Up from the Bottom or Down from the Top?

I recently updated my book, Teach and Learn: Reflections on Communicative Language Teaching, and made it available on Kindle and as an inexpensive paperback. To enjoy a read, please click here.

By Peter McKenzie-Brown

Psychology long ago began to debate two views – “top-down interpretation” and “bottom-up processing”– of how we understand language. Although the evidence is ambiguous, researchers generally believe they are distinct but complementary processes. For language instructors, the debate is less important than an appreciation of the roles these parallel processes play in classroom teaching and learning.

This discussion argues that good teaching practice accepts both views of language learning. However, their relative importance largely depends on the skills of the language learner.

To appreciate this, let’s begin with points of view. The advocates of “top-down interpretation” argue that background knowledge and previous experience of a situation, context, and topic play primary roles in helping us interpret meaning. We use prior knowledge and experience to anticipate, predict, and infer meaning. By contrast, the advocates of “bottom-up processing” believe language relies more heavily on decoding the sounds and letters of a language into words, clauses, sentences, and such. We then use our knowledge of grammatical, syntactic and lexical rules to interpret meaning. In this view, language users work from the bottom – the sounds they hear and the letters they encounter – to identify meaning.

To put that broad debate into context, consider that the primary focus of communicative language teaching is to develop communicative competence. CL teachers develop this competence through the use of materials and activities that focus on using language functions – for example, describing people and telling time. Because native-speakers use higher mental schema when they are processing language, language teachers develop activities that will enable their second-language learners to do the same. Broadly speaking, activities of these kinds involve top-down learning skills.

Is this always a good thing? No. Some language teachers are too quick to jump on the top-down bandwagon. In our view, better teachers are those who strike a conscious balance between top-down and bottom-up learning, which both have roles in language instruction.

According to Robert Norris, who uses listening activities to illustrate, “If we…require (our) students to use native speaker processing skills without first giving (them) a firm grounding in decoding the stream of sounds they hear, we run the risk of causing (them) more frustration and confusion than they can handle.”

We will return to Norris’s thoughtful discussion shortly. In the meantime, remember that bottom-up processing is particularly important when learners use the receptive language skills of listening and reading, because it plays a big role in making input comprehensible. And comprehensible input is the engine of effective language acquisition.

Bottom-up…: The bottom-up view assumes that listening is a process of decoding sounds and graphemes (the letters of the alphabet). We start with the smallest units, and gradually decode them until we understand the content of what we are listening to or reading.

The number of micro-skills involved is large. For example, when we listen we discriminate among the distinctive sounds of English, recognize stress patterns and the rhythmic structure of English, and discern how we use stress and intonation to signal information. Also, we need to identify words in stressed and unstressed positions and in reduced forms. We also have to recognize grammatical structures and typical word-order patterns. Meaning and comprehension are the last steps in the decoding process.

When we read, we use the building blocks of language to make meaning of what we see on the printed page. Bottom-up processes include sounds and graphemes -- the representation of sounds by letters. In English this involves word recognition for the countless irregular spellings and a sophisticated system of punctuation. We then need to to process written information through grammar and sentences. From these blocks we build comprehension.

…and Top-down: By contrast, top-down proponents believe that language processing involves the reconstruction of meaning through prior knowledge or “schema.”

Listeners actively reconstruct the original meaning of the speaker using incoming sounds and other signals like body language as clues. Prior knowledge of context and situation enables us to make sense of what we hear. A native speaker, for example, may completely zone out while hearing the news, then snatch a few brief cues that quickly draw him in. Similarly, when we begin a phone conversation to make an appointment, we shift into formal speech-patterns for such situations. This is another instance of schema guiding language use.

We also use schema to help us understand what we are reading. For example, the format of letters, emails and magazine ads are similar from culture to culture. Their format, whether in the reader’s first or second language, provides specific and useful information about what we can be likely to expect. Other top-down skills include surveying, skimming, scanning, reading for full comprehension, reading between the lines (inference), and reacting personally to reading texts. Teaching these learning strategies to your intermediate students can greatly improve their reading comprehension.

The Language Level Issue: In his excellent discussion of top-down and bottom-up teaching, Norris argues that the teaching community’s eagerness to focus on top-down teaching is sometimes misguided. “Many of the listening materials on the market today are concerned chiefly with helping learners become more adept at improving top-down skills by having them (identify relevant information while ignoring unnecessary details.)”

He adds, “In order to simulate the knowledge that native speakers bring to listening, learners are often provided with vocabulary lists prior to the task and told who the speakers are, what the situation is, and what the topic is about. However, scant attention is paid to the phonological characteristics that mark informal speech. This seems a bit like putting the cart before the horse.”

Norris makes a strong case that teachers must develop both bottom-up and top-down skills, especially at the lower levels. “Teachers are asking a lot from their students… when top-down listening tasks are given without first assessing the students' ability to do bottom-up processing.” His argument is sound. Learners need many micro-skills learners for bottom-up processing, and a good teacher neglects them at his peril. This applies especially to beginning and early intermediate students.

Wrap-up: The main conclusion of this discussion is that we need to feed both learning processes when we are teaching our students.

How and when can we use top-down processing? When you are teaching, make sure your students are aware of the format and general content of a reading, for example. Tell them they are going to read a ghost story, for example, and then elicit ideas about what the content might be, what vocabulary might occur, and so on. This switches on the ghost-story schema in their brains, and also begins activating their English skills. You can do the same with listening. Tell them you are about to listen to a sports broadcast on the Football World Cup. Elicit information about football and the vocabulary they might expect to hear, and so on. Also, of course, a CL teacher is constantly using authentic activities to teach. Thus, a role-play “in the restaurant” is by its very nature a top-down comprehension activity.

In these and many other ways, you can take advantage of your students’ ability to use top-down comprehension to get them ready for the upcoming learning activity. As your students advance, you can use more sophisticated top-down schema and strategies.

Bottom-up skills are different. As we have suggested throughout, they are usually more basic and therefore more important for lower-level students. With those students, you need to spend time helping them recognize reduced speech, for example, and irregular spellings. In the early stages of language acquisition, automaticity in word recognition is critical.

Communicative language teaching emulates real-life language acquisition, which means our work has a top-down bias. Your class needs to use authentic activities and materials to function effectively, and those materials tend to be top-down. However, focusing exclusively on top-down teaching creates problems. Especially with beginning students, spend time developing bottom-up skills.

Noam Chomsky's Black Box

I recently updated my book, Teach and Learn: Reflections on Communicative Language Teaching, and made it available on Kindle and as an inexpensive paperback. To enjoy a read, please click here.



By Peter McKenzie-Brown

What is language, and where does it come from? The answer to this question has great bearing on how we teach language.

Before the coming of Communicative Language Teaching, most thinkers believed that behaviourism – a powerful approach to psychology – explained language. In effect, according to this view, language is a set of habits we develop through mimicry of our parents and those around us.

We learn language because it brings us a reward of some kind. The basic model these thinkers – led by B.F. Skinner, a renowned Harvard psychologist – used to explain psychological phenomena is a simple one. If an organism responds in a certain way to a stimulus and gets positive reinforcement, its behaviour will be repeated. If it gets no reinforcement or negative reinforcement, its behaviour will stop. Thus, any kind of vocal output can become part of language, given sufficient reinforcement.

This model of language learning had a powerful impact on the teaching method known as audiolingualism. This method was based on the belief that language learning is basically a process of mechanical habit formation. It placed a strong focus on drills, on teaching through speech at the expense of the other skills, on inductive teaching (repeating a structure many times, then deciding for yourself how the grammar works) and on learning vocabulary within a linguistic and cultural context. This approach is still widely used in East Asia.

In effect, behaviourism denied the concept of human nature. It taught that people can be made to become almost anything, given the right sequence of stimuli and responses. On the matter of language, the logical conclusion was that there were no limitations to what language could be. Any linguistic habit that people somehow developed could become part of language. In academic circles these notions were widely accepted until Noam Chomsky turned them upside down.

The Language Instinct: In a famous phrase, Chomsky said he wanted to "look deeper" into some basic questions about human language. His early theoretical books and papers gave his answer to an extremely profound question: Is language infinitely changeable (the basis of audio-lingualism), or are all languages in some way a part of being human? In the following passage, one of Chomsky’s best-known admirers summarizes the thinker’s basic conclusions:

Language is a human instinct. All societies have complex language, and everywhere the languages use the same kinds of grammatical machinery like nouns, verbs, auxiliaries, and agreement. All normal children develop language without conscious effort or formal lessons, and by the age of three they speak in fluent grammatical sentences, outperforming the most sophisticated computers. Brain damage or congenital conditions can make a person a linguistic savant while severely retarded, or unable to speak normally despite high intelligence. All this has led many scientists, beginning with the linguist Noam Chomsky in the late 1950s, to conclude that there are specialized circuits in the human brain, and perhaps specialized genes, that create the gift of articulate speech.

Chomsky’s ideas explained the facts in a way that no other theory can. Acquiring language is not a normal mental problem. Everyone sees small children pick up language effortlessly. But few of us ever give a thought to what an amazing phenomenon this is. No one would expect a three-year-old to master calculus. Yet we are not in the least surprised when a three-year-old learns grammar, which is a much more difficult task. Children learn the rules of their native grammar (whatever it might be) by hearing a limited set of sample sentences. According to Chomsky's followers, the limited information they receive is mathematically insufficient for them to determine grammatical principles, yet somehow they are still able to do so. This is what is sometimes called “The paradox of language acquisition”. As Chomsky explained,

In a given linguistic community, children with very different grammar arrive at comparable grammars, indeed almost identical ones, as far as we know. That is what requires explanation. Each child has a different experience, each child is confronted by different data – but in the end the experience is essentially the same. As a consequence, we have to suppose that all children share the same internal constraints which characterize narrowly the grammar they are going to construct.

Chomsky proposed that each of us has a Language Acquisition Device (LAD) – what he sometimes called a “little black box” – that starts functioning when we are still infants. By the time we are five or six, that device has enabled us to vacuum from our immediate environment a native language based on universal grammar. Put another way, all languages are fundamentally the same, irrespective of the cultures we live in. They are the function of a large number of words (arbitrary symbols whose meaning is set by convention) and a limited number of grammatical rules that are somehow structured into our brains and minds.

Those who learn during the “critical period” – roughly, during early childhood – acquire language without effort. However, our little black boxes begin to function in different ways after we reach six years of age. We remain able to acquire language fairly easily until we reach puberty, after which our ability to acquire language seems to be reduced again.

Older learners must therefore learn languages in quite different ways from the effortless acquisition of young children. The ways in which language acquisition changes form are of vital importance to language teachers.

A helpful book in this area of study is How Languages Are Learned by academics Patsy M. Lightbown and Nina Spada. The two educational psychologists offer useful insights into the differences in learning styles. They say, “We have all heard people say that they cannot learn something until they have seen it. Such learners would fall into the group called ‘visual learners.’ Other people, who may be called ‘aural’ learners, seem to need only to hear something once or twice before they know it. For others, who are referred to as ‘kinaesthetic’ learners, there is a need to add a physical action to the learning process.”

These ideas are common in teaching circles. However, Lightbown and Spada go on to discuss two learning styles that are highly cultural in nature. “In contrast to these perceptually based learning styles, [there is] a cognitive learning style distinction between field independent and field dependent learners. This refers to whether an individual tends to separate details from the general background or to see things more holistically.” In general, westerners, who tend to see things as black or white, are field independent. By contrast, East Asians are field dependent inasmuch as they see things as part of a large mingling of relationships.

In their challenging text, Lightbown and Spada describe three theories about how we acquire our first language. The most important of those theories, innatism and behaviourism are covered at the beginning of this essay. A third approach is called interactionism.

The interactionist position is that “language develops as a result of the complex interplay between the uniquely human characteristics of the child and the environment in which the child develops.” A kind of umbrella that covers several different kinds of language theories, contemporary interactionism assumes the validity of Chomsky’s idea that innate structures in the mind are responsible for language learning, but has a variety of questions about the contributions of environmental factors to language learning.

In a summary of these questions, the authors propose that behaviourism may help explain our acquisition of vocabulary and other parts of grammar. For its part, innatism seems “most plausible in explaining the acquisition of complex grammar.” And “interactionist explanations may be useful for understanding how children relate form and meaning in language, how they interact in conversations, and how they learn to use language appropriately.”