Explicit grammar rules and

L2 acquisition
Paweł Scheffler and Marcin Cinciała

This article reports an empirical study that examines to what extent learners
can identify and understand the grammatical structures they produce when they
speak spontaneously. In the study, 20 upper-intermediate Polish learners of English
were interviewed in English by the researchers. The structures used accurately by
each learner were isolated and each of the participants was then administered
a separate test. The task in the test was first to identify correct sentences and then
to provide relevant grammar rules. The results show that in most cases, the
learners were able to identify and explain the grammar rules that accounted for
their own accurate L2 performance. In terms of second language acquisition
(S LA) theory, this means that there were few grammatical structures or categories
that the learners knew only implicitly. For teachers, the study indicates that explicit
grammar rules can, in an indirect way, contribute to S LA.

Introduction The notions of implicit and explicit L2 knowledge have generated a great deal
of research in the field of second language acquisition (SL A). Researchers
have looked, among other things, into ways of defining and measuring the
two types of knowledge (for example Ellis 2008), into the interface between
the two (for example Ellis 2005), and into the contribution of each of these
types of knowledge to language use (for example Macrory and Stone 2000).
There is also a huge literature on implicit and explicit learning processes
and the effectiveness of implicit and explicit instruction (for a review, see for
example DeKeyser 2003). In this paper, the focus is on the ways in which
explicit grammar teaching can facilitate L2 acquisition.

Implicit and explicit Ellis (2008: 6–7) defines implicit and explicit knowledge in the
L2 knowledge following way:
Implicit knowledge is intuitive, procedural, systematically variable, and
automatic and thus available for use in fluent unplanned language use.
It is not verbalizable. According to some theorists, it is only learnable
before learners reach a critical age (e.g. puberty).
Explicit knowledge is conscious, declarative, anomalous, and inconsistent
(. . .) and is only accessible through controlled processing in planned
language use. It is verbalizable, in which case it entails semi-technical or
technical metalanguage. Like any type of factual knowledge, it is
potentially learnable at any age.

E LT Journal Volume 65/1 January 2011; doi:10.1093/elt/ccq019 13

ª The Author 2010. Published by Oxford University Press; all rights reserved.
Advance Access publication April 23, 2010
 It follows from these definitions that explicit knowledge is drawn upon in
tasks not involving time constraints that preclude the planning of L2 output.
This means that explicit knowledge will not be activated in, for example,
spontaneous oral communication, in which time for planning is normally
very restricted or non-existent. An example of a typical context where explicit
knowledge could be employed by an L2 learner is an untimed multiple-
choice grammar test.
By contrast, since access to implicit knowledge is automatic, no restrictions
on its application will be imposed by the presence of time constraints: such
knowledge may thus be called upon by learners both on untimed grammar
tests and in spontaneous communication.
What follows from these definitions for language teachers is that the goal
of instruction should be implicit knowledge: in general, teachers want their
students to be able to use a foreign language fluently in spontaneous
discourse. What we would like to show in this paper is that explicit teaching
of grammar can help learners to achieve this goal.

Implicit and explicit The dominant view of L2 learning at the moment is that it is ‘a developmental
language learning process which is not subject to the learner’s conscious control’ (Willis and
Willis 2007: 18). That is, language learning is not about accumulating
grammar rules and through practice becoming more and more proficient at
deploying them. Instead, learners are supposed to follow their internal
predetermined syllabuses and to acquire L2 features when they are
developmentally ready to do so. Consequently, L2 instruction should engage
learners in various communicative activities that will trigger natural
acquisitional processes. For many SLA specialists, the most appropriate
type of communicative activity is the task (for example Ellis 2003).
Even if the above scenario is correct, it does not mean that explicit grammar
instruction cannot be used to support the process of implicit language
development. It seems that there are at least two ways in which it can
contribute.
First, knowing explicit grammar rules may lead to learners being able
to notice the structures that exemplify these rules in the input (for
example Ellis 2005). For many SLA researchers, conscious noticing of
formal L2 features is necessary for implicit language development
(for example Schmidt 1990). Further, explicit grammar knowledge may in
some cases help learners to obtain more comprehensible input: identifying,
say, the present continuous construction in spoken or written
discourse, and being able to relate it to the relevant rule, may lead to
increased comprehension. And as VanPatten (2004: 11) says, ‘increased
comprehensibility results in increased likelihood of a form being
processed in the input’.
Second, for many adolescent and adult learners, being able to understand
how a target language works, obviously including the rules underlying
their own production, is a vital part of the learning process. That this is the
case is often admitted by learners themselves: for example, Ellis (2002: 20),
in his analysis of diaries written by beginner learners of German as
a foreign language, was ‘struck by the depth of the learners’ concern to

14 Paweł Scheffler and Marcin Cinciała

 make sense of the grammar of German’. He says that their diaries were ‘full
of references to grammar—of their struggle to understand particular
rules and their sense of achievement when a rule finally ‘‘clicked’’’. This
means that if explicit grammar instruction can lead to a deeper
understanding of the target grammar, then it can contribute to the
process of learning by making learners feel more secure. If spontaneous
speech is indeed based on implicit knowledge, i.e. knowledge that is
intuitive rather than conscious, then an important question for teachers is
how much of the correct grammar their learners produce they can actually
understand.

The study The aim of the study reported on here is to investigate whether learners
Aim of the study can identify and understand the grammatical structures and rules that
underlie their spontaneous speech. In accordance with this aim, the
following research question was formulated:
Can L2 learners of English provide explicit rules for those
grammatical structures that they use accurately in spontaneous oral
performance?

Participants The participants were 20 secondary school Polish learners of English,

aged 16–18. Both the school and the learners were randomly selected for
the study. At the time of the interviews, they were attending
English instruction in their school at the upper-intermediate level. The
type of instruction they were exposed to could be described as the
weak version of communicative language teaching (Howatt 1984: 279):
this roughly means systematic and explicit treatment of English
grammar combined with a variety of practice and communicative activities.
There were 14 females and 6 males in the group. Their average exposure to
formal English instruction was 8.5 years. The average time spent by
a participant in an English-speaking country equalled one week (the
maximum was four weeks). All the subjects can thus be described as
genuine foreign (rather than second) language learners.

Procedure and data Each of the subjects was interviewed by one of the researchers. The
analysis topics that were talked about included the learners’ hobbies, their school
life, trips abroad, plans for the future, etc. Each of the learners was also asked
to describe two pictures. Any topic introduced by the learners was taken
up as well. The interviews took place in the school outside the regular
class times. They lasted between 20 and 30 minutes and were
recorded. All the data from the interviews were then transcribed
orthographically, and selected grammatical features were isolated and
analysed for accuracy. In selecting grammatical features, we focused
on those for which rules are given to learners in the process of
instruction. They included tenses (simple present, simple past, present
continuous, past continuous, and future simple), modal verbs, and
pronominal forms (for example subject, object, possessive, and relative
pronouns).

Explicit grammar rules and L2 acquisition 15

 In order to determine the accuracy of grammatical features in our learners’
performance, we employed a method called Target-Like Use (TL U). The
formula is as follows (for a discussion of this and other scoring methods
see Goldschneider and DeKeyser 2001):
n correct suppliance in obligatory contexts
TLU ¼
ðn obligatory contextsÞ + ðn suppliance in non- obligatory contextsÞ

We arbitrarily adopted 75 per cent TLU as an acceptable level of accuracy

for learner performance: it seemed to us that for an upper-intermediate
learner, a score of 75 per cent in a spontaneous production task is a clear
enough indication of L2 proficiency in a particular area. This also gave us
enough material to prepare the explicit knowledge test, in which all the
categories with scores of 75 per cent or more were used. In the test,
each learner was provided with pairs of sentences: one member of the pair
was a correct sentence used by that particular learner in the interview
(sometimes slightly modified in order to avoid any other
grammatical or lexical deviations), the other sentence was an incorrect
version invented by the researchers, in which the relevant grammatical
category was responsible for the deviation. The learners’ task was first to
identify the correct sentence in the pair and then give a rule or an
explanation in Polish that accounted for the contrast between the good and
the bad example. The rules that the subjects were supposed to provide
concerned both the form and the meaning of the grammatical
categories exemplified in the sentences. In the Appendix, for reasons of
economy, we provide one randomly selected set of test sentences. The
number of sentences that a learner received on the test depended on the
number of categories for which that learner’s TLU scores in the
interview equalled or exceeded 75 per cent. Overall, the length of the
tests varied from 6 to 12 pairs of sentences per learner.
The rules/explanations given by the learners were evaluated
independently by both of the researchers. The marking criteria for rules
generally followed those used by Green and Hecht (1992: 171–2). This
means that for a rule to count as valid, the learners did not need to
produce any technical language: informal descriptions of the phenomena
in question were also accepted. However, if technical language was
employed by a learner, then only correct terminology was accepted. This
was motivated by the fact that with incorrect metalanguage, it was
often difficult to determine whether a particular subject actually
understood the phenomenon he/she was referring to. Since judging the
validity of a rule is to some extent subjective, any discrepancies
between the researchers’ judgements that arose were arbitrated by
a third referee.

Results Table 1 contains the following data concerning each of the 20 subjects: the
success rate for the selection of correct sentences and the success rate for the
provision of correct rules. Table 2 presents the percentages for various
associations of rules and sentences. That is, the calculations are based on all
the 179 pairs of sentences used in the entire test.

16 Paweł Scheffler and Marcin Cinciała

 Subject number Success rates (%)
Selections Rules
1 83 83
2 100 71
3 100 64
4 100 100
5 100 90
6 100 100
7 100 50
8 100 43
9 100 88
10 100 83
11 90 60
12 100 92
13 100 91
14 100 75
15 100 100
16 100 100
17 100 100
18 89 56
ta b le 1 19 100 78
Success rates 20 83 75

Discussion The data in Table 1 indicate that the learners were generally successful in the
selection of correct sentences and the provision of correct rules or
explanations for their choices. Also, as both Tables 1 and 2 show, when the
learners selected a correct sentence, generally they also had the relevant
explicit rule: that was the case in 145 of 179 test items, i.e. in 81 per cent of the
cases. When the learners possessed a correct rule, the selection of a correct
sentence was virtually guaranteed: there was only one case (0.6 per cent) in
which this did not apply. Having no rule and being able to select a good
sentence were also rare.
As has already been said, it is often important for learners to make sense of
a target grammar. Our analysis of the data in this experiment shows that
learners employ devices ranging from precise metalinguistic formulations
to L1-based explanations when they need to describe the system they are
learning. The three examples below (translated here into English) are fairly

Percentage of S I T U AT I O N S where a correct rule 81

is associated with a correct sentence
Percentage of S I T U AT I O N S where a wrong rule 15
is associated with a correct sentence
Percentage of S I T U AT I O N S where no rule is associated 1.7
with a correct sentence
Percentage of S I T U AT I O N S where a correct rule is 0.6
ta b le 2 associated with an incorrect sentence
Associations of rules and Percentage of S I T U AT I O N S where a wrong/no rule is 2
sentences associated with an incorrect sentence

Explicit grammar rules and L2 acquisition 17

 traditional textbook rules:
Test item

You should eat regularly to stay healthy. correct

You should eating regularly to stay healthy. incorrect

Rule identified by a participant:

‘Should’ is always followed by an infinitive in English.

Test item

If I had a lot of money, I would travel around the world. correct

If I have a lot of money, I would travel around the world. incorrect

Rule identified by a participant:

Second conditional: past simple—‘would’ + infinitive.

Test item

Suddenly there was a waterfall and we had to go under it. correct

Suddenly there was a waterfall and we musted go under it. incorrect

Rule identified by a participant:

The past tense of ‘must’ is ‘had’. The form ‘musted’ does not exist.

It is interesting to note how the learners handled the so-called pro-drop

phenomenon. English and Polish are in this respect in the subset-superset
relation: in English, the subject must be present, whereas in Polish, null
subjects are allowed. This means that this is a contrast that may be very
difficult (if not impossible) for a Polish learner to learn from positive
evidence alone.
Despite the apparent difficulty posed by this phenomenon, our learners
were able to deal with English subject pronouns very well: in the interview,
all of them easily exceeded the 75 per cent TL U threshold with an average
TLU score of 98.6 per cent. The types of rules that the learners offered in the
test were of two main types: first, four learners stated that an English verb or
sentence must contain a subject, which is a rule often formulated in
descriptive and pedagogical grammars. Second, six learners said that ‘if
there is no person, it is not clear who or what is being referred to’. This may
be a simple meaning-based explanation of limited validity, but it seems to
have worked for the learners very well.
Although our subjects were in general able to handle metalinguistic
terminology quite successfully, metalinguistic terms were in a few cases
responsible for the problems that the learners had with giving correct rules.
For example, one of the subjects, after identifying the correct sentence in the

18 Paweł Scheffler and Marcin Cinciała

pair below, explained the difference by saying that the noun ‘decision’ was
uncountable.
Test item

I am glad that I’ve made this decision. correct

I am glad that I’ve made these decision. incorrect

When the relevant grammatical terms were lacking, the learners very
often appealed to their knowledge of L1. That was especially common in
handling distinctions in pronominal case forms and relatively
common in the case of modal verbs. In this way, by linking English to
Polish forms, the learners were able to explain the choices they made.
For example:
Test item

My Mum took me to the first dance lesson . . . correct

My Mum took I to the first dance lesson . . . incorrect

Rule identified by a participant:

‘me’ ¼ ‘mnie’

‘I’ ¼ ‘ja’

She took ‘mnie’, not ‘ja’.

Test item

To go to a disco in Spain you must be at least 16 years old. correct

To go to a disco in Spain you can be at least 16 years old. incorrect

Rule identified by a participant:

You ‘musisz’ (¼ ‘must’) and not ‘mo_zesz’ (¼ ‘can’)
to be 16. This is necessary.

Regardless of the means employed, our learners succeeded in accounting

for the structures used in their oral performance in the vast majority of
cases. The success rates for the most frequent categories on the written test
varied, but with the exception of modal verbs, they all equalled or exceeded
60 per cent.
The results presented in Table 3 largely confirm the findings of Green and
Hecht (1992), whose study is also concerned with the provision of
explicit grammar rules by learners. In the case of Green and Hecht’s
subjects, who were German learners of English as a foreign language, the
rules with high success rates ‘were those that (1) referred to easily
recognized categories; (2) could be applied mechanically; (3) were not

Explicit grammar rules and L2 acquisition 19

 dependent on large contexts’ (ibid.: 179). The pedagogical recommendation
that Green and Hecht make is that if grammar is to be taught explicitly, then
it is rules like these that should be the focus of instruction in foreign
language teaching.

Number Selections Rules

of test (%) (%)
items
Relatives 12 92 92
Second conditional 13 100 92
Object pronouns 20 100 90
Demonstrative pronouns 18 100 89
ta b le 3 Simple present 14 100 86
Success rates for the Simple past 10 80 70
most frequent Subject pronouns 20 100 60
grammatical categories Possessive pronouns 19 100 74
on the test Modal verbs 19 100 58
A similar recommendation can be made on the basis of the present results.
If we look at the top four categories, all of them meet at least one of Green
and Hecht’s criteria. In the case of rules formulated for relative pronouns, all
the criteria seem to be met:
Test item

I know a lot of people who speak English very well. correct

I know a lot of people which speak English very well. incorrect

Rule identified by a participant:

‘who’ is used for people, ‘which’ is used for things.

Typical second conditional rules, as shown by one of the examples above,

can also be applied mechanically, and object and demonstrative
pronouns can be described as ‘easily recognized categories’ that do not
depend on ‘large contexts’. As has already been said, object pronouns were
handled mainly through references to Polish case forms. The rules for
demonstrative pronouns normally involved references to the category of
number:
Test item

Paris was really nice, I would like to go there this year too. correct

Paris was really nice, I would like to go there these year too. incorrect

Rule identified by a participant:

We use ‘this’ when it refers to the singular (‘year’) and we would use
‘these’ with the plural (e.g. ‘years’).

20 Paweł Scheffler and Marcin Cinciała

Further, some of the rules concerning the use of tenses were formulated in
such a way that they could be applied mechanically, i.e. they referred to
specific lexical items. For example:
Test item

Computers are great because they usually do what you want correct
them to do.

Computers are great because they are usually doing what you incorrect
want them to do.
Rule identified by a participant:

‘Usually’ is used with the present simple.

However, as the following simple past example shows, rules in this area of
grammar were often of a more general nature:
Test item

Two years ago the Red Cross has sent me . . . incorrect

Two years ago the Red Cross sent me . . . correct

Rule identified by a participant:

The simple past tense, time of the action is given.

Rules like the one above concerning the present simple tense can, in our
view, be described as ‘informal pedagogical formulations of limited validity
and scope’ (Westney 1994: 77), i.e. they can be described as rules of thumb.
While in a few cases, they enabled our learners to make appropriate
selections, it seems that, in general, their applicability is rather limited: they
seem to be useful at lower levels of instruction, where only the most basic
differences in meaning are important. So, for example, as far as tenses are
concerned, linking frequency adverbs like ‘usually’ and ‘always’ with the
present simple and prepositional expressions introduced by ‘for’ with the
present perfect may work for beginners, but for more advanced learners, it is
certainly much too crude. In the case of advanced learners, what is needed is
probably an explanation rather than a mechanical rule and a lot of exposure
to the relevant structures in extended contexts.
Summing up, the discussion of the data shows that our subjects used three
types of devices to account for the grammar they knew: precise
metalinguistic formulations, rules of thumb, and L1-based explanations.
It seems to us, then, that all of them deserve a place in the teaching process.
While simple metalinguistic descriptions are commonplace in modern
foreign language teaching materials, the use of L1–L2 correspondences is
not, even in locally produced coursebooks. In our view, the contribution of
the mother tongue to the development of explicit knowledge could be much
greater than it is at the moment: as Swan (2007: 293) says, students
‘effectively know’ various aspects of English grammar before they start

Explicit grammar rules and L2 acquisition 21

 learning it, and ‘the existence of cross-language equivalents can
substantially reduce the teaching needed in some areas (. . .)’.

Conclusion In the study reported in this paper, the learners first participated in
a spontaneous oral interview and then they were tested on the rules
corresponding to the grammatical structures that they had used accurately
in the interview. According to the framework of Ellis (2008), this means that
they were asked to perform three tasks: first, to produce spontaneous output
based on implicit knowledge; second, to identify correct structures from this
output with the help of implicit and/or explicit knowledge; and third, to
analyse these structures appealing to explicit knowledge. As for tasks two
and three, most of the learners were able to identify correct structures in the
input and then to explain the grammatical choices they had made in the
interview. That is, there were few grammatical structures or categories that
the learners knew only implicitly. The answer to the research question
addressed in this paper is, therefore, affirmative.
The recommendation we would like to make on the basis of our results is
that language teachers should invest some classroom time in explicit
grammar instruction: our data show that at least some grammatical
phenomena can be successfully taught as simple rules. The success here
refers to benefits of two kinds. First, simple metalinguistic descriptions can
be meaningful to learners, in that learners should be able to notice in the
input some of the formal L2 features they have been taught and use them to
make sense of this input, i.e. to make it more comprehensible. It also seems
that many learners can use simple metalinguistic descriptions accurately
when discussing English grammar. Second, since grammar rules foster the
understanding of one’s grammatical output and, in this way, contribute to
a sense of security, confidence, and achievement on the part of the learners,
they contribute to the learning process in general.
Final revised version received January 2010

References testing’. International Journal of Applied Linguistics

DeKeyser, R. M. 2003. ‘Implicit and explicit learning’ 18/1: 4–22.
in C. J. Doughty and M. H. Long (eds.). The Goldschneider, J. M. and R. M. DeKeyser. 2001.
Handbook of Second Language Acquisition. ‘Explaining the ‘‘natural order of L2 morpheme
Oxford: Blackwell. acquisition’’ in English: a meta-analysis of multiple
Ellis, N. 2005. ‘At the interface: dynamic interactions determinants’. Language Learning 51/1: 1–50.
of explicit and implicit language knowledge’. Studies Green, P. S. and K. Hecht. 1992. ‘Implicit and explicit
in Second Language Acquisition 27/2: 305–52. grammar: an empirical study’. Applied Linguistics
Ellis, R. 2002. ‘The place of grammar instruction in 13/2: 168–84.
the second/foreign language curriculum’ in Howatt, A. P. R. 1984. A History of English Language
E. Hinkel and S. Fotos (eds.). New Perspectives on Teaching. Oxford: Oxford University Press.
Grammar Teaching in Second Language Classrooms. Macrory, G. and V. Stone. 2000. ‘Pupil progress in
Mahwah, NJ: Lawrence Erlbaum. the acquisition of the present perfect tense in French:
Ellis, R. 2003. Task-Based Language Learning and the relationship between knowledge and use’.
Teaching. Oxford: Oxford University Press. Language Teaching Research 4/1: 55–82.
Ellis, R. 2008. ‘Investigating grammatical difficulty Schmidt, R. 1990. ‘The role of consciousness in
in second language learning: implications for second language learning’. Applied Linguistics 11/2:
second language acquisition research and language 129–58.

22 Paweł Scheffler and Marcin Cinciała

Swan, M. 2007. ‘Why is it all such a muddle, and interests include S L A, modern English grammar,
what is the poor teacher to do’ in M. Pawlak (ed.). and corpus linguistics. He has published in a variety
Exploring Focus on Form in Language Teaching. of journals both in Poland and abroad. He also writes
Kalisz–Poznań, Poland: Faculty of Pedagogy and language teaching materials for Polish learners of
Fine Arts. English.
VanPatten, B. 2004. ‘Input processing in Second Email: [email protected]
Language Acquisition’ in B. VanPatten (ed.).
Processing Instruction: Theory, Research and Marcin Cinciała is a PhD student at Adam
Commentary. Mahwah, NJ: Lawrence Erlbaum. Mickiewicz University, Poland, where he received
Westney, P. 1994. ‘Rules and pedagogical grammar’ his MA in Linguistics in 2007. His MA thesis
in T. Odlin (ed.). Perspectives on Pedagogical Grammar. focused on contemporary approaches to Control
Cambridge: Cambridge University Press. Theory within minimalist syntax. His current
Willis, D. and J. Willis. 2007. Doing Task-based research interests include generative grammar,
Teaching. Oxford: Oxford University Press. contrastive Polish-English syntax, and the relation
between grammar and S L A.
The authors Email: [email protected]
Paweł Scheffler is a researcher and lecturer at Adam
Mickiewicz University, Poland. His research

Appendix
Subject 4: test
sentences

(1a) When I finally got those tickets I was very happy.

(1b) When I finally got that tickets I was very happy.
(2a) The girls in the corner sleeps.
(2b) The girl in the corner is sleeping.
(3a) Last year I went to her concert in Berlin.
(3b) Last year I have gone to her concert in Berlin.
(4a) I often listen to English pop music. I think it’s a good way to learn
English.
(4b) I am often listening to English pop music. I think it’s a good way to learn
English.
(5a) Mary wants to be an actress but I think is too shy.
(5b) Mary wants to be an actress but I think she is too shy.
(6a) I don’t want to go with they.
(6b) I don’t want to go with them.
(7a) I really like my teachers.
(7b) I really like me teachers.

Explicit grammar rules and L2 acquisition 23