The Case for Non-Targeted Comprehensible Input: The Net Hypothesis
Stephen Krashen
In nearly all foreign and second language classes, there is a “rule of the day”
as well as vocabulary that students are expected to focus on, often referred to
as “target” grammar and vocabulary. In traditional pedagogy, exercises are
aimed at the conscious learning of this targeted grammar and vocabulary.
They are also included in brief readings, which are generally packed with the
targeted items.
Targeted grammar and vocabulary is also present in TPRS, and in
“modified” Natural Approach, as manifested in the Dos Mundos textbooks,
although the goal in these cases is the subconscious acquisition of the target
items. TPRS provides longer, more interesting reading selections and
discussions, but typically utilizes a grammatical syllabus.
I present here the disadvantages of the grammatical syllabus and targeted
input in general, and then argue that comprehensible input effortlessly deals
with grammatical syllabus’ shortcomings.
Problems with the grammatical syllabus
The natural order problem. As is well-known, studies have shown that we
acquire the grammar of a language in a predictable order, and this order
cannot be broken. For an item of grammar to be acquired, the language
acquirer must be ready to acquire the item. It must, in other words, be at the
acquirers’ i+1, where i = aspects of grammar that were most recently
acquired.
Targeted grammar and vocabulary is also present in TPRS, and in
“modified” Natural Approach, as manifested in the Dos Mundos textbooks,
although the goal in these cases is the subconscious acquisition of the target
items. TPRS provides longer, more interesting reading selections and
discussions, but typically utilizes a grammatical syllabus.
I present here the disadvantages of the grammatical syllabus and targeted
input in general, and then argue that comprehensible input effortlessly deals
with grammatical syllabus’ shortcomings.
Constraint on interest. The goal of the language classroom is to provide
input that it genuinely interesting, so interesting that students, in a sense,
“forget” that it is in another language, or “compelling” (Krashen, in press).
The Compelling Input hypothesis maintains that language acquisition
proceeds best when all attention focused on the message to such an extent
that thoughts of anxiety and focus on form do not occur.
The Compelling Input Hypothesis is influenced by the concept of “flow,’
(Csikszentmihalyi, 1993). Flow is the state people reach when they are
deeply but effortlessly involved in an activity. In flow, the concerns of
everyday life and even the sense of self disappear – our sense of time is
altered and nothing but the activity itself seems to matter. “Forgetting” and
flow occur in reading when readers are “lost in a book,” when they are
aware only of the story or the message in the text. It is when this happens
that language acquisition occurs most effectively. Note that this position is
the opposite of the “focus on form” or “focus on forms” points of view.
It is very hard to create compelling messages when the hidden agenda is the
relative clause. In fact, it is hard enough to do it this when there are no
constraints on what vocabulary and grammar can be used.
The review problem.
Traditional second and foreign language methods
work through what is considered to be the basic grammar of a language the
first year. Once a grammar rule is presented and practiced, it may not be
seen again until the second year when we review the entire grammar again,
because students did not master it the first year.
The unteachable and untaught grammar problem. The grammar
presented in class is nowhere near the complete grammar of the language.
Even the most accomplished linguists concede that they have only described
fragments of languages. Moreover, language textbooks do not contain all
that linguists have described, and teachers rarely teach everything in the
texts.
Denial of i+l. The impoverished input provided by the grammatical
syllabus will result in students not getting input in structures they actually
are ready for. Grammatical syllabi typically place easily describable items
early in the sequence and more complex ones later, but the natural order of
acquisition runs on different principles. Some rules that look easy to the
linguist and teacher (e.g. the third person singular in English) are acquired
late, while others that look complex are typically acquired early. The earlyacquired
items must be in the input for their acquisition to take place.
Individual variation. There is individual variation in the rate of
acquisition, because of input factors (some students may have had additional
input in the language outside of class) and affective factors. Even if the rule
of the day happens to be at i+1 for some students, it will not be for other
members of the class.
Non-Targeted Comprehensible Input: The Net Hypothesis
An important corollary of the Comprehension Hypothesis is the “Net”
Hypothesis: Given enough comprehensible input, all the vocabulary and
structures the student is ready for is automatically provided. In other words,
“i+1″is automatically there. In Krashen and Terrell (1983) this was referred
to as the Net: “When someone talks to you in a language you have not yet
completely acquired so that you understand what is said, the speaker ‘casts a
net’ of structure around your current level of competence, your ‘i’. This net
will include many instances of i+1, aspects of language you are ready to
acquire” (p. 33).
The same, of course, goes for reading: If you understand the text, and you
read enough of it, you will get i+1.
Before looking at the evidence, let us for the moment assume that the Net
Hypothesis is correct and see how non-targeted comprehensible input
completely solves the problems of the grammatical syllabus.
The natural order problem: Non-targeted comprehensible input,
according to the Net Hypothesis, contains the aspects of language the
acquirer is ready for. This means we do not need to know the natural order.
Rather, grammatical competence will emerge in a natural order as a result of
getting non-targeted comprehensible input.
Constraint on interest: With non-targeted comprehensible input there are
no target structures and target vocabulary that must be used in creating
activities and stories. Anything goes, as long as the input is comprehensible
and interesting (or compelling).
The problem of providing comprehensible and interesting input is the
fundamental problem of beginning language teaching. It is easy to get input
that is interesting but not comprehensible, from the real world.
Unfortunately school tends to provide input that is comprehensible, but not
interesting. It is hard to get both, to say interesting things using limited
language, even if one is not required to use specific vocabulary and
grammar.
Denial of i+1: Non-targeted comprehensible input, according to the Net
Hypothesis, solves this problem easily: i+1 is always there, if there is
enough input.
The review problem: Non-targeted comprehensible input provides natural
review, especially if there is some topic continuity in the progression of
activities and reading.
The unteachable/untaught grammar problem: This is no problem for
non-targeted comprehensible input. “Unteachable rules” are only a problem
when the goal is conscious learning. Second language acquirers have always
been able to acquire rules that have not been taught and that cannot be
taught.
Individual variation: If the input is comprehensible for all members of the
class, everyone is getting what they need, even if i+1 is different for every
member of the class. See discussion of “picking out” i+1 below.
The evidence
The evidence supporting the Net Hypothesis comes originally from first
language acquisition. Caretaker speech to children is typically
comprehensible, but is not “finely tuned” to the child’s current linguistic
level. As the child develops linguistically, caretaker speech tends to get more
complex, but the relationship is not exact: The caretaker does not supply
precisely the next rule the child is ready for.
Evidence includes studies showing that the correlations between input
complexity and the child’s competence are usually positive, but are not
extremely high. Cross (1977) concluded that “… the syntax of mothers, even
to rapidly developing children, is not uniformly pitched just a step ahead of
the child in either linguistic or psycholinguistic complexity. Some utterances
are pitched at the child’s level, some even below this, and others are
considerably in advance of what the child themselves can say” (p. 180).
No studies of input to second language acquirers have examined input to this
level of detail, but we do know that teacher talk is roughly-tuned to the level
of students, not finely-tuned (Krashen, 1981). We also know that second
language acquirers improve from communicating with native speakers and
from reading authentic reading material (Krashen, 1981, 2004), input that is
certainly not finely tuned to the acquirer’s i+1.
Picking out i+1
There is, in addition, evidence that children are able to pick out the aspects
of the input that are relevant to their stage of development, that is, they can
pick out and make use of what is at their i+1.
First language researchers (Gleitman, Newport and Gleitman, 1984) studied
the relationship between the frequency of yes/no questions in caretaker input
and the development of the verb phrase auxiliary. A relationship was
suspected because in yes/no questions the verb phrase auxiliary in English is
often placed at the beginning of a clause and is often stressed, which makes
it very prominent (e.g. Is John playing the violin? Does Mary have a kite?).
They found that the frequency of yes/no questions was indeed very strongly
related to verb phrase auxiliary development for the older children in their
sample (23.9 to 24.8) months (r = .91) but was not significantly related to
verb phrase auxiliary development for the younger children (18.5 to 12.3)
months.
The two groups received similar input; for the older children, however, this
structure was at their i+1. For the younger group, it was beyond their i+1.
This did not, apparently, impair the younger children’s comprehension. This
suggests that the best input for acquisition is input that contains maximum
richness but remains comprehensible. Such data will contain, inevitably,
some i+n (input beyond i+1), as caretaker speech always does, in the form of
later-acquired aspects of grammar. Including this “noise” does not impair
communication, nor would deleting it make the input more comprehensible.
Rich input, as long as it is comprehensible, provides the acquirer with a
better sample to work with, more opportunities to hear and read structures he
or she is ready to acquire.
Roger Brown summarizes this point of view succinctly. After reviewing
research on how caretakers talk to children, Brown offered this advice in
answer to the question, “How can a concerned mother facilitate her child’s
learning of language?”
“Believe that your child can understand more than he or she can say,
and seek, above all, to communicate. To understand and be
understood. To keep your minds fixed on the same target. In doing
that, you will, without thinking about it, make 100 or maybe 1000
alterations in your speech and action. Do not try to practice them as
such. There is no set of rules of how to talk to a child that can even
approach what you unconsciously know. If you concentrate on
communicating, everything else will follow” (Brown, 1977, p. 26).
The same, I am hypothesizing, holds for second language acquisition.
Suggestions
The Net Hypothesis is, of course, a hypothesis. As is the case with all
scientific hypotheses, it could be refuted tomorrow. I suggest here
some modest ways of introducing non-targeted comprehensible input
into classes, and at the same time further test whether the hypothesis
is correct.
In class
I suggest we consider loosening up class discussions and in-class
stories. The focus in TPRS has been making input 100%
comprehensible, with students being able to understand, and translate,
every word (Ray and Seeley, 2008). Some beginners, because of bad
experiences in other classes, might require fully transparent input at
first, but it might be more efficient, and easier, to gradually relax the
transparency constraint and insist only that the input appear to be fully
comprehensible. I am suggesting that it is ok, and even desirable, that
the input contain a small amount of “noise,” or i+n.
Note that some late-acquired structures have little communicative
value. The third-person singular –s in English is hard to avoid in
English input, yet it is acquired very late. English acquirers have no
trouble understanding input containing –s because it contributes so
little to meaning. “Teaching” –s to beginners is useless, because it is
late-acquired, and “simplifying” the input to exclude it is hopeless.
This can be tested by examining teacher-talk in non-targeted classes.
The Net Hypothesis predicts that the appropriate grammar and
vocabulary will be included and that substantial language acquisition
will take place.
Readers
A modest first step is the creation of readers that are not targeted at
certain structures and vocabulary. Instead of writing stories that
include just those items that have been taught or are about to be
taught, writers can just try to make the texts interesting and
comprehensible, based on their own experience with students at the
beginning levels. If beginning students understand the texts (and like
them), then the texts are appropriate; the Net Hypothesis claims that
just the right aspects of language will be automatically included.
To see if the Net Hypothesis is correct, as suggested just above, we
can examine the texts of comprehensible/interesting readers written in
this way and determine what structures and vocabulary are covered.
We can also compare the achievement of classes using these texts
with those using readers matched to a grammatical syllabus and
vocabulary list.
Summary
This corollary of the Comprehension Hypothesis makes life much
easier and more interesting for teachers and students: If
comprehensible input, when provided in quantity, contains all the
structures and vocabulary the acquirer is ready for, we are liberated
from the constraint of targeting specific aspects of form and can focus
entirely on meaning, on providing input that is comprehensible and
compelling.
If only the feeling of full comprehension is required, if input is allowed to
contain some i+n, we no longer have to make sure that every word and even
every morpheme is completely transparent. If, in fact, if input is truly
compelling, it is likely that students will not even notice the “noise” or bits
of incomprehensible and nontransparent elements in the input.
*Thanks to Contee Seeley for helpful suggestions.
References
Brown, R. 1977. Introduction to Snow and Ferguson. In C. Snow and C.
Ferguson (Eds.), Talking To Children. pp. 1-27. Cambridge: Cambridge
University Press.
Cross, T. 1977. Mothers’ speech adjustments: The contribution of selected
child listener variables. In C. Snow and C. Ferguson (Eds.), Talking To
Children. pp. 151-88. Cambridge: Cambridge University Press.
Csikszentmihalyi, M. 1992. Flow: The psychology of optimal experience.
New York: Harper Perennial.
Gleitman, L., Newport, E. and Gleitman, H. 1984. The current status
of the motherese hypothesis. Journal of Child Language 11: 43-79.
Krashen, S. 1981. Second Language Acquisition and Second
Language Learning. New York: Prentice Hall. Available at
www.sdkrashen.com.
Krashen, S. 2004. The Power of Reading. Portsmouth, NH: Heinemann and
Santa Barbara: Libraries Unlimited.
Krashen, S. The compelling input hypothesis. The English Connection
(KOTESOL). In press.
Krashen, S. and Terrell, T. 1983. The Natural Approach: Language
Acquisition in the Classroom. Hayward, CA: Alemany Press.
Ray, B. and Seely, C. 2008. Fluency through TPR Storytelling.
Berkeley: Command Performance Language Institute. (Fifth Edition)