Krashen's Monitor and Occam's Razor

KEVIN R. GREGG
Matsuyama University
Matsuyama, Japan

Among the effects on second language acquisition research of the Chomskyan

revolution in linguistics has been the creation of something of a vacuum as far as theory
is concerned The older behaviorist model of acquisition has been discarded, for the
very good reason that it was hopelessly inadequate, but no new model of comparable
scope has been put forward, although a great deal of interesting work has been done in
specific areas of acquisition research.' There are several possible explanations for this
state of affairs. One is that second language acquisition may simply be too difficult and
too complex to be dealt with in a single theory. This seems like a reasonable idea,
especially as Chomsky (1980:90) has suggested the same sort of thing for first
language acquisition, where there are arguably fewer variables to worry about
. . . we should not expect a unitary answer to the question 'What is knowledge of human
language and how does it arise?' Rather, we will find that the question was wrongly put; it
does not identify a natural kind.
Another possible reason for this absence of an overall theory of second language
acquisition, I believe, is the growingriftbetween second language acquisition research
and generative linguistic theory, which itself is a consequence of the increasing
difficulty of linguistics, and of what is perceived as the increasing irrelevance of
theoretical linguistics to acquisition research. This, if true, is to be regretted, since a
theory of language acquisition without a linguistic theory is doomed to
inconsequentiality. Both of these reasons should be kept in mind when we examine, as
I propose to examine, what is probably the most ambitious and most influential
attempt in recent years to construct an overall theory of second language acquisition;
namely, Stephen Krashen's 'Monitor Theory'.2
Krashen presents five main hypotheses that 'make up a coherent theory of second
language acquisition' (1982:2).31 propose to examine these hypotheses one by one, to
look at the evidence and argumentation that Krashen offers in support of them, and (to
let the cat out of the bag) to show that he has not in fact presented us a coherent theory
of second language acquisition. The hypotheses are:
1 the Acquisition/Learning Hypothesis
2 the Monitor Hypothesis
3 the Natural Order Hypothesis
4 the Input Hypothesis, and
5 the Affective Filter Hypothesis.

1. THE ACQUISITION/LEARNING HYPOTHESIS

The Acquisition/Learning Hypothesis 'states that adults have two distinct and
independent ways of developing competence in a second language' (1982:10):
acquisition, which is 'subconscious', and learning, which is conscious.4 Evidently,
'acquisition' is to be identified with Chomsky's language acquisition device (LAD),
although Krashen is not very clear on this: 'acquisition' is 'a process similar, if not
Applied Linguistics, VoL 5, No. 2

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 80 KRASHEN'S MONITOR AND OCCAM'S RAZOR

identical, to the way children develop ability in their first language' (1982:10).5
'Learning', on the other hand, refers to 'conscious knowledge of a second language,
knowing the rules, being aware of them, and being able to talk about them' (1982:10).
The distinction between conscious and unconscious knowledge, or between
conscious and unconscious mental processes, is certainly by now uncontroversiaL But
Krashen of course recognizes this, and his hypothesis goes much further. Specifically,
he claims that (1)'adults can access the same natural "language acquisition device"
that children use" (1982:10), and (2) that' learning does not" turn into" acquisition'
(1982:83).
One effect of these two claims is to reduce drastically the difference in language-
learning ability between children and adults, since on the one hand both have the same
LAD, and on the other the adult's superior cognitive abilities, memory, pragmatic
knowledge, etc., are apparently next to useless. Thus the Acquisition/Learning
Hypothesis places a heavy burden on Krashen to explain just why it is that so few
adults successfully acquire a second language. This is where the Affective Filter
Hypothesis comes in, which we will consider later.
First of all, it is important to see just what is involved in the claim that acquisition in
adults is the same as acquisition of a first language by children. Although Krashen
does not make himself clear on this, he seems to be giving the LAD a scope of
operation much wider than is normally the case in linguistic theory. Indeed, he seems
to equate the LAD with unconscious acquisition of any sort 6 This is certainly not in
the spirit of the inventor of the term: as Chomsky and others have argued (see, e.g.,
Chomsky 1980:40-4,54-5,92ff.; Lightfoot 1982:42-9), the mind is modular, that is,
LAD is but one of various' mental organs' that interact with each other and with input
data to produce linguistic competence. Conceptual knowledge, real-world knowledge,
common sense, pragmatic competence, etc., are all necessary for understanding and
using language, but they are not part of LAD.
In any case, the LAD as a construct is intended to describe the child's initial state,
before being presented with primary linguistic data (PLD). The LAD is constrained
by ianate linguistic universals (Universal Grammar, or UG) to project grammars to
account for the PLD. Indeed, the terms LAD and UG are in effect interchangeable
(see, e.g., Chomsky 1981:34-5). It is not immediately clear how this concept of LAD
can be applied to an adult Not only is an adult not in an initial state with respect to
language, but he also is endowed with a much richer set of cognitive structures, which
theoretically at least could enable him to violate the constraints of UG (and thus put
him at a disadvantage vis-a-vis the child in terms of language acquisition). For
instance, to take an example frequently used by Chomsky (e.g., 1975:30-3):
1 The man is talL
2 Is the man tall?
3 The man who is tall is in the room.
4 Is the man who is tall—in the room?
5 *Is the man who—tall is in the room?
As Chomsky points out, given( l)-(3) as PLD, it would be perfectly logical to form a
hypothesis about yes/no question formation in English that would lead to the
production of (5) rather than (4). In fact such a hypothesis is simpler than the correct
hypothesis. None the less, children do not form such hypotheses, presumably because
they cannot their hypothesis-forming powers are restricted by UG; here, by the
principle of structure-dependency. But—what is to the point here—an adult can form

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 81

such a hypothesis; Chomsky did it, and so can you. Whether in fact adult second
language learners do violate UG is another question; a very interesting question in
fact, and one that is worth investigating.7 Krashen provides no evidence one way orthe
other. In any case, it is not enough merely to say that adults have a language
acquisition device, without specifying what that device does.
Of course, some adults do attain native-speaker competence in a second language,
and of course such competence is largely unconscious; no one has ever denied this. In
this sense, the Acquisition/Learning Hypothesis is not a hypothesis, but simply an
observatioa But Krashen also claims that 'learning' cannot become' acquisition', and
this is what makes the hypothesis interesting. Unfortunately, it also makes the
hypothesis a bit inconsistent, since if'learning' cannot become' acquisition', and if—
as Krashen goes to unnecessary lengths to remind us (1982:92-4)—most of our
knowledge of a second language is necessarily unconscious, then it makes little sense
to call 'learning' one of'two distinct and independent ways of developing competence
in a second language'. But that is by the way. The question is, is the claim a valid one?
On the face of it—and this is what makes it the interesting part of the hypothesis—it
is nonsense, and Krashen himself seems to be aware of this: 'The idea that we first
learn a new rule, and eventually, through practice, acquire it, is widespread and may
seem to some people to be intuitively obvious It was, I thought, exactly the way I
learned languages myself (1982:83). Well, it certainly does seem intuitively obvious
that at least some rules can be acquired through 'learning1. I learned the rules for
forming the past tense and gerundive forms of Japanese verbs by memorizing the
conjugation chart in my textbook and—like most of my classmates—was pretty much
error-free after a couple of days, with no input other than a bit of drilL I learned the
super-polite forms of Japanese adjectives by asking a friend for a few examples and a
short explanation. I know these rules, I am aware of them, I can talk about them. These
seem to me to be cases of' learning' becoming' acquisition', and fairly typical cases at
that 8
It is worth emphasizing not only that I 'learned' these rules, but that I 'acquired'
them—if error-free, rapid production is a criterion for' acquisition'. And I ' acquired'
them instantaneously, to all intents and purposes, and in the absence of virtually any
input I stress this point since Krashen insists on several occasions that'acquisition'
' always takes time and requires a substantial quantity of input data' (1982:114).' This
seems plausible on the face of it, but it is not enough simply to make such an assertion
dogmatically, in the absence of evidence. After all, first language acquisition proceeds
on the basis of a rather small quantity of PLD, and at such a rapid pace that linguists in
the Chomskyan tradition feel justified in idealizing to instantaneous acquisitioa And,
as Roeper (1981:5) points out, 'later stages in acquisition are often marked by a
sudden ability in children to move from no response to a correct response (in an
experiment) with no evidence of intermediate trial hypotheses'. (See, e. g., Carey 1978
on lexical acquisition in three- and four-year-olds.)
The burden is thus on Krashen to disprove the intuitively obvious proposition that
' learning' can become' acquisition'. Unfortunately, he does not do this. He does show
(1982:84-7) that'leaming^ need not precede'acquisition', but he does not show that it
cannot He gives three arguments: (1) sometimes there is 'acquisition' without
'learning'; (2) sometimes 'learning* never becomes 'acquisition'—e.g., someone who
knows the rule but still keeps breaking it; (3) no one knows anywhere near all the rules.
These are all true, but are not evidence that 'learning' cannot become 'acquisition'.
More specifically, Krashen has not shown that presentation of rules, explanation, etc.,

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 82 KRASHEN-S MONITOR AND OCCAM'S RAZOR

cannot facilitate the acquisition of a second language, which is the very strong claim
that he is making.I0
One of the problems with the Acquisition/Learning Hypothesis is that Krashen
plays fast and loose with his definitions. 'Acquisition' is a process leading to
competence; 'learning5 is evidently a state ('knowing the rules, being aware of them,
and being able to talk about them'). If Krashen means that a given state cannot become
a process, he'll get no argument from me, or from anyone else. But the position that he
is attacking is the position that learning (not 'learning") can lead to acquisition (not
'acquisition') of what one has learned. Krashen is simply wrong when he says, 'Some
second language theorists have asumed that children acquire, while adults can only
learn' (1982:10). He is confusing learning with 'learning'; I know of no one who has
maintained that adults can only' learn', and to argue against such a position is to attack
a straw man.
Similarly, Krashen fails to make clear what he means by 'conscious' and
'subconscious', as McLaughlin (1978) has pointed out Specifically, does
'subconscious' mean 'not accessible to the conscious', or simply 'not conscious at a
given momenf? Does 'conscious' entail 'incapable of becoming unconscious'? If by
definition the 'subconscious' is inaccessible, and conscious 'learning' is always
accessible, then Krashen's claim that 'learning7 does not become 'acquisition' is of
course trivially true, but uninteresting. On the other hand, if (some) unconscious
knowledge is capable of being brought to consciousness, and if conscious knowledge is
capable of becoming unconscious—and this seems to be a reasonable assumption—
then there is no reason whatever to accept Krashen's claim, in the absence of evidence.
And there is an absence of evidence.
As it stands, then, the Acquisition/Learning Hypothesis, far from being'potentially
the most fruitful concept for language teachers that has come out of the linguistic
sciences' in recent years (Stevick 1980:270), is either clearly false or trivially true.
The triviality of the distinction becomes all the more striking when we turn to the
second of Krashen's five hypotheses.

2. THE MONITOR HYPOTHESIS

The Monitor Hypothesis claims that 'Learning has only one function, and that is as a
Monitor, or editor5 (1982:15). Further, the Monitor cannot be used unless the
following three necessary but not sufficient conditions are met (1) time ('In order to
think about and use conscious rules effectively, a second language performer needs to
have sufficient time'); (2) 'focus on form' ('The performer must also be focussed on
form, or thinking about correctness'); and (3) knowledge of the rule (1982:16).
Notice first that the Monitor Hypothesis claims that'learning' is available only for
use in production, not in comprehension.11 This is an interesting—nay, an
extraordinary—claim to make, and, once again, one for which Krashen offers no
evidence whatever. (Again I should mention that McLaughlin (1978) also makes this
point, and that Krashen does not answer it in his reply (1979) to McLaughlin.) It
should be superfluous to offer evidence against this claim, but none the less I will give a
couple of examples, by no means atypical, from my own experience with Japanese.
The other day while listening to the radio, I heard the announcer announce wagunaa
nokageki, kamigami no kasoware. Knowing that kageki = 'opera' andthat&ami =
either'god1 or'hair1 or'paper", and knowing that there is a (fairly unproductive) rule in
Japanese for pluralizing by reduplication, I concluded that kamigami must be the
plural of kami 'god', that therefore wagunaa must be Wagner and kasoware must

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 83
1
mean' twilight , and that I was in danger of hearingjD/e Gotterdammerung. Of course I
was not quite right there is no word kasoware, if s tasogare. But the point is that I was
using a rule that I had 'learned* (and never used productively), and using that rule
consciously (and quickly enough to turn the radio off in time). Again, hearing someone
utter a sentence ending ia-nakaroo, I deduced (correctly, as it turned out) that this
was the equivalent of-nai daroo, and that I could therefore expect to hear takakaroo
instead oftakai daroo, etc., from this speaker at least" Of course in both cases I was
using acquired knowledge, but I was also consciously making use of rules that I had
'learned'. Which suggests that'learning' can indeed be used in comprehension, as no
one before Krashen would have doubted.
But let us, for the sake of argument, accept this 'fundamental claim of Monitor
11160^(1981:2); what follows? For one thing, the Acquisition/Learning distinction
becomes even less important, especially since—as we will see when we look at the
Input Hypothesis—we learn a language through comprehension. The distinction
becomes all the more pointless when one examines the three necessary conditions for
using the Monitor. Condition 1, time, is fairly uncontroversial; one needs time to
produce an utterance, and the more one relies on conscious knowledge of rules, the
more time will be required. Condition 2, focus on form, is really a false distinction;
with a few exceptions (e.g., in English, the -s ending of the third person singular
present, plural markers, etc.) focusing on form is focusing on content The differences
betweenaimerez zndaimeriez, betweenyobaseru andyobareru andyobas(er)areru,
between site iru and sine iru, between Tanaka-san and Tanaka-kun, are all formal
differences, but they are also differences in content13 If on the other hand 'focus on
form' means no more than 'thinking about correctness', then there is no distinction
being made between form and content, and the condition boils down to trying to say
what one means to say. This is not such a nitpicking argument as it may seem, as we
can see when we look at some of the putative evidence for Monitor use. For instance,
Krashen distinguishes between morpheme acquisition studies that permit or require
Monitor use and those that do not; the latter are 'studies that used Monitor-free
instruments, that is, . . . studies where language was used for communication'
(1981:57). Krashen's idea of'communication' is truly bizarre. Take, for example, the
following two types of test question:

1 Here is a wug. Here are two of them.

Here are two
2 Here are two (wug) .
(1), which is the format of the SLOPE test (Fathman 1975), is evidently a
'communicative' test, whereas (2), used by Larsen-Freeman (1975), is not The
difference is that Fathman's results, but not Larsen-Freeman's, support the Natural
Order Hypothesis. Thus the subjects taking the SLOPE test were evidently trying to
'communicate', while Larsen-Freeman's subjects were focused on form, or thinking
about correctness. This, I submit, is absolute nonsense.14
Condition 3, 'know the rule', is of course correct in a sense, but again only in a rather
trivial way, it is less restrictive a condition than Krashen suggests, since the language
learner's idea of a rule need not coincide with the linguist's. If I want to use the verb
/asob/ in the past tense, I must say asonda. Assuming that I haven't 'acquired' this
form yet, I can make appeal to my conscious' learning', which might supply any of the
following 'rules': the past tense of/asob/ is asonda; the past tense of verbs ending in
/b/ is /nda/; the past tense of verbs ending in /b/, /m/, or/n/ is/nda/; the past tense of

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 84 KRASHEl>rS MONITOR AND OCCAM'S RAZOR

verbs ending in a labial consonant is /nda/, etc. None of these rules is correct, as far as
a linguist is concerned; that is, while they are all observationally adequate, none of
them attains descriptive, let alone explanatory, adequacy. " Still, they all do the job of
getting therightsounds uttered. When Krashen reminds us that' Linguistics has taught
us that the structure of language is extremely complex, and they claim [sic] to have
described only a fragment of the best known languages' (1982:16), or when he asks
rhetorically, 'How many rules did Jespersen (ever) know, how much of English have
scholars such as Noam Chomsky described?' (1982:92), he is simply missing the
point The rules that Chomsky is looking for are not the rules the language learner
needs. As Chomsky himself says, 'the concerns of a linguist seeking to determine the
nature of language are virtually complementary to those of a person trying to learn a
language' (Chomsky 1980:98-9).
Still, these objections to the formulation of the three conditions on use of the
Monitor are not necessarily objections to the conditions themselves. What follows
from these conditions? According to Krashen, 'for most people, even university
students, it takes a real discrete-point grammar-type test to meet all three conditions
for Monitor use . . . ' (1982:18). Which is to say that under normal conditions, the
Monitor cannot be used. Since the Monitor is the only way in which conscious
knowledge of rules ('learning') can be used, conscious knowledge of the rules is of no
use as far as acquisition is concerned. But if conscious knowledge of the rules is of no
use for acquisition, then it makes no sense to say that there are two ways,' acquisition'
and 'learning', to gain competence in a second language. Therefore, the Acquisition/
Learning Hypothesis is false. Q.N.E.D.
The Acquisition/Learning Hypothesis and the Monitor Hypothesis taken together
are self-contradictory, since Krashen offers no evidence for either ofthese hypotheses,
it is hard to reconcile the contradiction, and I am not going to try. Instead, let us move
on to the next hypothesis.

3. THE NATURAL ORDER HYPOTHESIS

The Natural Order Hypothesis suggests that' Second language acquirers acquire (not
learn) grammatical structures in a predictable order* (Krashen 1980:169). Here, as
far as evidence goes, we are on more solid ground there has been no end of morpheme
studies, both in first and second language acquisition research, and most of the studies
seem to support this hypothesis, at least as far as the morphemes studied are
concerned. There have been occasional challenges to the validity of these morpheme
studies (e.g., Porter 1977, Rosansky 1976), but I am going to assume that Krashen's
defense (see, e.g., 1981:55-63) is correct; that is, that the orders obtained are not
artifacts of the test instruments used, and that the orders obtained are indeed
acquisition orders. This is a far cry, however, from assuming that the hypothesis is
correct; and granting that the hypothesis is correct still leaves us with the problem of
interpreting the hypothesis.
First of all, what is a 'structure'? Krashen is singularly unhelpful here: the learner
'acquires the different structures that make up a language: tense endings, plural
markers, negative sentences [sic], complements, and so forth' (Dulay, Burt, and
Krashen 1982:200). As Krashen himself points out, 'of all structures of English,
morphology [sic] is the most studied" by those testingfor acquisition orders (1982:12):
that is, grammatical morphemes such as Past Tense, the Progressive -ing, etc. The fact
that these morphemes are acquired in (roughly) the same order by various learners
does not entail that the acquisition of, say, relative pronouns, counterfactuals, indirect

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 85

object placement, or epistemic modals is ordered, in respect to those grammatical

morphemes or in respect to each other. Furthermore, as Wode et al (1978:184) point
out,' there is no reason to assume in advance that such items as the various inflections,
prepositions . . . , or the copula, are acquisitionally comparable.... Moreover, if it
does not matter what is compared in terms of morpheme order, then there is no reason
why one should not determine the order of acquisition for, say/r/ vs. negation.' In an
earlier book Krashen had acknowledged this point, in a footnote—'Wode et al note
that ordering studies might be overly concerned with determining relative order of
acquisition of items that are formally quite different" (1981:63)—but in later writings
this objection is ignored. And in any case Krashen misses the point it is not that the
morpheme studies (which are the only evidential underpinning for the Natural Order
Hypothesis) 'might be overly concerned' with formally different items; rather, they
make no principled distinction between comparable and non-comparable' structures'.
Grammatical morphemes have been used in so many acquisition studies because they
are easy to use; one can count the number of correct utterances of such morphemes on
obligatory occasions (see Brown 1973:255). But in the absence of a rich enough
linguistic theory, there is no a priori reason to assume that, say, progressive -ing and
third person-s are comparable, let alone /r/ and negation; that is, there is no reason to
assume that there is significance to the acquisition of the one before the other. Since,
aside from paying lip service to LAD, Krashen has no linguistic theory to relate to the
data from the morpheme studies, we are left with a hy; othesis in search of an
interpretation.
Actually, if we are to extrapolate from grammatical morphemes to'structures' lato
sensu, the Natural Order Hypothesis seems absurd. The strong form of the hypothesis
would have the learner acquiring the, say, 3,217 structures of English by progressing
from 1 to 3,217; or, for the unfortunate majority, stopping at 2,983 or 1,705 or
wherever. (Alternatively, the learner progresses from one group of structures to the
next group, with random order within the group.) On this view, the acquisition of a
second language is rather like the acquisition of dinner we start with the soup, proceed
to the fish, then the meat, and so oa Of course, there is individual variation; some of us
will eat all the asparagus before starting in on the mutton, while others will take
alternate nibbles of meat and vegetables; and some of us will (for affective reasons)
turn down the port and cigars. This is to me an extraordinarily unlikely model of
acquisition. (See, e. g., White 1981 or Lightfoot 1982 for comment on less silly forms
of this model.) In fact Krashen seems to recognize this, again in a footnote, when he
says that 'a strictly linear view of the natural order hypothesis, that there is only one
stream of progress that acquirers follow in strict sequence' is incorrect Rather,
'several streams of development are taking place at the same time' (1982:53-4)."
What this is supposed to mean is hard to say, as Krashen gives us no explanation of
what a 'stream of developmenf could be, nor does he set any limit on the number of
such 'streams'. One definite consequence of allowing more than one 'natural order1,
however, is to vitiate the Natural Order Hypothesis. If the structures of, say, English
are divided into varying numbers of ordered sets, the number of sets varying according
to the individual, then it makes little sense to talk abouta 'natural order*. If the number
of sets varies from individual to individual, then the membership of any given set will
also vary, which makes it very difficult to compare individuals, especially since the
content of these sets is virtually completely unknown. If the set of sets of structures is
claimed to be invariant across individuals—that is, if it is claimed that there is one
(unknown) fixed number of streams of developmenf—the problem of comparability

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 86 KRASHEN'S MONTTOR AND OCCAM'S RAZOR

would be removed, but the problem of empirical support for the hypothesis remains; in
fact it becomes even greater. If on the other hand one rejects the criterion of
comparability across individuals, the Natural Order Hypothesis has no empirical
content for any individual it would claim that the acquisition of structure n+1 was the
next structure to be acquired, and this would be true ex hypothesi, but it would be
impossible to prove it or falsify it As we shall see, the problem of separate 'streams'
poses problems for the Input Hypothesis as welL
It is difficult to say where Krashen himself stands with respect to the Natural Order
Hypothesis. Although the footnote quoted above seems to indicate that he is, in effect,
abandoning the hypothesis, on other occasions he seems to be accepting the
implications of the strong form of the hypothesis. Take, for example, the much-cited
' F (Krashen and Pon, 1975), an example of the successful Monitor user, a Chinese
woman who writes and speaks almost flawless English, but whose spoken English
contains occasional mistakes: 'e.g. omission of the third person singular ending,
incorrect irregular past tense form, failure to make the verb agree with the subject in
number...' (Krashen and Pon 1975:126; cited in Krashen 1981:13). How do we
account for these occasional mistakes? Krashen of course considers P to be an
example of'learning" (of these broken rules) without'acquisition'. (An alternative
explanation that Krashen does not consider is that P has indeed acquired the rules she
broke, and that her errors are simply performance errors. Since only eighty errors in all
were gathered over a three-week period, and since presumably P often (usually?) did
not break the rules involved, this seems as plausible an explanation as invoking the
failure of a putative Monitor. After all, Krashen himself seems to have some difficulty
with subject-verb agreement, and in presumably Monitored situations!17) His
explanation is instructive:' her acquisition, while very advanced, had simply not gone
the final few steps in syntax and morphology* (1982:86). In other words, 'structures'
such as third person -s are at or near the end of the' natural order". Or else, they are at or
near the end of one 'stream', but still this stream is evidently later than most other
streams in the natural order of streams.
In any case, what follows from this hypothesis? The original studies of morpheme
acquisition in second languages (e.g., Dulay and Burt 1974, Bailey, Madden, and
Krashen 1974) showed a similar acquisition order for certain English morphemes
regardless of the subject's first language, a result strongly suggesting that the
Contrastive Analysis Hypothesis, at least in its strong form, was incorrect; this-was an
important finding. But beyond that, and beyond showing that there are evidently
universal principles of some sort in second language acquisition, it is not clear what
theoretical conclusions should follow. As Dulay, Burt, and Krashen point out
(1982:214), 'we still have not discovered the principles underlying the L2 order by
itself nor [sic] the LI order by itself. One likely reason for this failure is that the
researchers ignore grammatical theory, and thus deprive themselves of any means of
interpreting their findings. Indeed, one of the most striking things about the Natural
Order Hypothesis, as it is presented by Krashen, is the almost stunning lack of
intellectual curiosity it reveals. This is starting to change in first language acquisition
research (see, e.g., Tavakolian 1981 or Baker and McCarthy 1981), but not yet in
second language acquisition research, as far as I can see.18
One practical consequence of the Natural Order Hypothesis that might occur to us
would be to organize the syllabus to coincide with the' natural order*. (Of course, there
is a practical difficulty in that no one knows what this 'natural order* is. But if the
Natural Order Hypothesis is correct, it should in principle be possible to determine the

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 87

order.) But, although he chides thoughtless teachers for violating the order (Dulay,
Burt, and Krashen 1982:58, 200-1), Krashen maintains that we should 'reject
grammatical sequencing in all cases where our goal is language acquisition' (Krashen
1982:14); that is to say, always. Just why we should not follow the 'natural order1,
aside from the fact that we can't, is not self-evident; for Krashen's explanation we must
turn to the Input Hypothesis.

4. THE INPUT HYPOTHESIS

The Input Hypothesis, says Krashen, 'may be the single most important concept in
second language acquisition today", since' it attempts to answer the crucial question of
how we acquire language' (1980:168). The problem, as Krashen sees it, is this: 'given
the correctness of the natural order hypothesis, how do we move from one stage to
another?... how do we move from stage i, where i represents current competence, to
H-l, the next level?'(1982:20-1). As we have seen, it is hard to grant the correctness
of the Natural Order Hypothesis, but even if we do, we still do not have an explanation
of acquisition, since we have as yet no mechanism for moving along any given' stream
of progress'.19 Krashen at least seems to recognize this problem, and the Input
Hypothesis aims at providing a mechanism which will explain the learner's progress.
The Input Hypothesis is presented in four parts, which I will quote in their entirety:
1 The input hypothesis relates to acquisition, not learning.
2 We acquire by understanding language that contains structure a bit beyond our
current level of competence (H-l). This is done with the help of context or extra-
linguistic information.
3 When communication is successful, when the input is understood and there is
enough of it, H-l will be provided automatically.
4 Production ability emerges. It is not taught directly. (1982:21-2.)
I will ignore (1) and (4); the latter follows from the former, and the former is
superfluous, given the Acquisition/Learning Hypothesis. We may note in passing that
(3) is the main justification for not organizing the syllabus according to the 'natural
order1.
As always, Krashen is not terribly clear; his H-l, for instance, sometimes seems to
mean the next structure to be 'acquired', and sometimes to mean the learner's
competence at one 'stage' after stage i.20 Still, there are a couple of empirical claims
being made here. One of them is that output does not help 'acquisition' except
indirectly (1982:57, 60-2). Another is that we 'acquire' a previously 'unacquired'
structure if and only if (1) that structure is present in input that we understand; (2) that
structure is 'due' to be 'acquired' next; and (3) that structure is presented (in
understood input) sufficiently oftea A third claim is that we 'acquire' grammatical
structures through extra-linguistic means: 'How can we understand language that
contains structures that we have not yet acquired? The answer to this apparent
paradox is that we use more than our linguistic competence to help us understand. We
also use context, our knowledge of the world, our extra-linguistic information to help
us understand language directed at us' (1982:21).
The first claim is one of many that Krashen makes without bothering to give any
evidence. Since there is a fairly widespread opinion (which may of course be totally
unfounded) that 'practice' is necessary for second language acquisition, it is all the
more incumbent on Krashen to back up his assertion. Even within the framework of
Krashen's theory, it would seem at least theoretically possible that output could be

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 88 KRASHEKS MONITOR AND OCCAM'S RAZOR

available as input; there is no a priori reason to assume that a learner systematically

ignores his own utterances. If output is available as input, and if Monitoring can
increase the incidence of correct utterances of a given structure, then it would seem
that output is being used to further acquisition, and thus that the Monitor can be used
for acquisition. Krashen entertains this idea briefly (1982:84; 1981:118), but only to
reject it The only reason that I can see for rejecting it is that it conflicts with that part of
the Acquisition/Learning Hypothesis that claims that 'learning' does not become
'acquisition'. Since that claim is not backed up by evidence either, it hardly serves as
support for this claim.
As for the second claim, (1) it seems plausible enough that a structure that is never
present in the input will not appear in the output (This does not, however, go without
saying; it is at least possible that in at least some cases a learner could acquire
appropriate structures without any input of those structures.) On the other hand, it
remains to be proven that we need to understand the utterance containing structure X
in order to understand or acquire structure X (e.g., -nakaroo). (2) Since we don't
know when a structure is ready to be acquired, it is hard to demonstrate whether this is
true or not C learly, if a structure has been acquired it was acquirable, but that does not
mean it was' ready" for acquisition, since we have no definition of ready". (3) This too
is plausible, but ambiguous: either frequent exposure is necessary because the learner
must in some unspecified sense 'practise' the structure before he can' aquire' it, 21 or
else frequent exposure is necessary because we don't know which (single) exposure
will do the trick. In the latter case, if we knew the' natural order" we might be able to do
without extensive exposure. (Of course, the two interpretations of(3) are not mutually
exclusive; it is also possible that even if we could avoid 'waste' exposures we would
still need a large number of exposures in order to' acquire' the structure i+1.) Krashen
does not make himself clear in this regard.
The third claim—that we acquire through extra-linguistic knowledge—is an
astonishing one. I find it difficult to imagine extra-linguistic information that would
enable one to 'acquire' the third person singular -s, or yes/no questions, or indirect
object placement, or passivization. Of course one may understand a given utterance
without understanding all of the structures of the input; indeed, this happens all the
time. But it by no means follows from this that one has acquired the unknown structure.
For example, if one has acquired the basic SVO pattern of English sentences, then,
being presented with sentences like John was bitten by the dog, and with the
appropriate contextual evidence—John with tooth marks on his arm, a dog, etc.—one
might be able to understand the sentences, and even unconsciously develop a rule of
passivization. But understanding thaUohn was bitten by the dog means the same thing
as The dog bit John does not mean that one has acquired the rule of passivization.
Such presentation of input would not in itself enable a learner to avoid producing
sentences like */ was rained, *I was died by my father, or conversely *The river
situated the house. Although Krashen pays Up service to the idea of a LAD, in fact he
seems to be offering a kind of parody of a 'developmental* theory of acquisition of a
Piagetian sort He cites MacNamara (1972) in support of his claim, for instance,
saying that 'we acquire by "going for meaning" first, and as a result, we acquire
structure!... As MacNamara (1972) pointed out, the child does not acquire grammar
first and then use it in understanding. The child understands first, and this helps him
acquire language' (1982:21,23). Aside from the dubiousness of the analogy between
first and second language acquisition here—the adult second language learner
presumably already has the semantic or cognitive structures that the child is acquiring

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 89

—neither Krashen nor MacNamara gives us any explanation of how such a step from
understanding to acquisition is made. Simply to state that understanding helps
acquisition is to make an uncontroversial observation, but one without any empirical
content"
Krashen's use of 'caretaker speech' (CS—also known as 'motherese' or 'baby
talk")—that is, the speech of mothers and other caretakers to young children—as
evidence for the Input Hypothesis is illustrative of this faulty reasoning. Krashen tells
us that C S has three characteristics relevant to the hypothesis: (1) it is not a deliberate
attempt to teach language; (2) 'while it is syntactically simpler than adult-adult
speech', CS 'is "roughly-tuned" to the child's current level of linguistic competence,
not "finely-tuned" '; (3) CS follows the 'here-and-now principle': 'caretakers talk
mostly about what the child can perceive, what is in the immediate environment"
(1982:22-3).
Before asking the obvious question— so what?— it would be well to look a little more
closely at these three characteristics. First of all, as Newport, Gleitman, and
Gleitman (1977), Krashen's main source of data on CS, point out, CS is not
syntactically simpler than adult-adult speech (AS): it has a far lower proportion of
canonical SVO declaratives than AS (30 per cent and 80 per cent respectively);23 it
has many more 'deformed' clauses (Le., clauses showing deletion or movement of
elements) than AS (72 per cent, 55 per cent); it has a much higher proportion of
questions and imperatives than AS (62 per cent, near 0). 'Overall then, "syntactic
simplicity" is a pretty messy way to characterize Motherese' (p. 122). And, in the face
of these data, to say that CS is syntactically simpler than AS is a pretty messy way, to
say the least, to use data as evidence for one's hypotheses. The one way in which CS
can be said to be simpler than AS, according to Newport et al, is in terms of
propositional content (number of verbs or noun phrases per utterance). But, as they
point out (pp. 132-3), and as Krashen omits to point out (1982:22-3), there is no
significant correlation between CS input and growth in propositional complexity of
the child's utterances. 'The child's growth in the use of complex sentences . . . is
unaffected by the aspects of Motherese examined here.' In other words, in the one area
where CS and AS can be said to be significantly different in terms of simplicity, there is
no evidence of tuning, rough or fine, on the part of the caretaker.
As for the 'here-and-now principle', it is worth quoting Newport et al in full
(p. 126): 'three special characteristics of Motherese (brevity, well-formedness and
intelligibility) arise for the purpose of here-and-now communication with a limited
and inattentive listener, and cannot be described in terms of a language-instruction
motive; this begins to suggest that they may not serve a language-learning purpose'
(emphasis added). Krashen seems at one point to acknowledge this point, but then
blithely ignores it 'While there is no direct evidence showing that caretaker speech is
indeed more effective than unmodified input, the input hypothesis predicts that
caretaker speech will be very useful for the child' (1982:23).24 Elsewhere (1981:124)
he goes so far as to suggest that CS 'might encourage language acquisition on the part
of the child'. As if encouragement were necessary, or even relevant!
Now we can ask the obvious question about CS: So what? What follows from the
fact that (some) caretakers speak differently—more carefully, more slowly, more
clearly, even perhaps more simply in some senses—to (young) children than to adults?
What consequences are there for first language acquisition theory, let alone for second
language acquisition theory? There seem to be none. No one—certainly not Krashen
—has shown that CS makes any significant contribution toward the acquisition of a

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 90 KRASHEN'S MONTTOR AND OCCAM'S RAZOR

grammar by a child, and no one has shown that the existence of CS has any bearing
whatever on second language acquisitioa As Wexler and Culicover (1980) put it,
'there is no theoretical justification for the claim that simplified input aids language
learning1 in children (p. 66; see pp. 60-84 for an excellent discussion of CS in
particular and of the role of input data in first language acquisition in general).
Krashen is of course right when he says that context helps a child (or an adult)
understand an utterance. Who has ever doubted this? Who has ever doubted that
speaking clearly and slowly to a foreigner will, ceteris paribus, be helpful to him?
But this does not explain how the child (or the adult) goes from understanding an
utterance to 'acquiring1 ' i + l ' . The only evidence that CS provides for the Input
Hypothesis is evidence that children receive input, scarcely earth-shaking news.
The 'evidence from second-language acquisition' is no more compelling
(1982:24-7). It consists mainly of repeating the claims of the Acquisition/Learning
Hypothesis and the Natural Order Hypothesis, and of noting that native-speakers
sometimes modify their speech, so that learners can also receive modified input, just
like children. This is true, of course; one thing that any good teacher, or any other
sensitive person, leams to do is to adjust his speech when talking to a foreigner."
Whether this modified output is, or should be, similar to CS is quite another question.
(For discussion, see Krashen 1981:128-32.) The other second language 'evidence'
that Krashen offers is the so-called'silent period': children learning a second language
commonly speak very little in the target language for the first several months. This
seems to be a fairly well-established phenomenon, but it is hardly evidence for (or
against) the Input Hypothesis. According to Krashen( 1982:27),' In accordance with
the input hypothesis, speaking ability emerges on its own after enough competence has
been developed by listening and understanding1. But the Input Hypothesis makes no
prediction about the silent period. The fact that production ability emerges does not
mean that there should be no production for several months. The silent period is not
inconsistent with the Input Hypothesis, but neither is it evidence for it
If Krashen wishes to argue that acquisition requires comprehensible input, no one
could object But he seems to be saying something more specific: that for children
learning their first language or adults learning a second language, acquisition is caused
by understanding the input In the absence of any evidence supporting this claim, and
in the absence of any theory explaining how we proceed from understanding to
acquisition, the Input Hypothesis must be rejected. As it stands, the Input Hypothesis
has no more explanatory power than Moliere' s doctor* s explanation that opium makes
one sleepy by virtue of its dormitive powers.

5. THE AFFECTIVE FILTER HYPOTHESIS

The final hypothesis to be examined is the Affective Filter Hypothesis. This was
originally proposed by Dulay and Burt (1977), who used the term 'affective
delimitors'; Krashen has revised it somewhat in an attempt to incorporate the so-
called ' affective variables' into his second language acquisition theory.26 Krashen lists
three major categories into which most of the affective variables studied can be placed
motivation, self-confidence, andanxiety(1982:31). The Affective Filter Hypothesis
claims, first of all, that there is such a thing as an affective filter, which is 'that part of
the internal processing system that subconsciously screens incoming language based
o n . . . the learner's motives, needs, attitudes, and emotional states' (Dulay, Burt, and
Krashen 1982:46)." The hypothesis also claims that'the effect of affect is "outside"
the language acquisition device proper" (Krashen 1982:32). Krashen offers two

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 91

models of second language acquisition (I present both because they are not quite
equivalent).
There are several problems with these models, if we are to take them seriously.
First, note that the LAD can be bypassed, at least in Model B. Model A does not
permit this, but Krashen says of learners with a 'high' Filter that 'even if they
Filter Monitor
Acquired
Input '• » LAD Competence Output

Figure 1: Model A (from Krashen 1982:16, 32)

i— -\1
1

Language Filter Organizer Monitor j. Verbal

—i (LAD)
environment ' performance

Figure 2: Model B (from Dulay, Burt, and Krashen 1982:46)

understand the message, the input will not reach that part ofthe brain responsible for
language acquisition, or the language acquisition device' (1982:31). Having been
filtered and having bypassed the LAD, what happens to the input? It is apparently
(optionally) Monitored before emerging as an utterance! This is a very unusual input-
output device, to say the least; the only situation to which I can see it applying
successfully would be a repetition drilL Model A makes more sense in that the output
ofthe LAD is competence not utterances, but it contradicts what Krashen says about
bypassing the LAD. These inconsistencies are not irreparable, but one wonders why
Krashen didn't bother to do the repairs. These problems arise, as I said, if one takes the
models seriously, but then perhaps Krashen doesn't really mean them to be taken
seriously. When he tells us that, on reading a textbook, he 'felt [his] affective filter
going up' (1982:166), we are justified in having doubts.
Both models have the Filter acting on all input—there is no bypassing the Filter,
although input presumably can get through totally intact The filtering power of the
Filter varies according to the individual and, presumably, within each individual
Here we have the answer to the question of why some learners never acquire full
competence: 'The filter hypothesis explains why it is possible for an acquirer to obtain
a great deal of comprehensible input, and yet stop short (and sometimes quite short) of
the native speaker level... When this occurs, it is due to the affective filter.... Child-
adult differences in attainment are not due to any change in the "language acquisition
device" (LAD) but are due to the filter . . . ' (1982:32, 45). This would be very
convenient, if true. Adults acquire a second language in the same way that children
acquire afirst(or second) language, and provided that they get enough ofthe right kind
of input, they should therefore become as competent as children become. If they don't,
ifs because they have some affective difficulty: lack of motivation, dislike orfearof the
culture of the target language's speakers, feelings of inferiority, what have you.

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 92 KRASHEN"S MONITOR AND OCCAM'S RAZOR

Once again, there is no denying the importance of the so-called affective variables in
second language acquisitioa Indeed, one of the most important differences between
first- and second-language acquisition is precisely that affect is important in the latter
and not in the former. But to go from this uncontroversial (and, in itself, uninteresting)
observation to the positing of an Affective Filter is a large step to take, and I think an
unjustifiable one.
For one thing, if the Affective Filter is what is responsible for incomplete mastery of
a second language, it is necessary to explain why the Filter does not work in children.
Krashen surely would not claim that children do not have affect, that they never suffer
from feelings of insecurity, anxiety, lack of confidence, inferiority, etc. Of course they
don't have motivation to acquire their first language (although Krashen seems to think
that they do, since he talks about CS encouraging children to acquire), but they
certainly have motivations of varying strengths. Why are all these emotions irrelevant
to the child's acquisition yet crucial to the adult5 s? We cannot appeal to differences in
the structure or function of the LAD, since Krashen has denied any such differences.
Nor can we appeal to changes in cognitive capacity, since these should give the
advantage to the adult Thus we are left with the problem of motivating the postulation
of an unconscious filtering mechanism. Krashen makes no serious attempt to explain
how or why this Filter should suddenly manifest itself at around the age of puberty.
The only attempt he does make consists of a perfunctory and generally irrelevant
discussion of Piaget's 'formal operations' stage of cognitive development, which
Krashen suggests is related to the development both of the Monitor and of the
Affective Filter 'significant Monitor use is only possible after the acquirer has
undergone formal operations [sic], a stage in cognitive development that generally
occurs at about puberty* (1982:44). Having reached this stage, the adolescent now
'has a meta-awareness [sic] of his ideas and can use abstract rules to solve a whole
class of problems at one time. It is thus plausible that the ability to use a conscious
grammar . . . comes as a result of formal operations [sic]' (1981:35). But, says
Krashen, following Elkind (1970), this cognitive stage, which includes 'a greater
capacity to conceptualize the thoughts of others', 'leads to the increased self-
consciousness, feelings of vulnerability, and lowered self-image that are associated
with [adolescence]. In our terms, it leads to an increased [sic] affective filter and a
subsequently lowered ability to acquire a second language' (1981:35).
There are several things to notice about this explanation. First of all, the appeal to
Piaget seems to be as gratuitous as the appeal to Chomsky. Since Krashen makes no
serious attempt to integrate the theories of either scholar into his theory, the references
to 'formal operations', like those to 'LAD', amount to nothing more than name-
dropping. If all Krashen means is that adolescents and adults have certain cognitive
abilities that children lack—such as 'meta-awareness' of grammar—there is no need
to drag Piaget into the picture. And since Piaget" s theories of language acquisition are
in conflict with Chomsky's, so that if Piaget is right Chomsky is wrong,28 Krashen
would do well to get rid of one or the other of them, if he is not willing to do the work
necessary to try to reconcile the two.
Assuming that ElMnd's analysis is correct, we still have an explanation only for
certain kinds of affective barriers to acquisition, and then only for adolescents. If we
follow Krashen's exegesis, adults should be superior to adolescents as acquirers, since
most of us do ultimately overcome the unpleasant consequences of undergoing formal
operations' at puberty. Or is Krashen arguing that the average adult, even though he no
longer suffers from the self-consciousness, vulnerability, loss of self-esteem, etc., that

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 93

plagued him as a teenager, still has these feelings unconsciously to a degree sufficient
to hinder language acquisition? If so—and of course Krashen offers no such
argument—we still face the original problem of why pre-pubescent children are
immune to the Filter, why Oedipal conflicts and sibling rivalry don't affect language
acquisition. And we still do not have an explanation of such affective variables as
motivation. In short, Krashen gives no explanation of the development of the Affective
Filter, and no account of why it develops only with puberty."
There are other problems. Take 'P', for example, the Chinese woman with almost
native-like competence in English. Evidently we are to believe that F s 'non-
acquisition' of certain rules, such as third person singular -s, is owing to the Affective
Filter. What could this mean? What is F s problem? Low motivation? Anxiety? Lack
of self-confidence? These seem odd characteristics to predicate of a person who took a
B. A. in linguistics at an American university even though she started learning English
only at the age of twenty. Perhaps she has an unconscious dislike of certain
morphemes? Once we move from the general to the particular, once we try to put a
little content, a little specificity into the hypothesis, we run into difficulty.
And just what could the Affective Filter be? What kind of mental structure could
determine all of the following: '(1) which target models the learner will select; (2)
which parts of the language will be attended to first;- (3) when language acquisition
efforts should cease; and (4) how fast a learner can acquire the language' (Dulay, Burt,
and Krashen 1982:46)? If we are to take the concept of a Filter seriously, we will have
to assume that the Filter filters; that is, that its effect is a negative one. For instance, if a
learner selects certain target models, it will be by (unconsciously) filtering out the
input from other models. Similarly, if a learner somehow selects which parts of the
language will be attended to first, it is by unconsciouslyfilteringout the other parts. In
the first case, as long as there is enough input of the right sort from the unfiltered
model(s), theFilter should not prevent'acquisition' of native-like competence; itneed
not even delay' acquisition'. In the second case, on the other hand, the Filter will have
to adjust itself continually to let in previously filtered-out input, if'acquisition' is to
progress. (The Affective Filter Hypothesis apparently permits the Filter to filter out
input that had been previously let in. In other words,' attending to' various' parts of the
language' need not involve a progressive'lowering" of the Filter, but rather a shifting of
the holes, as it were.) One problem with this negative interpretation—which is, I
repeat, the only possible interpretation of a 'filter1—can be seen in the case of
motivation. Under ideal circumstances, the Filter does not function. As far as
motivation goes, ideal circumstances would mean the highest possible motivation to
learn the language. Anything less than this— say, simply a solid respect for the culture
of the speakers of the target language, a healthy interest in the language, an absolute
absence of any prejudices or misconceptions about that culture— should still result in
somefilteringof input This seems counter-intuitive to me, and it certainly seems to
run counter to the standard interpretation of the motivation studies. The other
possibility is equally undesirable—that the Filter is completely 'down' under neutral
motivation conditions, so that there is no difference in acquisition ability between the
highly motivated and the merely not unmotivated.
A more serious difficulty with the idea of an Affective Filter is suggested by (2)
above: the Filter determines' which parts of the language will be attended tofirst".Two
questions (at least) come to mind: What does 'part of a language' mean? How does an
affective filter recognize different'parts of a language'? Once again, Krashen is of no
help, and I can only speculate as to what he intends by the construction of such a filter.

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 94 KRASHEN'S MONITOR AND OCCAM'S RAZOR

The notion 'part of a language', insofar as it has any meaning at all, presupposes some
sort of grammatical theory that the Filter must have access to. Even in the most
extremely unlikely, simple-minded interpretation of (2)—say, a morbid fear of
adjectives—the Filter must' know" what an adjective is, and such knowledge can by no
stretch of the imagination be called 'affective'. Any such grammatical knowledge
could only be a part either of LAD (UG) or of the acquired competence produced
through the interaction of UG and input Thus the model of acquisition outlined in A
and B above seems to make no sense.
Note that input filters have also been proposed in first language acquisition research
(for discussion see White 1982:43, and the references mentioned therein). But unlike
Krashen's Affective Filter, these filters are either connected in some way with the
child's grammar (e.g., Roeper 1978), or they are filters on parts of utterances (e.g.,
Slobin's 'processing strategies' (1973) or Newport et al. (1977)); they are not
affective filters. (On the other hand, they are filters; that is, the child is not, on these
accounts, a superior acquirer to the adult because he is filter-free.30) Thus, whether
correct or not, such accounts of child acquisition are not subject to the same sort of
criticism that Krashen's Affect Filter is subject to.
The Affective Filter on input could conceivably delay acquisition (but see note 30),
but could it 'determine . . . when language acquisition efforts should cease'? It could
presumably determine that acquisition efforts should cease; but in the absence of
grammatical knowledge, how could it actually make the efforts cease? As far as I can
see, it could do this only by shutting out all input whatever. If it filters selectively,
letting in new data, it must have a basis for selection; it must know what is new and
what isn't, and this presupposes access to the output of the LAD. If it doesn't have
such access, and if it doesn't filter out all input, then some new input is going to sneak in
through the Filter, and is going to be acquired willy-nilly.
Once again, it is uncontroversial to claim that affect affects adult acquisition of a
second language; most people would accept the claim that, ceteris paribus, an
unmotivated learner will acquire less than a motivated one, a nervous learner less than
a relaxed one, a self-hating learner less than a self-respecting one. But this by no means
justifies a theory postulating an Affective Filter the growth of which and the function of
which are not explained, and for the existence of which there is no evidence. On the
contrary, given the lack of compelling evidence, and given the incoherence of the
construct itself, Occam's Razor requires that we reject the Affective Filter
Hypothesis.
We have seen that each of Krashen' sfivehypotheses is marked by serious flaws:
undefined or ill-defined terms, unmotivated constructs, lack of empirical content and
thus of falsifiability, lack of explanatory power. His second language acquisition
theory is not a coherent theory, it is indeed incoherent to the point that it seems
inappropriate to apply the word 'theory* to it It is not that I think Krashen's ideas are
wrong, although of course I do strongly disagree with some of his ideas. Indeed,
although there is much that he seems not to understand about language and language
acquisition, I think he is often right on the important questions: I agree with him that
most language learning is unconscious, that comprehensible input is vital for learning
and that a teacher's most important job is to provide that input, that affective barriers
can prevent successful acquisition of a second language and that a teacher has the duty
to try to lower those barriers wherever possible. But then, does anybody disagree?
It is to Krashen's credit that he has tried to go beyond these observations and to
unify them into a single theory. The manner in which he has gone about trying to do

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 95

this, however, is not at all to his credit A theorist of second language acquisition—or
of anything else, for that matter—should define his terms with at least a minimal
degree of rigor, and should use them consistently. Krashen uses such terms as
acquisition, learning, structure, or LAD either without defining them at all or else
without holding his definition constant A theorist should present evidence for his
theory, and should take into account possible counter-examples, although he is not
bound to deal with all counter-evidence. Krashen often makes dogmatic statements
without evidential support (e.g., that 'learning' does not become 'acquisition' or that
'acquisition' always requires large amounts of time and input data), and glosses over
or distorts evidence to suit his purposes (e.g., his use of the data of Newport et ah on
CS, or his explaining away the SLOPE test as a Monitor-free instrument).31
Perhaps most importantly, a theory of second language acquisition must include
some sort of linguistic theory if it is to have any value as a theory. Aside from throwing
out a reference to 'LAD' from time to time, Krashen has no such theory, and this is
what makes him unable to interpret the morpheme studies, or to present a coherent
account of what 'i+V could meaa As Chomsky says, 'It is absolutely suicidal for a
field to define itself the way psychology of language almost invariably does, as dealing
with processes but not with the structures that might enter into them, or to deal with the
observed stages of growth and development, but not with the systems that underlie
them' (Chomsky 1982:69).
It may be objected that a bad theory is better than no theory at all, and that it is
unreasonable to demand of Krashen or anyone else that he give us a full-blown theory
of grammar before presenting a theory of second language acquisition. To this I would
reply, first, that I am not asking Krashen to give us a complete theory of grammar, but
rather simply to avail himself of some theory, because in the absence of any such
theory a theory of second language acquisition is doomed to triviality. Secondly, I
would argue that while an incorrect theory is certainly better than no theory at all,
much better in fact, a bad theory is not At best, a bad theory is a waste of time. At the
worst, it can discourage more fruitful research, and can mislead the unwary into
thinking that questions have been answered and problems solved that in fact remain
unanswered and unsolved. We do not yet have a theory of second language
acquisition. It may be the case that second language acquisition is simply too hard, too
complex a field to be susceptible of coverage by a unified theory. My suspicion, for
what if s worth, is that this is the case. Krashen's theory, in any case, brings us no
closer to answering that questioa
When the bad theory is a theory that has immediate applications, there is a further
reason to prefer no theory. In discussing the role of'learning' in teaching, Krashen
(1982:87-8) gives an illustration that he evidently often uses in lectures: he explains
the 'rule' for the English present progressive 'tense', specifically that this 'tense' has
' three meanings: (1), a current, on-going action that would soon be completed, (2), an
action that began some time ago in the past and may or may not be taking place at the
moment, and would end sometime in the future, and (3) future tense', so that a
sentence like/o/m is playing the violin is 'three ways ambiguous'. Krashen remarks
that his auditors, ESL teachers, usually react with a 'By George, that's true!' or 'I
never noticed that before!' or some such comment Krashen's point is that these people
had 'acquired' the 'rule' without 'learning' it I would like to draw a rather different
moral First of all, Krashen's analysis of the present progressive is simply wrong, and
the sentence is not ambiguous. Second, most of his listeners accepted his incorrect
analysis; after all, he is a professor of linguistics and an expert in TEFL. If I may quote

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 96 KRASHEN'S MONITOR AND OCCAM'S RAZOR

Chomsky once more, 'the willingness to rely on "experts" is a frightening aspect of

contemporary political and social life. Teachers, in particular, have a responsibility to
make sure that ideas and proposals are evaluated on their merits, and not passively
accepted on grounds of authority, real or presumed' (Chomsky 1966:31). The danger
of a bad theory of second language acquisition is that it can tempt us to overlook that
warning.
(Received May 1983)

NOTES
1 1 do not regard the writings of such people as Gattegno, Lozanov, or Curran as worthy of serious
consideration as theories of language acquisition—for that matter, they are hardly worthy of serious
consideration, period, although they all do contain some useful teaching techniques. See, e.g., Curran 1972,
1976; Gattegno 1972, 1976; Lozanov 1979. For an excellent—and damning—review of Lozanov, see
Scovel 1979.
2 The most recent and complete sources of information on Krashen's theory are his Second Language
Acquisition and Second Language Learning (1981) and Principles and Practice in Second Language
Acquisition (1982), and Dulay, Burt, and Krashen'sLanguage Two (1982). I'm not sure whether Krashen
snll uses the term 'Monitor Theory'.
3 Krashen makes other claims and other hypotheses that are not always clearly associated with one of these
five, but these five are intended to be the foundation of his theory
4 To avoid confusion I will put learn and acquire and their cognate forms in single quotes when I use them in
Krashen's sense. Note, however, that this distinction is not made in Dulay, Burt, and Krashen 1982, where
leant, learner, etc., are used in their more or less everyday senses.
5 Dulay, Burt, and Krashen (1982:54) say of their 'Organizer* that it is 'similar to Chomsky's (1975) [sic]
" language acquisition device" '. The 'Organizer" is one of the three components (along with the Monitor and
the (affective) Filter) that comprise the learner's 'internal processing system' (p 46).
6 He even goes so far as to equate it with a part of the brain: 'the input will not reach that part of the brain
responsible for language acquisition, or the language acquisition device' (Krashen 1982:31).
7 I lmow of only one study in this area, Ritchie (1978), which suggests that Japanese subjects learning
English do not violate the principle of subjacency. There certainly is some evidence that second-language
lean ers do make 'impossible' errors (e.g., *The student of chemistry is older than the one of physics; see
Baki r 1978), but it is quite possible that such errors can be explained without having to claim that U G is being
violated.
8 The past tense forms of verbs vary according to the final consonant of the verb; the gerundive forms are
identical to the past tense forms except for the final voweL The present tense of what I'm calling the super-
polite form of an adjective, e.g. /haya/, /atu/, /suzusi/, would be—rather than the normal hayai atui,
suzusii—ohayoo gozaimasu, oatuugozaimasu, osuzusyuugozaimasu. Aside fromohayoogozaimasu'it
is early1, which is the normal way of saying' good morning', these forms are seldom used, especially by men. I
have never heard them used, and I never use them; but I could if the need arose. It will be seen that a rule
cannot be generalized from ohayoo gozaimasu.
I should make it clear that I am not a very fluent speaker of Japanese. More specifically, I am not claiming
that I always use the past tense (of those verbs in my lexicon) correctly. What I am claiming is that I have
acquired the rule for forming the past tense, and that whenever I use a verb in the past tense I do it right The
'rule' of past tense use in Japanese, as in English, is really a set of rules, some grammatical and some
pragmatic (which, incidentally, makes it all the harder to interpret the morpheme acquisition studies).
9 Since Krashen is comparing 'acquisition' with 'learning', and since no one 'learns' a language, I assume
that he is talking about the acquisition of a rule or structure and not of a language—that is certainly the
implication of the comparison he makes in Principles and Practice, pp. 113-4.
10 Actually, Krashen does admit that'learning' can precede'acquisition', but he denies that it can become
' acquisition'. When' learning' precedes' acquisition', if s really a form of fudging. See e. g., Krashen's mind-
boggling explanation of this phenomenon in Second Language Acquisition, pp. 117-8.
11 He does once say (Dulay, Burt, andKrashenl982:59),'Wheneverconscious linguistic processingtaies
place, the learner is said to be using the monitor. Similarly, when a learner performs a drill that requires

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R, GREGG 97

conscious attention to linguistic form, or when he memorizes a dialogue,... the monitor is being used'; but
this is evidently a careless slip.
12 This, I found out later, is a local (Matsuyama) dialectal form. The negative ending -na is formally an
adjective (I 'learned" this from a textbook). Thus I could generalize from, say, Mary dya nakaroo = Mary
dya nai daroo 'if s not Mary, is it?' to takakaroo = takai daroo 'if s tall, isn't it?' e t c This being a non-
standard form, I would never use it, which means I could never 'Monitor1 it.
13 These are the causative, passive, and passive-causative forms of/yob/'calP; 'is doing* and'knows' (the
difference being one of consonant length); and Mr/Ms Tanaka and (Mr) Tanaka (difference in status),
respectively.
14 Since Krashen makes so much of communication—he even goes so far as to claim that communication is
' what language was designed for5 ( 1 9 8 1 1 1 ) — it should be noted that he doesn't seem to have any idea of what
communication is. In what sense can filling in blanks on a test—whatever kind of test—be called
communication? In what sense can a mother*s baby talk be called communication (1981:102—see also the
dialogue on p. 119)? In what sense can writing a description of a cartoon in an English class be called
communication? To ask the questions is to answer them; Krashen has deprived the word of any useful
meaning. And as for what language is 'designed for*, it really is about time to call a halt to this kind of
nonsense, at least in the pages of learned journals; Chomsky and others have criticized this view over and over
again ad nauseam (see e.g., Chomsky 1975, pp. 56ff, Chomsky 1980, pp. 229-30; Lightfoot 1982,
pp. 31-2), and their criticisms have not been answered.
15 Krashen inadvertently gives an example himself (1982:87-8), when he presents what he calls the 'rule'
for the present progressive 'tense'. The 'rule' is hopelessly inadequate as a description of the present
progressive, but it could in many cases work to produce the right form at the right time.
16 But ct his explanation of'learning' preceding 'acquisition' (1981:117-18), which suggests a 'single-
stream' interpretation.
17 E.g., 1982:23,51, 111, 112, 113, 140, 159; 1981:35, 1979:162.
18 Note that the Natural Order Hypothesis claims that for every language there is a set of structures, and that
these structures are' acquired* in a set order. Assuming this to be the case, and given that Krashen's definition
of'structure' includes (at the very least) grammatical morphemes, Krashen cannot fob off on to the LAD the
explanation for this order. UG makes claims about what is a possible grammar, hence what is a possible rule.
It includes rules and principles that underlie all human languages; but it says nothing relevant about -s.
19 This is one of the fundamental criticisms made of Piagef s stages of cognitive development See e.g., the
comments by Mehler and by Fodor in Piattelli-Palmarini 1980 (pp. 350 and 143-8 respectively) and by
Fodor 1974, pp. 87-95.
20 Compare, say, ' i + l , a rule of the second language' (1982:27), with 'a comparison between i and
i+V (1982:28). In the latter case ' i + l ' must be the learner's competence, or else it is a category mistake.
21 This is implied e.g., in Krashen's use of the word'review* (1982:24); 'Roughly-tuned input provides [a
child] built-in review*.
22 For a cogent criticism of MacNamara 1972 and of this kind of explanation generally, see Atkinson 1982.
23 The CS and AS corpora are based on the speech of the same mothers with their children and with the
researchers, respectively. The A S data may of course be distorted by the situation, but that is irrelevant here.
24 Note that Krashen is here extending the Input Hypothesis to include first language acquisition.
25 It should be kept in mind, however, that there are bad teachers and insensitive or inept native-speakers.
Mrs. Plomish in Little Dorrit is representative of all too many people. And of course no one has
demonstrated that CS is universal I make this (to me obvious) point only because Krashen seems to want to
deny i t * . . . it appears to be the case that we make these adjustments automatically when we focus on trying to
make ourselves understood* (1982:65).
26 The term 'affective delimiters' was used to refer to 'conscious or unconscious motives or needs of the
learner* (Dulay and Burt 1977:99). Krashen's Affective Filter (and Dulay, Burt, and Krashen's Filter) is
'subconscious'.
27 I quote from Dulay, Burt, and Krashen 1982 because Krashen doesn't bother to state the hypothesis in
Principles and Practice. Remember that learner here (and throughout Dulay, Burt, and Krashen 19 82) does
not mean 'learner* (see note 4).
28 See e.g., the debate in Piattelli-Palmarini 1980, especially the contributions by Piaget, Chomsky, and
Fodor. There is a tendency in the literature to claim both Chomsky and Piaget as the patron saints of

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 98 KRASHEN'S MONITOR A N D OCCAM'S RAZOR

psycholinguistics. In a way this is legitimate, but only to the same extent that one could claim both Adam
Smith and Karl Marx as joint founders of modern economics.
29 Actually, Krashen implies that the Affective Filter exists in children, too, although he gives no evidence
for this: cf. his reference to 'an increased affective filter* (1981:35) and to 'the strengthening of the affective
filter at about puberty" (1982:44).
30 The child may well be a superior acquirer by virtue of, rather than in spite of, these filters, and it is at least
theoretically possible that input filters would be useful to a second-language learner in the same way, by
limiting input to a manageable amount For Krashen, however, the Affective Filter is only a hindrance to
acquisition
31 Some additional examples: (1) Discussing studies of the use of self-correction, which Krashen wants to
use as evidence for Monitor use, he acknowledges that these studies could either be overestimating Monitor
use (since self-correction can also be done unconsciously) or underestimating it( since 'Monitoring' can be
done before the utterance is uttered), but then sweeps this problem aside by assuring us that the studies
'probably give us an approximation of the efficiency of conscious learning and Monitoring1 (1982:108).
(2) When comparing teaching methods, Krashen maintains that the grammar-translation method provides
'only scraps of comprehensible input* (1982:128), whereas Suggestopedia—in which 'new material is
presented in a somewhat traditional way, with the necessary grammar and translation' (1982:143, quoting
Bancroft 1978, p. 170)—'comes very close to completely matching the requirements for optimal input"
(1982:146). (3) When discussing the relationship between length of residence in the U.S. (LOR) and
proficiency, he alternately denies and assumes a causal relation between the two, depending on the needs of
his argument 'Some [subjects] may have lived in the United States many years without much
comprehensible input' (1982:40)— hence the low correlation between LOR and proficiency in this study. On
the other hand, the subjects in Stauble's (1978) study, 'since they had been in the United States for many
years,.. had had considerable comprehensible input* (1982:48)—hence the low correlation here between
LOR and proficiency supports the Affective Filter Hypothesis. Neither study in fact shows how much input
there actually was; what interests me is Krashen's ability to switch assumptions to suit his purposes.

REFERENCES
Atkinson, M. 1982. Explanations in the Study of Child Language Development Cambridge:
Cambridge University Press.
Bailey, N., C Madden, and S. D. Krashen. 1974. 'Is there a "natural sequence" in adult
second language learning?'. Language Learning, 24: 235-43.
Baker, C L. 1978. Introduction to Generative-Transformational Syntax. Englewood Cliffs,
N.J.: Prentice-HalL
Baker, C L. and J. J. McCarthy(eds.). 1981. The Logical Problem of Language Acquisition.
Cambridge, Mass.: MIT Press.
Bancroft, J. 1978. 'The Lozanov method and its American adaptations'. Modem Language
Journal, 62: 167-74.
Brown, R. 1913. A First Language: The Early Stages. Cambridge, Mass.: Harvard University
Press.
Carey, S. 1978. "The child as word learner1 in M. Halle, J. Bresnaa, andG. A. Miller(eds.).
Linguistic Theory and Psychological Reality. Cambridge, Mass.: MIT Press.
Chomsky, N. 1966. 'Linguistic theory" in J. W. Oiler, Jr., and J. C. Richards (eds.). Focus on
the Learner. Rowley, Mass.: Newbury House, 1973. (Originally published in R. G. Mead
(ed.). Northeast Conference on the Teaching of Foreign Languages. Menasha, Wis.: George
Banta, 1966.)
Chomsky, N. 1975. Reflections on Language. New York: Pantheon.
Chomsky, N. 1980. Rules and Representations. New York: Columbia University Press.
Chomsky, N. 1981.' Principles and parameters in syntactic theory* in Homstein and Lightfoot
(1981).
Chomsky, N. 1982. Noam Chomsky on the Generative Enterprise. Dordrecht, The
Netherlands: Foris.
Corran, C A. 1972. Counseling-Leaming: A Whole-Person Model for Education. Apple
River, I1L: Apple River Press.

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 KEVIN R. GREGG 99

Curran, C A. 1976. Counseling-Leaming in Second Languages. Apple River, IIL: Apple

River Press.
Dulay, H. C and M. K. Bart 1974. 'Natural sequences in child second language acquisition'.
Language Learning, 24: 37-53.
Dulay, H. and M. Burt 1977. 'Remarks on creativity in language acquisition' inM. Burt, H.
Dulay, and M. Finocchiaro (eds.). Viewpoints on English as a Second Language. New
York: Regents.
Dulay, HL, M. Burt, and S. Krashen. 1982. Language Two. New York: Oxford University
Press.
EHdnd,D. 1970. Children and Adolescents: Interpretative Essays on Jean PiageL New York:
Oxford University Press.
Fathman, A. K. 1975. 'Language background, age and the order of acquisition of English
structures' in M. Burt and H. C. Dulay (eds.). New Directions in Second Language
Learning, Teaching and Bilingual Education. Washington, D.C.: TESOL.
Fodor, J. A. 1975. The Language of Thought Cambridge, Mass.: Harvard University Press.
Gattegno, C 1972. Teaching Foreign Languages in Schools the Silent Way. New York
Educational Solutions.
Gattegno, C 1976. The Common Sense of Teaching Foreign Languages. New York:
Educational Solutions.
Hornstein, N. and D. Lightfoot (eds.). 1981. Explanation in Linguistics: The Logical
Problem of Language Acquisition. London: Longman.
Krashen, S. D. 1979. 'A response to McLaughlin, "The monitor model: some methodological
considerations" '. Language Learning, 29: 151-67.
Krashen, S. D. 1980. "The Input Hypothesis' in Georgetown University Round Table on
Language and Linguistics, 1980.
Krashen, S. D. 1981. Second Language Acquisition and Second Language Learning.
Oxford Pergamon Press.
Krashen, S. D. 1982. Principles and Practice in Second Language Acquisition. Oxford:
Pergamon Press.
Krashen, S. D. and P. Pon. 1975 'An error analysis of an advanced ESL learner*. Working
Papers on Bilingualism, 7: 125-9.
Larsen-Freeman, D. 1975. 'The acquisition of grammatical morphemes by adult ESL
students'. TESOL Quarterly, 9: 409-20.
Lightfoot, D. 1982. The Language Lottery: Toward a Biology of Grammars. Cambridge,
Mass.: MIT Press.
Lozanov, G. 1979. Suggestology and Outlines of Suggestopedy. New York: Gordon and
Breach.
MacNamara, J. 1972. "The cognitive basis of language learning in infants'. Psychological
Review, 79: 1-13.
McLaughlin, B. 1978. 'The Monitor model: some methodological considerations'. Language
Learning, 28: 309-32.
Newport, E L , H . Gleitman, and L. R. Gleitman. 1977. 'Mother, I'd rather do it myself:
some effects and non-effects of maternal speech style' in C. E. Snow and C. A. Ferguson
(eds.). Talking to Children: Language Input and Acquisition. Cambridge: Cambridge
University Press.
Piattelli-Palmarini, M. (ed). 1980. Language and Learning: The Debate between Jean
Piaget and Noam Chomsky. Cambridge, Mass.: Harvard University Press.
Porter, J. H. 1977.' A cross-sectional study of morpheme acquisition in first language learners'.
Language Learning, 27: 47-61.
Ritchie, W. C 1978. "The right roof constraint in an adult-acquired language' in W. C. Ritchie
(ed). Second Language Research Issues and Implications. New York Academic Press.
Roeper, T. 1978. 'Linguistic universals and the acquisition of gerunds' in H. Goodluck and L.
Solan (eds.). Papers in the Structure and Development of Child Language, University of
Massachusetts Occasional Papers in Linguistics, VoL 4.

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017
 100 KRASHEN'S MONITOR AND OCCAM'S RAZOR
Roeper, T. 1981. 'Introduction' to Tavakolian (1981).
Rosansky, E. J. 1976. 'Methods and morphemes in second language acquisition research'.
Language Learning, 26: 409-25.
Scovel, T. 1979. Review of Lozanov(1979). TESOL Quarterly, 13: 255-66.
Slobin, D. L197 3.' Cognitive prerequisites for the development of grammar' in C. A. Ferguson
and D. Isaac Slobin (eds.). Studies of Child Language Development. New York: Holt,
Rinehart and Winston.
Stanble, A.-M. E. 1978. 'The process of decreolization: a model for second language
development*. Language Learning, 28: 29-54.
Stevick, E. W. 1980. Teaching Languages: A Way and Ways. Rowley, Mass.: Newbury
House.
Tavakolian, S. L. (ed.). 1981. Language Acquisition and Linguistic Theory. Cambridge,
Mass.: MIT Press.
Wexler, K. and P. W. Cnlicover. 1980. Formal Principles of Language Acquisition.
Cambridge, Mass.: MTT Press.
White, L. 1981. 'The responsibility of grammatical theory to acquisition^ data', in Homstein
and Lightfoot
White, L. 1982. Grammatical Theory and Language Acquisition. Dordrecht, The
Netherlands: Foris.
Wode, IL, J. Bahns, H. Bedey, and W. Frank. 1978. 'Developmental sequence: an alternative
approach to morpheme order". Language Learning, 28: 175-85.

Downloaded from https://academic.oup.com/applij/article-abstract/5/2/79/315322

by University of Bath user
on 21 December 2017