate verbs are the correct ones. According
to automaticity theory, practice "improves the chunking of
information about the outputs, goal states, and inputs of the
situation" (Schneider and Fisk, 1983, p. 122). In addition, teachers
must try to use multiple measures of comprehension accuracy and
reading speed as students engage real texts -- accuracy in
understanding and reading speed are important characteristics of
automatic performance. It is not possible to have reading proficiency
without automaticity; automaticity of certain features -- word and
syntactic recognition -- are necessary (but not sufficient) to
reading for meaning. Thus, in applying automaticity theory into the
actual Japanese EFL reading instruction, much emphasis should be
placed on the repeated practice of text-driven operations and the
online evaluation.

Another technique and activity that would seem to have both
usefulness as a procedure for assessment as well as practice is
actually one that has been around for a while. This is called "read
and look up" (West, 1960, in Bruder and Henderson, 1986). As Bruder
and Henderson (1986) explain it, it is a technique where "the student
looks at a sentence or part of a passage, says it silently, looks up
from the page and says the sentence aloud. Unless the student
understands the grammatical structure and the message of the
sentence, it is impossible to remember the string long enough to
repeat it back. We frequently use the technique at beginning and
intermediate levels to check comprehension and short-term memory (p.
36)." The "read and look up" tasks does not deviate too much from
what normal reading is, and so should prove useful in assessing what
learners can do. While this is not a new technique, it is easy to see
that it fits with many recent assertions about language learning and
processing: students read a phrase, clause or sentence to themselves
(and teachers might prepare a text with the chunks marked for
students learning the procedure) -- a CHUNK of meaning -- and then
look up from the text and say that chunk out loud. If the can do this
with some fluency, then chances are they understand what they are
reading. If they cannot put the chunks into working memory and repeat
out loud what they are holding in immediate memory, then the
language -- the words, the structure -- is too difficult. That is,
even if somewhat known from previous study, there is insufficient
automaticity in recognition of lexical and syntactical features for
fluent reading. This is also an effective way to monitor
comprehension without resorting to cross-linguistic translation.

According to AT, "Practice improves chunking of information about the
outputs, goal states, and inputs of the situation" (Schneider and
Fisk, p. 122). In addition, teachers who do direct reading
instruction need to measure comprehension accuracy along with reading
speed when students read texts, which are the other characteristics
of automatic performance. To measure all these aspects is important
to get an accurate picture of the learners' reading proficiencies.

Conclusion

Automaticity theory (AT) and how to apply it to a FL curriculum have
been introduced. AT is justified as an application to the Japanese
EFL instruction because of the learners' age and the learning
environment. A curriculum to develop the Japanese EFL learners'
reading proficiencies has been recommended using the concepts and
framework of AT. This general theory of learning emphasizes the
importance of training at the associative stage, where learners come
to coordinate individual skills in more varied, complex tasks which
display automaticity and fewer breakdowns in mental processing and
overloads of working memory. How to realize automaticity theory in
actual classroom instruction is still in its tentative, exploratory
stages and more research and insights from reflective practice need
to be accumulated and shared with the rest of the field.