What is meant by automaticity and automatization?

automatic

• largely or wholly involuntary, especially as with a reflex
• acting or done spontaneously or unconsciously

automatize

• to make automatic [noun derivations: automatization the process of making automatic; automaticity the state or condition of being automatic]

In some literature, the word automization is used rather than automatization.

For our purposes as teachers, automatization is the process of making language processing and production automatic. The skill of automatization is the ability of the language learner to get things right [with both spoken and written production] under circumstances when there is no attention available or any other provision for getting those things right.

Automization is a central concept within the academic study of cognitive skills acquisition: Shiffrin and Dumais [1981], for example, identify automatization as a fundamental component of skill development. As EFL teachers in the Arab world, we should be clear exactly why it is important, not only to the development of skills in general but to the development of language skills in particular.

Let us start by examining the process of automatization and the role of automaticity in skill development.

Whenever a skill is newly learnt, its performance takes up a great deal of conscious attention [or channel capacity]. For example, learner drivers who have recently been taught to change gear with a manual gearbox [floor shift] will at first probably only be able to do so if they concentrate on that task and nothing else. This state of affairs is clearly unsatisfactory, and it may well be dangerous if the novice is actually driving, because there are, simultaneously, higher-level driving skills in operation that require the available channel capacity [that is the driver's conscious attention]. All drivers need to anticipate the movement of other vehicles and pedestrians, and they certainly need to be paying attention to what is going on around them. Channel capacity can only be made available for these higher-level skills if lower-level skills [such as changing gear] have been automatized to the extent that they take up no room in the driver’s conscious mind. When novice drivers achieve automatized gear changing, they will be able to perform the action without ever being aware that they are doing it. So the role of automatization in motor skill learning is to free important channel capacity for tasks that require it.

The role of automatization in language learning is similar. Learners automatize the use of lexical items and sentence patterns and structures [especially lexicalised grammar] so that they can then concentrate their effort where it is needed most, in comprehending and communicating meaning clearly. This is so for all areas of language use, but it is probably more important in some areas than in others [Bialystok, 1982]. This leads us to an important insight: the automatization of predictable items of language is highly likely to precede the acquisition of more generative functions. In terms of English Language Teaching, this automaticity is rooted in the handling of probable language rather than possible language, in lexical production and in the production of lexicalised grammar. Bialystok suggests that different tasks in fact place different demands in terms of the degree of automaticity [or fluency] required. Thus, in fluent and interactive conversation, the user certainly needs high automaticity with lexis. In such conversational interactions, great store is placed on rapid response and turn taking. However, in formal written production, where users normally have sufficient time to think and plan and edit text, automatized language performance will be less important.

The value of automatization is that it frees conscious attention for higher-level skills. Skilled language performers automatize their production of language. They automatize their responses in interactive situations. This frees conscious attention for higher-level activities. This is particularly true of the lower level or mechanical aspects of a task [and of lower order language skills], which require less conscious attention in order to perform. One result of automatization is that skilled performers are less conspicuously aware of performing activities that they have automatized. Just observe an experienced teacher of English in front of a class if you want to see this phenomenon in action.

Automatization is an important concept for teachers of English to grasp and respond to. As EFL teachers, we are familiar with learners of English who may be able to produce [for example] present perfect tense structures accurately in a classroom drill, in which there is a clear focus on form, but may not be able to reproduce or transfer these structures subsequently. The skill of automatization ensures that learners perform the target structure accurately in communicative situations in which there is a focus on message and where there may be distracting circumstances. The skill of automatization is typically what those learners we identify as false beginners do not possess to any useful degree.

Reading is an extremely important skill required in our vocational training program. As teachers, we seek to produce skilled readers who are ultimately capable of reading and understanding complex technical vendor manuals or comparable texts. We may need to re-examine our approach to the teaching of reading [and to other aspects of our language teaching program] in the light of recent research into automaticity. Urquart and Weir [1998] cite both Carrell et al [1988] and Stanovich [1981] as identifying automaticity as one of the hallmarks of a skilled reader, and they argue that some of the time devoted in class to working out the meaning of words in context might be better spent on activities promoting automaticity [Page 188]. They point out that one of the most significant contributions by cognitive psychologists to reading research and therefore to the teaching of reading is the finding that good readers use context much less than poor readers when recognizing printed words. Good readers are able to recognize words without conscious thought. In other words, they have automatized their lexis.

Urquart and Weir [1998] observe that increased importance has been attributed to automatic word recognition in second language reading. They point out that previously a great deal of faith had been placed on decoding by means of context. They cite Haynes [1984] as pointing out that teachers of reading needed to improve the level of automatized vocabulary rather than focusing on decoding in context. Haynes suggests that bottom-up, input constrained processing, particularly the rapid and precise recognition of letters and words, must be mastered before fluent reading can take place. She found evidence from first language studies that reading fluency is achieved by increasing the bottom up processing of text and decreasing semantic and syntactic guesswork. Haynes questions the use of guessing the meaning[s] of unknown words from context as the primary approach to learning vocabulary, and points out that context is often inadequate to support accurate inferencing. Also cited in Urquart and Weir, Bensoussan and Laufer [1984] argue that, in many instances, only a minority of word meanings can be recovered from context.

Beck [1981] argues that the use of basic recognition exercises to improve reading speed and accuracy should form an important component of an effective second language reading program, because this will promote the development of automaticity.

Urqart and Weir [1998] suggest that if automatic word recognition is more important to fluent processing of text than context clues, the large-scale development of recognition vocabulary may be crucial to reading development. They suggest that poor readers have not acquired automatic decoding skills and so spend excessive processing time thinking about words and context, rather than automatically recognizing them and then accessing the meaning of the text.

Bernhardt [1991] argues that automaticity is the ultimate goal, and that it is only automatic processing that allows a good reader to think about the meaning of text, to recover the message and to relate to new information in the text. It is only this kind of local processing that engenders global meaning and true comprehension. Bernhardt further argues that a major bottom up skill is reading as fast in the target language as knowledge allows, in relation to reading purpose. This suggests that, where appropriate, teachers need to dedicate some time to the rapid, automatized recognition of lexical and [lexicalised] grammatical form.

It is widely recognized that ESL and EFL learners generally read texts more slowly than first language English readers. This is demonstrably so for the vast majority of our readers in our industrial and vocational training programs. Haynes [1984] suggests that it is the length of fixations that slows down reading rather than merely the number of fixations and regressions, and suggests that second language readers take longer to access meaning or to remember what a word means. This is because second language learners of English by their very nature have neither large and well-practised vocabularies nor the years of experience of recognizing words in print that first language learners have. Consequently, it takes such readers longer to decide whether a given word is known or unknown and, in the latter case, crucially, whether to ignore it or not. Such readers have not sufficiently automatized their lexicon.

Urquart and Weir [1998] point out that although the importance of automaticity in decoding is gaining general recognition in ELT, there remains less agreement on how to achieve the goal of automaticity in second language reading [Page 193]. Haynes [1984] argues that the importance of word unit processing should be recognized, and advocates dictionary work as a way to separate words that appear similar and to help readers distinguish new words efficiently in lexical memory. Nuttall [1996] warns that the use [or over-use] of dictionaries may slow reading speed and reduce effectiveness in reading.

Juel [1991] makes the point that automaticity in most skills derives from over-learning through repeated practice and drilling. This may have implications for the EFL classroom, especially in the provision of frequent and regular practice. As Thornton [1984 onwards] points out, with beginners of English particularly in mind, what learners need is a little a lot, not a lot a little. He uses the magic word: ALALNALAL.

Paran [1996] suggests the need for a greater focus on bottom-up processing. Paran argues that good readers do not rely on hypothesis formation and prediction as much as is commonly thought and points out that visual input and bottom-up processing during reading are of great importance [Page 25]. He observes that the top-down approach promoted by Goodman [1967] still dominates ELT reading materials despite the identification of more comprehensive and interactive models of reading. Paran goes on to argue that contextual guessing is not an appropriate strategy for the lower levels of language processing, that is for word or phrase recognition.

Bialystok [variously 1982, 1988 and 1990] distinguishes between different areas of language use, and particularly between declarative and procedural knowledge. It is important to understand how declarative and procedural knowledge relate to the concept of automatization. Automatization is seen as the process by which declarative knowledge is converted into procedural knowledge, giving the advantages of the procedural knowledge and eliminating the disadvantages of the declarative. Given this central role of the process of automatization, we would expect a model of skill learning to account for how it occurs.

What is the evidence for automatization?

If automatization is the process of making something procedural, then research that attempts to find evidence of automatization at work will be research that shows how declarative knowledge becomes procedural. Such research will be comparative, examining different groups of individuals doing the same tasks at the same time, or the same group of individuals doing different tasks, or doing the same tasks at different times, and identifying differences in performance that can be accounted for by the process of automatization.

Towell [1987] is representative of a large amount of psycholinguistic research aimed at identifying aspects of information processing at work. His research was based on a large corpus supplied by five undergraduate students of French and collected over a four-year period. The research focused on specific syntactic constructions, for example, one of them being the choice of preposition a, de or [zero] following the adjective difficile. Towell noted a great deal of persistent variability over the entire period of the study. In order to account for this variability, he utilized the concept of control [see Hulstijn and Hulstijn, 1990]. He postulated a diachronic dimension to control, suggesting that the learner’s mastery of the factors in controlling language improved over time. In other words, automatization took place.

Towell used four measures of language performance in his research:

• The speaking rate was calculated by dividing the total number of syllables produced by the total time taken to produce the target utterance [including pause time] and multiplying the result by 60 to give an expression of syllables per minute.
• The articulation rate was arrived at by dividing the number of syllables produced by the number of seconds taken to produce them [not including pause time] to give an expression of syllables per second.
• The pause to time ratio gave an expression of the percentage of time spent pausing.
• The length of runs was a straightforward measure of the uninterrupted stretches of speech.

The results for one of the subjects were representative of the whole sample. The subject’s speaking rate increased by 65%, her articulation rate increased by 20%, her pause to time ratio increased by 37%, and her length of runs between pauses increased by 95%. In addition, the number of utterances between one and four syllables fell by 30.4%, the number of utterances between five and ten syllables increased by 37.9%, and the number of utterances of eleven or more syllables increased by 19.91% [Towell, 1987, Page 125]. It is clear that this student over time improved her performance on the control dimension considerably. This is evidence of the existence of automatization.

McLaughlin [1987 & 1990] presents a cognitive theory in which the two central principles are automatization and restructuring. McLaughlin [1987] also contains a useful summary of recent information-processing research, listing it under three headings: lexical retrieval; syntactic processing; and reading. Here are six examples of information-processing research in which some form of automatization was involved [selected and summarized from McLaughlin, 1987].

• Henning [1979] studied errors in lexical processing and suggested that more advanced learners group lexical items using semantic criteria, unlike less advanced learners who seem to cluster acoustically, not having fully automatized target lexical items.
• Dornic [1979] looked at bilinguals and found that, even with balanced bilinguals, encoding was slower in a learner’s second language than in the first language. The explanatory factor was thought to be the extent of automatization. Dornic also found that external factors such as noise interfered with processing significantly more in the second language than in the first, a finding that, together with research such as that by Hulstijn and Hulstijn, [1990], suggests that task complexity is a viable concept in language teaching.
• Hatch et al [1970] asked native and non-native speakers to cross out certain letters in a text [for example, all the instances of letter e]. The researchers found that the native speakers ignore the target letters more often in function words than in content words, but that this is not the case with the non-native speakers. This suggests that native speakers do not focus so much attention on function words, having automatized their use, and concentrate channel capacity on content words.
• Rossman [1981] compared the memory of semantic and syntactic features of a text with native and non-native speakers. With changes in semantic features, native speakers display greater recognition than non-native speakers. With changes in syntactic features, non-native speakers display greater recognition than native speakers. Rossman hypothesizes that this is because native speakers automatized syntactic features and were able to concentrate on semantic features, unlike the non-native speakers.
• Wolfe [1981] studied native speakers of English learning French as a second language. The task was to read a set paragraph and then identify given sentences as being the same or different from those in the text. The similarity or difference was to do with either language or content. The more proficient L2 speakers identified a greater number of different sentences where the difference was semantic. The less proficient L2 speakers had higher scores when identifying syntactic differences. This suggested a lower degree of automaticity for the less proficient speakers, together with a greater concentration [or preoccupation] with form as opposed to content.
• Segalowitz [1986] found that fluent bilinguals demonstrated lower automaticity in word recognition and slower second-language reading in comparison to first language performance.

Much of the work in this area of cognitive psychology [information-processing concepts] was originally just concerned with the identification of the types of human processing and with the development of models to account for them. Then, as Langley and Simon [1981] note, at the beginning of the 1980s, attention began to switch to attempts to develop learning theories that took account of these issues. One such model was developed by the cognitive scientist Anderson [et al]. The Anderson Model exists in several slightly different versions, but an accessible description of the model is found in Anderson [1983]. It is, in fact, largely based on earlier work by Fitts [1964] and by Fitts and Posner [1967].

The Anderson Model was intended to provide a model of learning. It was developed to explain skills [such as those in geometry] where specific instruction [declarative knowledge] is given. It is explicitly acknowledged [Anderson, 1982] that information is provided through instruction at the declarative stage. We learn declaratively first and then automatize over time. The strength of the model is that it provides a specification of how automatization takes place through the subsequent processes of composition and proceduralization. Composition speeds up action as a combination of two productions takes less time to perform than the same two productions performed sequentially [Neves and Anderson, 1981]. Proceduralization diminishes the required memory space as the relevant parts of the database are incorporated into the production. The production encoding is then tuned. This sequence provided by the model is a learning sequence: in the sequence, declarative encoding leads to procedural encoding, which in turn leads to tuning [by generalization, discrimination, and strengthening].

In the Anderson Model, it is the process of proceduralization that leads to automatization. It follows that encodings entering the system in proceduralized form will of necessity rapidly become highly automatized. Consequently, and perhaps especially for initial and relatively basic learners, these encodings should occur most frequently in tasks that entail high automaticity. For relatively low level English language learners in a language impoverished environment, it is likely that there would be greater acquired output in fluent conversation than in writing and particularly with lexical items and items of lexicalised grammar.

A word of caution is needed here. Acquired encodings, which enter the system in an already proceduralized form, rapidly become highly automatized and impermeable to change. It is [observably] extremely difficult to change knowledge once it has been proceduralized. This may be desirable in some respects but it also makes such acquired proceduralized encodings a risk. If a production is in some way faulty, a great deal of effort will be needed to remediate or eradicate it. Perhaps one way to view the phenomenon of false beginners and the related fossilization of interlanguage is to see them as the result of [over] rapid automatization of faulty productions.

A further word of caution is also needed. Proceduralized encodings are by their nature inflexible and non-generative, because the relevant knowledge is contained in the production itself. This knowledge is not part of a database and is not available for other encodings. If language acquisition were characterized by directly proceduralized encodings, then acquired knowledge about language would be inflexible and non-generative too. As teachers of English, we are well aware of the place of a flexible and generative function of language, broadly the area of much grammar and some phonology. Nevertheless, the acquisition of automatized language remains the essential foundation upon which flexible and generative language can develop.

A further problem with the Anderson Model is that it attempts to impose a sequence from declarative to procedural on all language acquisition. It may well be that this formulation is too rigid as a model for the acquisition and mastery of language skills. The mastery of skills in general may at times involve the direct proceduralization of knowledge, without going through a declarative stage. For example, in learning to play a piano, there are initial procedural representations when the novice pianist acquires fixed configurations of notes as a first stage [Karmiloff-Smith]. Just think of playing ‘Chopsticks’ on a piano. Novices learning to play a musical instrument will certainly find that they can play some notes or sequences of notes only when their full attention is on these. When those notes or sequences occur within a piece of music, where attention is required for many other things at the same time, mistakes may be made. The novice player has the ability to get things right under circumstances when there is attention or available for getting them right but not to get things right under circumstances when there is no attention available. A similar point can be made in respect of learning to play chess. People learn how to play chess by practice and are very often unaided by lessons in the theory. The same is true for some second language production. The writer fondly remembers one young girl in Africa who ritually exchanged greetings in English with him every time she saw him, quite purposefully as a social gesture but without any background of formal ELT instruction whatsoever.

Bibliography

Anderson J R [1982] Acquisition of Cognitive Skill [Psychological Review 89.4]

Anderson J R [1983] The Architecture of Cognition [Harvard University Press]

Beck I [1981] Reading Problems and Instructional Practices [In Mackinnon G E & Waller T G [Eds] Reading Research: Advances in Theory and Practice Academic Press]

Bensoussan M and Laufer B [1984] Lexical Guessing in Context in EFL Reading Comprehension [Journal of Research in Reading Vol. 7] [Cited in Urquart & Weir, 1998]

Bernhardt E B [1991] Reading Development in Second Language: Theoretical, Empirical and Classroom Perspectives [Ablex Publishing Corporation]

Bialystok E [1982] On the Relationship Between Knowing and Using Linguistic Forms [Applied Linguistics 3.3]

Bialystok E [1988] Psycholinguistic Dimensions of Second Language Proficiency [In Rutherford W & Smith M S Grammar and Second Language Teaching Newbury House]

Bialystok E [1990] The Competence of Processing [Paper at TESOL Convention 1990]

Carrell P L, Devine J & Eskey D E [Eds] [1988] Interactive Approaches to Second Language Reading. [Cambridge University Press] [Cited in Urquart & Weir, 1998]

Dornic S [1979] Information Processing in Bilinguals: Some Selected Issues [Psychological Research 40]

Fitts P M [1964] Perceptual Motor Skill Learning [In Melton A W Categories of Human Learning Academic Press]

Fitts P M & Posner M I [1967] Human Performance [Brooks Cole]

Goodman K S [1967] Reading: a Psycholinguistic Guessing Game. [Journal of the Reading Specialist Vol. 6]

Hatch E , Polin P & Part S [1970] Acoustic Scanning or Syntactic Processing [Paper at WPA San Francisco]

Haynes M [1984] Patterns and Perils of Guessing in Second Language Reading [In Handscombe J et al On TESOL 83: the Question of Control TESOL] [Cited in Urquart & Weir, 1998]

Henning G H [1979] Remembering Foreign Language Vocabulary: Acoustic and Semantic Parameters [Language Learning 23.1]

Hulstijn J H and Hulstijn W [1984] Grammatical Errors as a Function of Processing Constraints and Explicit Knowledge [Language Learning 34.1]

Johnson K [1996] Language Teaching and Skill Learning [Blackwell]

Juel C [1991] Beginning Reading. [In Barr R et al Handbook of Reading Research Longman]

Karmiloff-Smith A [1992] Beyond Modularity [MIT Press]

Langley P and Simon H A [1981] The Central Role of Learning in Cognition [In Anderson J R Cognitive Skills and their Acquisition Lawrence Erlbaum]

McLaughlin B [1987] Theories of Second Language Learning [Edward Arnold]

McLaughlin B [1990] Restructuring [Applied Linguistics 11.2]

Neves D M & Anderson J R [1981] Knowledge Compilation: Mechanisms for the Automatization of Cognitive Skills [In Anderson J R Cognitive Skills and their Acquisition Lawrence Erlbaum]

Nuttall C [1996] Teaching Reading Skills in a Foreign Language [Heinemann Educational]

Paran A [1996] Reading in EFL: Facts and Fictions [ELT Journal Vol. 50.2]

Rossman T [1981] The Nature of Linguistic Processing in Reading a Second Language [PhD thesis: Concordia University cited in Johnson K 1996 Language Teaching and Skill Learning Blackwell]

Segalowitz N [1986] Skilled Reading in the Second Language [In Vaid J Language Processing in Bilingual Psycholinguistics and Neuropsychological Perspectives Lawrence Erlbaum]

Shiffrin R M and Dumais S T [1981] The Development of Automatism [In Anderson J R Cognitive Skills and their Acquisition Lawrence Erlbaum]

Stanovich K E [1981] Attentional and Automatic Context Effects in Reading. [In Lesgold A & Perfetti C Interactive Processes in Reading Lawrence Erlbaum] [Cited in Urquart & Weir, 1998]

Towell R [1987] Variability and Progress in the Language Development of Advanced Learners of a Foreign Language [In Ellis R Second Language Acquisition in Context Prentice Hall]

Urquart A H & Weir C J [1998] Reading in a Second Language: Process, Product and Practice. [Longman]

Wolfe S [1981] Bilingualism: One or Two Conceptual Systems? [PhD thesis: San Francisco State University] cited in Johnson K [1996] Language Teaching and Skill Learning [Blackwell]

Related posts

• The characteristics of spoken language (0)
• Teaching speaking: fluency or accuracy? (0)
• Speaking activity outcomes - a rough guide (1)
• Some more speaking activities (0)
• You Say and I Listen (0)