Chapter 1

Input Processing in Second Language

Acquisition

Bill VanPatten
University of Illinois at Chicago

Without a doubt, second language acquisition (SLA) is complex. It is complex

for at least two reasons. It involves the acquisition of a complex implicit
linguistic system consisting of lexical entries and their features and forms, an
abstract syntactic system, a phonological system, and rules on pragmatic use of
language, among other components related to language. In addition, acquisition
cannot be reduced to a single process. SLA is best conceived of as involving
multiple processes that in turn may contain subprocesses that work at every
stage of acquisition.
This chapter is concerned with only one of the processes involved in SLA,
the initial process by which learners connect grammatical forms with their
meanings as well as how they interpret the roles of nouns in relationship to
verbs. This process is termed input processing (cf., Chaudron, 1985). In earlier
work, I have discussed input processing vis à vis four principles that guide
learner attention to linguistic form in the input. Here I will review those
principles and expand on them. Before doing so, several points need
clarification. The first is that any model of input processing is not per se a model
or theory of acquisition. As mentioned previously, acquisition consists of
multiple processes. Thus, the mechanisms responsible for how learners
restructure grammars (e.g., reset parameters, regularize forms and structures)
fall outside the domain of input processing. Likewise, how learners come to be
able to produce language for communicative purposes also falls outside the
domain of input processing, as do whatever factors or mechanisms are involved
in the acquisition of fluency and accuracy in output.
A second very important point is that a model of input processing such as the
one presented here is not intended as a final state model; that is, I am not
attempting to describe an L2 parser and how it operates. L2 parsers develop over
time. The present model attempts to capture under what conditions learners may
or may not make connections between a form in the input and a meaning and the
processes they initially bring to the task of acquisition. (I take this point up in
more detail later.)
A third important point is that a focus on input does not suggest that there is
no role for output in acquisition in its more general sense as some have come to
6 PROCESSING INSTRUCTION

wrongly conclude after reading work on input processing. Both input and output
have roles in acquisition and I touch upon this later in this chapter. As a
preview, I argue that input and output play complementary roles but that we
cannot get around the basic fact that the fundamental source of linguistic data
for acquisition is the input the learner receives.
A fourth and final point before beginning is that input processing is not about
instruction or about classroom-based learning only. As a model of one process
involved in acquisition, what we observe about input processing should be true
regardless of context. It is true that research on input processing with so-called
“naturalistic” learners is scant, at best. But as Gass (1990) argued, there is no
reason to believe that the internal mechanisms learners bring to acquisition
should somehow be different based on context or language to be learned. To be
sure, insights from input processing research can be and have been used to
develop intervention techniques for instructed SLA (VanPatten, 1996; Wong,
chap. 2, this volume). But instructional developments are not equivalent to a
theory or model on which the developments draw.

FIRST, SOME DEFINITIONS

Given the plethora of terms used in SLA these days to talk about similar and
related phenomena, it would be appropriate here to discuss the term processing.
As I use it, processing refers to making a connection between form and
meaning. That is, a learner notes a form and at the same time determines its
meaning (or function). The connection to meaning may be partial or it may be
complete (for example, given the complexity of verb endings in Spanish, a
learner may “realize” that a form denotes pastness but has not grasped the
aspectual meaning also encoded in the inflection).
Processing is not the same as perception of a form or noticing. Perception of
a form refers to the acoustic signal registration that happens to all auditory
stimuli. This occurs prior to assignment of meaning and in a number of cases
something perceived may get deleted before assignment of meaning to a
sentence (see, for example, the discussion in Wolvin & Coakley, 1985).
Noticing, as I understand it, refers to any conscious registration of a form, but
not necessarily with any meaning attached to it (Schmidt, 1990). Terrell (1991),
for example, very clearly illustrates his ability to notice a form in the input but
an inability to assign any meaning (or function to it). In my terms, he noticed the
form but did not process it. Thus, processing implies that perception and
noticing have occurred, but the latter two do not necessarily imply that a form
has been processed (linked with meaning and/or function).
I should also define what I mean by intake. I use the term intake to refer to
that subset of the input that has been processed in working memory and made
available for further processing (i.e., possible incorporation into the developing
 1. INPUT PROCESSING 7

system). As we will see, intake is not just filtered data (i.e., a mere subset of the
input) but it may include data processed incorrectly (i.e., the wrong form-
meaning connection may be made). I do not use intake to refer to internalized
data. How data makes it into the developing system and the impact this has on
the grammar is the subject of what I term accommodation and restructuring and
lie outside the present discussion.
In short, processing is about making form-meaning/function connections
during real time comprehension. It is an on-line phenomenon that takes place in
working memory. I turn attention now to the “what” of input processing.

GETTING MEANING WITH THE RESOURCES YOU HAVE

I take as a point of departure the following claims: that during interaction in the
L2 (1) learners are focused primarily on the extraction of meaning from the
input (e.g., Faerch & Kasper, 1986; Krashen, 1982), (2) that learners must
somehow “notice” things in the input for acquisition to happen (Schmidt, 1990
and elsewhere), and that (3) noticing is constrained by working memory
limitations regarding the amount of information they can hold and process
during on line (or real time) computation of sentences during comprehension
(e.g., Just & Carpenter, 1992). We begin with the most basic and overarching
principle in input processing:

Principle 1. Learners process input for meaning before they

process it for form.

What this principle says is that learners are driven to look for the message or
communicative intent in the input. Although this is true of all human
communication, for the second language learner undergoing acquisition this
“push to get meaning” combined with limited resources for processing input,
means that certain elements of form will not get processed for acquisitional
purposes. This observation is consistent with a number of perspectives on both
first and second language acquisition (e.g., Faerch & Kasper, 1986; Klein, 1986;
Sharwood Smith, 1986; Peters, 1985; Wong Fillmore, 1976; among others).
That learners would process input for meaning before form would lead one to
ask, “Well, just what is it that learners are processing, especially in the early
stages of acquisition?” Again, consistent with observations in both first and
second language acquisition, the most logical place for learners to extract
meaning is in content words. Although children may come to the task of
acquisition with a universal grammar (UG) that distinguishes between lexical
and functional categories, they do not come to the task like second language
learners who already have the cognitive construct of “word” based on their first
language experience (see, for example, the discussion in Peters, 1985, on “units
8 PROCESSING INSTRUCTION

of acquisition”). In short, second language learners in particular know there are

“big words” that can help them get the meaning of what is being said to them
and their internal processors attempt to isolate these aspects of the speech stream
during comprehension. “Little words”, inflections on verbs and nouns, may be
skipped over or only partially processed and then dumped from working
memory as the processing resources in working memory are exhausted by the
efforts required to process lexical items. (Some little words and inflections may
be processed but not in isolation; that is, they may be fused with the content
word they occur with. This is called chunking and results in formulas and
routines.) The focus on processing content lexical items, then, leads us to posit a
sub-principle of Principle 1, one that we might call “The Primacy of Content
Words.”

Principle 1a. Learners process content words in the input before

anything else.

To be sure, the primacy of lexical items in learners’ search for meaning in the
input is aided by prosodic factors. In most natural languages, content lexical
items tend to receive stronger stress than noncontent items and very often, the
roots of content items receive stronger stress than any surrounding particles,
functors or inflections. Thus, learners may use prosodic cues as a means of
helping them locate and process the major “units of meaning” in a given input
string.
The primacy of processing lexical items, although helping learners get
meaning, creates a secondary problem when it comes to acquisition. Natural
languages often involve a great deal of redundancy of meaning. By this I mean
that semantic notions are often encoded more than once in either a sentence or in
discourse (across sentences). One of the clearest examples is standard English
present tense third person singular marking. A basic rule of English is that if a
sentence contains a third person singular subject, the verb “must agree” (i.e.,
repeat the same semantic and/or functional information). If we examine the
following sentences, we see how.

(1) John (third-person sing) talks (third person sing) too much.
(2) What does (third person sing) John (third person sing) do?
(3a) He (third person sing) talks (third person sing) too much.
(3b) Talk (non finite) too much.
(3c) *Talks (third person sing).

In (1)–(3a), the third person singular referent is marked twice; once by the
(pro)noun and once by the verb form. This is called redundancy. Note how only
 1. INPUT PROCESSING 9

(3a) and (3b) can be answers to (2). (3b) is fine without a noun or pronoun
because the verb is nonfinite. (3c) is ruled out because the -s on the verb cannot
stand alone in a declarative; it requires the obligatory presence of a noun or
pronoun that marks the same meaning. This is redundancy.
A fallout, then, of relying on content lexical items is that if a content lexical
item and a grammatical form both encode the same meaning, the learner’s
processing mechanisms need only rely on the lexical item for that semantic
information and not the form. In the case of third person -s, the noun or pronoun
subject is the content lexical item that carries the information that someone else
is being talked about. To be sure, not all forms are redundant nor do all forms
carry semantic meaning (see the discussion in VanPatten, 1996) but a good
many do. The point is that learners may not have to rely on grammatical markers
to get the semantic information. Thus, a second subprinciple of Principle 1
would be:

Principle 1b. Learners will tend to rely on lexical items as

opposed to grammatical form to get meaning when both encode
the same semantic information.

This is a slightly restated version of the original Principle P1b which used the
term “prefer” (instead of “tend to rely on”) and simply “semantic information”
(instead of “same semantic information”). In short, the original wording did not
capture the nature of redundancy by omitting the use of “same.” This particular
subprinciple can be called “The Lexical Preference Principle.”
To be clear, it is not the case that learners “choose” to ignore form (one
reason to change the word “prefer”) any more than it is the case that learners
“choose” to process one form over another. Like most aspects of acquisition,
learners may voluntarily pay attention to language in order to comprehend, but
processing basically happens to the learner. It is also not the case that the
processing mechanisms sweep through an input string and eliminate all
grammatical form and attend only to lexical items as some have (mis)interpreted
(e.g., DeKeyser, Salaberry, Robinson, & Harrington, 2002). To interpret this
subprinciple and the next one correctly, one must keep in mind what we mean
by process and processing. To process a form means to connect that form with
its meaning and/or its function. The position taken here is that because of the
constraints on working memory, these connections may not happen (or may
happen only under certain conditions). The learner may very well perceive the
form and notice it, but because no connection to meaning or function is made,
the form is dropped from further processing. One must keep in mind how
effortful comprehension and processing are for beginning and even intermediate
learners. Processing lexical items simply drains the resources in working
memory that would allow for making a connection to meaning for any possibly
perceived form. (Again, this may vary under conditions, a point I discuss later.)
10 PROCESSING INSTRUCTION

In previous work, I have discussed a third subprinciple for Principle 1. This

particular principle is based on the idea that not all forms are equal in terms of
the meaning they express. Specifically, some forms carry meaning and some do
not. Adjective agreement in languages such as the Romance languages does not
carry semantic information when the object of agreement is a noun expressing
an inanimate object, as in the following example:

(4) La casa blanca fue pintada.

The white house was painted.

In this sentence, the head noun casa is of feminine grammatical gender. Gender
may have originated as a semantic notion (e.g., biologically or socially
determined maleness/femaleness), but as a grammatical feature is devoid of
semantic information: inanimate nouns such as “house” and “rock” are not
biologically or socially constructed with gender. The grammatical form as an
inflection, then, is meaningless in terms of semantic information it carries.
Previously we saw that third person -s carries a semantic notion (another
person that is neither the speaker nor the person the speaker is addressing).
However, it is redundant and learners may not make the particular form-
meaning connection in the input right away. Alternately, the progressive marker
-ing is not redundant; it tends to be the sole marker of progressive aspect (a
semantic notion) in English as in sentences such as “What are you doing?” “I’m
baking a cake. Why are you asking?” There are no lexical items that carry the
same information. In the third subprinciple to Principle 1 I suggested that
learners would attend to more meaningful items before less meaningful items
with the scale being this:
• items with semantic information are meaningful that are not redundant with
lexical items bearing the same semantic information;
• items with semantic information that are redundant are less meaningful;
• items with no semantic information are not meaningful.
The particular subprinciple was stated as follows:

P1c. Learners prefer processing “more meaningful” morphology

before “less meaningful” morphology.

At the time, I defined “meaningfulness” as the relative communicative value a

form had, which meant the overall meaning that form contributed to sentence
comprehension. Communicative value was operationalized by an intersection of
the features [+/− semantic information] and [+/− redundancy] as described
above (see Lee, 1987a, for an excellent study that demonstrates how learners
skip items of low communicative value during processing). However, this
 1. INPUT PROCESSING 11

subprinciple is misleading as stated. First, not all grammatical form is

morphological in the sense that it is inflected on nouns and verbs. As described
soon, form can also exist at the sentence level in terms of word order. In
addition, the subprinciple failed to separate redundant form with meaning and
form (whether redundant or not) that does not carry meaning. (Meaning, again,
refers to a semantic notion.) Here I refine the subprinciple to be two different
ones:

New P1c. The Preference for Nonredundancy Principle.

Learners are more likely to process nonredundant meaningful
grammatical form before they process redundant meaningful
forms.
New P1d. The Meaning-Before-Nonmeaning Principle.
Learners are more likely to process meaningful grammatical
forms before nonmeaninful forms irrespective of redundancy.

The question arises, once one discusses that forms have differential levels of
“meaningfulness” (or what I used to term communicative value), “If learners are
driven to get meaning, how is it that they process the form that does not
contribute much to overall sentence meaning?” Recalling that processing form
(i.e., linking it with meaning or function) is constrained by the limits of working
memory and the taxing nature of comprehension for early and intermediate stage
learners, I developed a second principle that said:

P2. For learners to process form that is not meaningful, they

must be able to process informational or communicative content
at no or little cost to attention.

This principle fell out of the theorem in SLA that increased comprehensibility of
input may result in increased acquisition (e.g., Blau, 1990; Hatch, 1983; Long,
1985); in this case, however, we would restate the matter as increased
comprehensibility results in increased likelihood of a form being processed in
the input. However, as stated, the principle fails to capture that there is also the
matter of redundant but meaningful forms that may be difficult to process (a
fallout of P1c), What is more, the principle is actually not a principle in its own
right, but a derivative of the preceding subprinciples. For these two reasons, the
principle is restated here as a new subprinciple:

New Principle P1e. The Availabity of Resources Principle. For

learners to process either redundant meaningful grammatical
forms or nonmeaningful forms, the processing of overall
sentential meaning must not drain available processing
resources.
12 PROCESSING INSTRUCTION

Just what provides for the availability of processing resources? One obvious
answer is proficiency level and the nature of learners’ ability to access lexical
items they have already incorporated into their developing linguistic systems. If
lexical retrieval is not laborious during comprehension, in principle resources
are not as strained as they would be at an earlier stage. At the same time,
familiarity of words in the input would make a difference. Simply put, fewer
unknown words in the input string requires less processing of novel lexical items
that in turn means releasing resources to process grammatical form.

One Role for Output

Less obvious to some is that task demands also affect processing resources. It is
here that one of the roles of output in acquisition becomes evident. Some have
argued that negotiated interaction facilitates acquisition. Gass (1997), for
example, has said “negotiation is a facilitator of learning; it is one means but not
the only means of drawing attention to areas of needed change. It is one means
by which input can become comprehensible and manageable” (pp. 131–132).
What Gass is arguing, it seems, is that interaction alters the task demands placed
on a learner during input processing. The change in task demands frees up
attentional resources allowing learners to process something they might miss
otherwise. It must be made clear that this position does not suggest that by
producing the form in question during the interaction the learner is acquiring or
has acquired the form; the position is that by interacting the learner gets crucial
data from another interlocutor (more on this later). The following example is
illustrative and was overheard in a locker room after a tennis match. “Bob” is a
native speaker of English and “Tom” is a non-native speaker with Chinese as a
first language.

Bob: So where’s Dave?

Tom: He vacation.
Bob: He’s on vacation?
Tom: Yeah. On vacation.
Bob: Lucky guy.

In this particular interaction, Bob’s clarification/confirmation request allowed

Tom to notice the use of on with vacation. Why? Bob’s second question did not
contain a new message. That is, Tom did not have to process it for its
informational content. This freed up the resources in working memory for him
to process the preposition. That Tom incorporated it subsequently into a
confirmation does not mean that he has acquired it; what it shows us is that he
has noticed it, something that Gass claims may be part of the process of
acquisition or, in this case, input processing (see also Schmidt, 1990, 1995).
 1. INPUT PROCESSING 13

Consistent with Gass’s position is one articulated by Swain (1998). She states
the following;

“I have hypothesized that, under certain circumstances, output

promotes noticing. This is important if there is a basis to the
claim that noticing a form in input must occur in order for it to be
acquired.” (p. 66)

In addition to altering the attentional demands of the task and thereby getting
important data from the input, the learner may also notice that something he or
she says is not the same as what was just heard in the input. Thus, the
immediacy of juxtaposing one’s output with another’s input may trigger noticing
that is useful for making form-meaning connections. This is one way to explain,
for example, the possible benefits of recasts (Ortega & Long, 1997) and the
interactions that surround them (see the Bob/Tom just described).

Back to Principles
In later writings (e.g., VanPatten, 1997), I reviewed research on processing that
is related to but not the same as a change in task demands. That research focuses
on the location of formal elements within a sentence. The research (e.g.,
Barcroft & VanPatten, 1997; Klein, 1986; as well as some of the literature in
cognitive psychology) strongly suggests that elements that appear in certain
positions of an utterance are more salient to learners than others, namely,
sentence initial position is more salient than sentence final position that in turn is
more salient than sentence internal or medial position. Translated into
processing grammatical form in SLA, what this observation means is that
grammatical form or linguistic elements in sentence initial position are more
likely to be processed than elements in other positions (all other processing
issues being equal). Why would this be so? Again, the explanation may reside in
processing resources. Elements at the beginning are, by definition, the first on
which available resources are applied to process an input string. If the resources
are constrained then that means the resources may be gobbled up to process that
initial item(s) and may not be available for medial items. As the learner
approaches the end of the input string (i.e., once again redirects attention to
processing the string), the resources may now be available and thus an element
in final position gets processed or has chances of being processed. Of course,
sentence length may interact with this principle. Processing something like “Is it
cold outside?” is a lot different for the early stage or “resource depletion prone”
learner than “Is it cold outside or do you think I can go out with just a shirt on?”
The principle of location, then, can be captured in a principle such as P1f:
14 PROCESSING INSTRUCTION

New Principle P1f. The Sentence Location Principle. Learners

tend to process items in sentence initial position before those in
final position and these latter in turn before those in medial
position.

To summarize so far, we have discussed one major principle and five

subprinciples that fall under the general matter of the primacy of meaning in
input processing. I list all of the principles here in their most current incarnation
for the benefit of the reader.
Principle 1. The Primacy of Meaning Principle. Learners process input for
meaning before they process it for form.

Principle 1a. The Primacy of Content Words Principle.

Learners process content words in the input before anything else.
Principle 1b. The Lexical Preference Principle. Learnes will
tend to rely on lexical items as opposed to grammatical form to
get meaning when both encode the same semantic information.
Principle 1c. The Preference for Nonredundancy Principle.
Learners are more likely to process nonredundant meaningful
grammatical form before they process redundant meaningful
forms.
Principle 1d. The Meaning-Before-Nonmeaning Principle.
Learners are more likely to process meaningful grammatical
forms before nonmeaninful forms irrespective of redundancy.
Principle le. The Availability of Resources Principle. For
learners to process either redundant meaningful grammatical
forms or nonmeaningful forms, the processing of overall
sentential meaning must not drain available processing
resources.
Principle 1f. The Sentence Location Principle. Learners tend
to process items in sentence initial position before those in final
position and those in medial position.

WHO DID WHAT TO WHOM?

Until now we have discussed grammatical form as inflections and noncontent

words (“little words”). However, if we expand the notion of form, we see that
sentences can also have form. Languages can have different word orders when it
comes to relationships of nouns to a verb. English, for example, is almost
exclusively SVO in both declaratives and questions. Languages like Spanish and
Hungarian are not so rigid, allowing for SVO, SOV, OVS and others. Thus, as a
sentence is processed, learners must assign both grammatical (e.g., subject vs.
 1. INPUT PROCESSING 15

nonsubject) and semantic (e.g., agent vs. nonagent) roles to nouns in order to get
the intended meaning of the speaker.
A good deal of research on L2 sentence processing suggests that learners
begin acquisition by tagging the first noun in an NP-V-NP sequence as the
subject or agent. We see learners of various first languages tending to interpret
(5) as (6).

(5) The cow was kicked by the horse.

(6) The cow kicked the horse.

We suspect that this misinterpretation is not due simply to L1 transfer. First, the
L2 learners tested come from various L1 backgrounds, some of which include
passives and some that don’t. Some allow for flexible surface word order (e.g.,
in addition to being SVO they may also allow OVS, for example). In one study
(Ervin-Tripp, 1974), learners of L1 English going into L2 French tended to
interpret the first noun of a passive structure as the agent even though English
and French have the exact same sentence structure for passives. Finally, we also
know from typological research that SVO and SOV are overwhelmingly the
preferred canonical patterns of languages from around the world suggesting that
the human mind may be predisposed to placing agents and subjects in a first
noun position.
This particular tendency to tag the first noun or noun phrase as the subject or
agent is called the first noun principle.

P2. The First Noun Principle. Learners tend to process the first
noun or pronoun they encounter in a sentence as the subject or
agent.

In terms of consequences for language learning, this principle may cause a delay
in the acquisition of passives, any OVS structures, case marking, among others.
L2 learners of French, for example, are known to misinterpret causative
structures (e.g., Allen, 2000; VanPatten & Wong, this volume). In French, the
causative is formed with the verb faire.

(7) Jean fait nettoyer la chambre à Marc.

John makes Mark clean the room.
(lit: John makes to clean the room to Mark.)

When asked, “Who cleans the room?” learners tend to respond that John does,
thereby interpreting the sentence as something like “John cleans the room for
16 PROCESSING INSTRUCTION

Mark.” L2 learners of Spanish are known to misinterpret object pronouns when

the subject is omitted or placed after the verb (e.g., Lee, 1987b; VanPatten,
1984). Thus, they misinterpret sentences such as the following:

(8) Lo conoce María.

him—OBJ knows Mary—SUBJ
incorrect interpretation: He knows Mary.
correct interpretation: Mary knows him.

(9) Se levanta temprano.

self-REFL raises early.
incorrect interpretation: He (=se) gets up early.
correct interpretation: He (=self) gets up early [as opposed to someone
getting him up early].

LoCoco (1987) found that learners of German ignore case markings and
misinterpret sentences in German. Sentences such as Den Lastwagen schiebt das
Auto (‘the-OBJ truck pushes the-SUBJ car’ ‘The car pushes the truck’) are
incorrectly interpreted by learners as “The truck pushes the car.” This occurs
even after learners have formally studied case marking in German and have been
formally tested on it. (See Gass, 1989, for a study on L2 learners of English and
L2 learners of Italian.)
In short, this particular principle may have a variety of consequences in a
variety of languages. It is not just that learners may get word order wrong, it is
also that they may not process case markings for some time, will have
difficulties with the pronoun system in some languages, and so on.
However, there is also research showing that the First Noun Principle can be
attenuated in certain circumstances. For example, lexical semantics may come
into play. Our earlier example of the passive, repeated as (10) involves two
entities equally capable of performing the act of kicking. However, in (11) the
lexical semantics of the verb require that an animate being with legs be the
subject/agent of the verb ruling out a misinterpretation of who did what to
whom.

(10) The cow was kicked by the horse.

(11) The fence was kicked by the horse.

The effects of lexical semantics can be captured as a subprinciple to P2.

 1. INPUT PROCESSING 17

P2a. The Lexical Semantics Principle. Learners may rely on

lexical semantics, where possible, instead of on word order to
interpret sentences.

Another way in which the First Noun Principle may be attenuated is with event
probabilities. Event probabilities refer to the likelihood of one noun being the
subject/agent as opposed to another. Given two nouns such as man and dog and
the verb bite, it is more likely in the real world that a dog would bite a man than
the other way around. Thus, learners would likely interpret a sentence such as
(12) correctly given what they know about the real world.

(12) The farmer was kicked by the horse.

The principle that captures this aspect of sentence interpretation is the following:

P2b. The Event Probabilities Principle. Learners may rely on

event probabilities, where possible, instead of word order to
interpret sentences.

In previous work I combined P2a and b into one principle but because lexical
semantics and event probabilities are really two different semantic concepts, I
have come to realize it is best to discuss them separately. At the same, time we
have conducted research based on contextual cues in L1 sentence processing
that show that contextual cues in L2 sentence processing may also attenuate the
First Noun Principle. In this research (VanPatten & Houston, 1998), we gave
learners sets of identical sentences with the only difference being that one set
had contextual information that would push them away from interpreting the
targeted clause the wrong way. For example, we gave them sentences like (13)
and (14). The targeted clause is underlined.

(13) Gloria contó a su amiga que la atacó Ramón en casa.

Gloria told her friends that her-OBJ attacked Ramon-SUBJ at home.
(14) Roberto está en el hospital porque lo atacó María con un cuchillo.
Robert is in the hospital because him-OBJ attacked Mary-SUBJ with a
knife.

In (14) “Robert is in the hospital” suggests he is there because something has

happened to him. This would lead to more interpretations of Mary attacking him
with a knife rather than the other way around; that is, “Robert is in the hospital
because he attacked Mary with a knife” makes little sense. In (13) the preceding
18 PROCESSING INSTRUCTION

phrase “Gloria told her friends” does not constrain interpretation; the following
phrase could equally be that she attacked Ramon or that Ramon attacked her.
Our results were overwhelmingly clear. Participants’ reliance on the First Noun
Principle was significantly less for the sentences with constraining context and
some participants showed no reliance at all on the First Noun Principle when a
constraining context was present. I have thus added another subprinciple to P2.

P2c. The Contextual Constraint Principle. Learners may rely less

on the First Noun Principle if preceding context constrains the
possible interpretation of a clause or sentence.

The reader will note the use of the term preceding. We have yet to research any
effects that a post target context may have on sentence interpretation. For
example, if we took (14) and reversed the order of presentation, we would get
something like (15). Again, the targeted clause is underlined.

(15) A María la atacó Ramón y por eso ella está en el hospital.

Mary-OBJ her-OBJ attacked Ramon-SUBJ and that is why she is in the
hospital.

We cannot predict based on our results that learners would necessarily also
interpret the first part of the sentence correctly because there is context present.
In this case, given aspects of working memory limitations, it is open to empirical
investigation whether learners would have the capacity to backtrack and
reprocess what they heard in the first part of the sentence, and under what
conditions they can.
To summarize this section, I have outlined the principles that account for
basic sentence interpretation (who did what to whom). These consist of a First
Noun Principle and three subprinciples that account for its possible attenuation.
The principles are all listed below:
P2. The First Noun Principle. Learners tend to process the first noun or
pronoun they encounter in a sentence as the subject/agent.

P2a. The Lexical Semantics Principle. Learners may rely on

lexical semantics, where possible, instead of word order to
interpret sentences.
P2b. The Event Probabilities Principle. Learners may rely on
event probabilities, where possible, instead of word order to
interpret sentences.
P2c. The Contextual Constraint Principle. Learners may rely
less on the First Noun Principle if preceding context constrains
the possible interpretation of a clause or sentence.
 1. INPUT PROCESSING 19

OTHER MATTERS

In this section, I examine and discuss a number of issues related to input

processing that have either arisen in discussions with colleagues or have arisen
in scholarship by others.

Looking at More Than one Principle

I should clarify here two aspects of input processing that merit attention. First,
the principles do not operate in isolation; sometimes (if not often) several may
act together or one may take precedence over the other. It should be clear, for
example, that The Availability of Resources Principle allows for learners to
process something they might normally miss if the task demands alter the
demands placed on processing resources. As the Bob and Tom episode clearly
suggests, interactional aspects of language may free up resources when no new
information is conveyed. Thus, learners may process something in the input they
had missed before. (Remember that process here is not equal to acquisition; we
cannot say that just because something is processed that it becomes part of the
implicit L2 linguistic system.)
In languages such as Spanish, we have evidence that L2 learners may skip
over a non-redundant case marker that is in initial position because the First
Noun Principle is such a powerful guide for sentence interpretation. Learners
hearing a sentence like A María no la ha visto Juan por muchos años ‘John has
not seen Mary for many years’ often miss the object marker a (which is not
redundant in this particular case; the clitic pronoun la is). Even the Sentence
Location Principle does not help them to process the case marker (the marker a
is in initial position). Instead, they are guided completely by the First Noun
Principle (VanPatten, 1984).
In some cases, principles “collude” to delay acquisition. The subjunctive in
French, for example, is hindered by at least two processing problems; The
Lexical Preference Principle (mood is indicated in the matrix clause by a phrase
rendering the subjunctive redundant in noun clauses) and The Sentence Location
Principle (it tends to occur in medial position in a sentence).
The point here is that one must look at a variety of factors that influence
processing rather than at one single principle or problem to determine why a
form may be difficult to process in the input. Obviously, such things as
frequency in the input ultimately come to bear on acquisition, but we also know
that frequency cannot be the only factor (e.g., various papers in N.Ellis, 2002;
Larsen-Freeman & Long, 1991).
20 PROCESSING INSTRUCTION

The Developmental Nature of Processing

Some have taken the model of input processing to be similar to an L1 parsing
model and thus should follow the same constraints in terms of construction and
theory (e.g., DeKeyser, Salaberry, Robinson, & Harrington, 2002). They have
argued that the input processing model makes use of an outdated model of
attention as well as a parser that does not consider the standard accounts in L1
parsing, namely, that parsing is structure based. The present model of input
processing clearly has a strong meaning-based foundation as Principle 1 and all
the subprinciples within it suggest. At first glance, the criticism that this model
looks far different from any L1 model of sentence processing seems to be true.
However, my argument is that we cannot import wholesale models from L1
studies without concern whether such wholesale importation is appropriate for
SLA (we learned this from behaviorism). Let’s take parsing first.
That L1 parsing relies on structure precedence is predicated upon ambiguity
resolution in L1 sentence processing (e.g., Clifton, Frazier, & Rayner, 1994;
Cuetos, Mitchell, & Corley, 1996). These models attempt to account for how
native speakers get tripped up and resolve processing problems such as that
presented by the verb raced in the horse raced past the barn fell. The problem
here is that the verb can either be a main verb or part of a reduced passive in a
relative clause. Other parsing ambiguities in L1 studies include whether people
use early or late closure in processing sentences such as Mary bought a present
for the nephew of the man who speaks French. When asked “Who speaks
French?” the answer could be either the nephew or the man and the response
depends on whether one has a preference for early closure or late closure. One
final example of a parsing issue occurs in languages like Spanish and Italian that
allow null-subjects. The Spanish sentence Vi a Juan cuando [pro] iba a la playa
is ambiguous in that it could mean “I saw John when I was walking to the
beach” or “I saw John while he was walking on the beach.” (The verb iba can be
used with either first person or third person singular.)
Although important matters for L1 parsing models, such issues hardly speak
to the basic L2 question of “What forms get processed under what conditions?”
In addition, developing L2 learners do not have intact L2 parsers like L1
speakers. L2 learners may eventually wind up with a parsing mechanism that
can be described and tested using L1 models but this does not mean they begin
with those parsers. Child L1 acquirers certainly don’t (e.g., Crain, Ni, &
Conway, 1994). Likewise, L2 learners must build L2 parsers. If we briefly
examine the research of Fernández (1999), for example, we see one of the
limitations of the application of parsing research to SLA. Fernández investigated
native, early non-native and advanced non-native interpretations of sentences
with ambiguous prepositional phrases such as Roxanne read the review of the
play that was written by Diane’s friend. When asked the question What was
written by Diane’s friend? The review or the play? native speakers of English
tend (about 60%) to answer the play, revealing what is called a “late closure
 1. INPUT PROCESSING 21

strategy.” Native Spanish speakers, with the exact same sentence in Spanish
(literally, word for word) respond (again, about 60%) the review revealing a
preference for “early closure.” What she found was that the learners tended to
use the L1 preference in interpreting the L2 sentences. She answers her main
research question—Do adult learners process L2 linguistic input in the way
monolingual speakers of the target language do?—“No, they don’t.” The
question for us in this chapter is “What does this have to do with making form-
meaning connections?” Certainly nothing regarding the initial processing of
inflections, functors, and other surface features of language and certainly not the
initial processing of basic syntax. It does tell us that learners have L1
preferences for interpretation of ambiguous sentences, something that may or
may not be important in the long run. But note that the preferences of native
speakers hovers around 60%. Do we want to say that the remaining 40% of
native speakers’ responses are wrong because they don’t fit the pattern? Clearly
both interpretations are possible in both languages and both sets of native
speakers use them. My point is that although I find this study highly interesting
for its basic premise (that we must study processing in order to understand
acquisition), and it is certainly a solid piece of research uncovering processing
preferences for ambiguous sentences, I am not sure where this line of parsing
research will lead in relation to the basic questions: “What do learners process in
the input and why to they process that and miss something else?” Perhaps as we
get more sophisticated in relating parsing to form-meaning connections, my
skepticism will change.
It should be noted that my skepticism does not contradict the cautious
optimism of Harrington (2001) who believes that the application of L1
processing frameworks to L2 processing contexts may prove insightful. After
reviewing three different approaches to sentence processing research (a
principle-based approach, a constraint-based approach, and a referential
approach) he concludes, “Research on ambiguity resolution has generated a
large body of [L1] empirical findings and has been a primary source of evidence
for current [L1] models of sentence comprehension. However, with more
interest directly on learning processes, the focus is starting to shift to modeling
acquisition processes in which the focus is on modeling normal learning
processes” (pp. 123–134). The current model of input processing is attempting
to do precisely this by starting with the premise that comprehension and
processing for natives cannot and is not the same process as that for beginning
non-natives who must not only comprehend but also come to discover linguistic
data in what they comprehend.
Harrington (2001) is useful here to help us understand the domain of input
processing. He discusses three components or mechanisms of sentence
comprehension: algorithms, heuristics, and representations. Algorithms are
stated as IF-THEN procedures for processing outcomes and as Harrington states
“are responsible for transforming linguistic input into meaning” (p. 98).
22 PROCESSING INSTRUCTION

Heuristics are pre-existing principles that constrain how the algorithms work.
Harrington offers the examples of the late closure principle (related to the
Fernández study described above) and Slobin’s operating principles such as
“Pay attention to the ends of words” (Slobin, 1973). Representations refer to the
linguistic and non-linguistic knowledge structures of the learner. As Harrington
puts it “[representations] are the ‘stuff’on which algorithms and heuristics
operate” (p. 99).
If we take something like our Lexical Preference Principle in L2 input
processing by which learners tend to rely on lexical items as cues to meaning
rather than their corresponding grammatical markers, this is clearly not an
algorithm or a representation. It functions much like the heuristics Harrington
refers to. But what is important to note here is that the Lexical Preference
Principle is not meant to be a final state principle for sentence comprehension
but a constraint on processing (making form-meaning connections) during
acquisition. It attempts to describe a particular strategy that learners take to the
task of comprehension that affects processing.
To be sure, the current model of input processing does not address the issue
of how learners come to process syntactic information such as co-reference and
reduced clauses vs. main verbs. (I discuss this in the section on L1 influence.)
Thus, the concept of input processing cannot be equated with the term sentence
processing as it is normally used in the psycholinguistic literature.
As for capacity, although it may be true that some models of attention no
longer describe fixed resources, the issue of capacity is simply inescapable in
current discussions of attention. In his 1998 book, Pashler reviews attention in
great detail as well as the various models that have evolved over several decades
of research. At almost every level of discussion in his book, the issue of capacity
(that is, limited resources) surfaces in the terms “gate keeping” (to describe what
gets excluded during processing), “selective attention,” “interference of one
detected stimuli with another,” “competition,” and others. What is clear from his
conclusions is that limited resources and limited capacity, as constructs within
attention, are alive and well: “Assuming these conclusions are roughly correct,
many basic questions still remain to be resolved. One question is the nature and
source of the capacity limitations that arise with perceptual overload” (Pashler
1998, p. 403).
The issue of attentional limits and capacity constraints in language processing
has been addressed by Just and Carpenter (1992). They have proposed a model
of capacity limitations that is comprehension-oriented. One of their most
fundamental claims is that language comprehension is one of the most capacity
robbing behaviors that humans engage in and that capacity affects skill in terms
of reading and listening (see also the discussion in Gathercole & Baddeley,
1993). Their claim is that capacity is not the same for everyone and that there
are individual differences in capacity that affect the on-line computations a
reader or listener makes during comprehension. In short, some people have
 1. INPUT PROCESSING 23

greater capacity than others. If a language-based L1 model suggests that

comprehension is capacity robbing, we can only begin to imagine the drain on
resources for the L2 learner attempting to comprehend in a new language and
the impact this has on input processing. In short, any model of L2 input
processing would have to consider in some way the impact of capacity issues in
working memory on what learners can do at a given point in time.
As one more point, no matter what weaknesses reside in the explanations
behind learner processing behaviors, the fact remains that the processing
principles do describe what L2 learners actually do. There is published and
unpublished research using on- and off-line measures that show that learners do
indeed prefer lexical items to grammatical markers for getting semantic
information (e.g., Lee, Cadierno, Glass, & VanPatten, 1997; Mangubhai, 1991;
Pondea & Wong, 2003; Musumeci, 1989), that a First Noun Principle is active
(e.g., see Allen, 2000; VanPatten, 1996), that there are position preferences for
processing (e.g., Barcroft & VanPatten, 1997; Klein, 1986; Rosa & O’Neill,
1998), and so on.
To conclude, a model of L2 input processing must capture the developmental
and non-static nature of how learners get form from input and how they process
basic syntax. The model must address the basic questions of what gets
processed, why it gets processed, and the conditions that affect the processing.
The current model attempts to do this.

First Language Influence

The problem of L1 influence in SLA has long been acknowledged, but its
manner of operation is debated. Within input processing one might ask the
question “Doesn’t the nature of the L1 influence how learners process input?”
This is a legitimate question. For example, shouldn’t French speakers process
the subjunctive sooner and better in Spanish than learners with English as the
L1? Given they already have grammars with the subjunctive they can use this
knowledge to process the subjunctive in the input. Although this may be true, it
does not mean that the learners will process the subjunctive before non-
redundant forms, for example. It does not mean that the learners do not have
preference for location in sentences, for example. It simply means that when
they finally do notice the form in the input, they are more likely to process it
(connect form and meaning/function) than their English L1 counterparts.
Several other questions about L1 influence surface in the present discussion:
what about the First Noun Principle? Isn’t this principle dependent on the first
language? What about learners whose L1 is object first (e.g., OVS)? Wouldn’t
they process sentences thinking the first noun is the object? These, too, are
legitimate questions. Those working within the Competition Model would
certainly claim that learners begin with L1 processors. The Competition Model
says that sentence processing involves the use of weighted cues and that in a
24 PROCESSING INSTRUCTION

given L1, certain cues are preferred over others because in that language they
are more reliable and valid. For example, in English word order is preferred over
cues such as animacy as a cue to agency and patience because in natural
language data English word order is highly reliable as a cue. MacWhinney
(1997) has made the following claim about SLA and processing: “We find,
uniformly, that the learning of sentence processing cues in a second language is
a gradual process. The process begins with L2 cue weight settings that are close
to the L1” (p. 129). To cut to the chase, L2 learners transfer L1 cue weights.
Although this claim is backed by considerable research (see MacWhinney,
1997, for the summary), I have wondered about the foundation of the
Competition Model. Cue weight preferences are determined by giving speakers
sentences in which cues are put into deliberate conflict. For example, NVN
sentences are fairly unrevealing to understand how people assign agency or
subject-status to nouns because the world’s languages are overwhelmingly
subject-first, or rather subject-before-object, in terms of canonical order. So
what those working with the Competition Model do is create conflicts with
NVN in which animacy is thrown in as the conflict. For example, participants
are asked to respond to sentences such as rock-throw-monkey. What the
Competition Model research finds is that English speakers tend to select the rock
as the subject more often than speakers of other languages, thus showing a
strong preference for word order. Italian speakers would select the monkey more
often because it is animate (see the Lexical Semantics Principle earlier in this
chapter).
What puzzles me about this research is what it actually says about non-
conflict situations. When given sentences such as monkey-bite-baboon, all
speakers regardless of language, pick the monkey as the subject. This suggests
to me that there is something much more primitive and default about word order
(the first noun as the subject) and that differences only surface when we
deliberately put conflict into sentences. Thus, some languages develop cues in
addition to word order but everyone has word order at the core. If this
interpretation is correct, then English speakers do not transfer word order to L2
processing; they simply use the default cue. Speakers of other languages have
cues in addition to the basic cue and thus may transfer those while retaining the
core cue of word order.
The only way to test this hypothesis, of course, is to work with learners of
OVS languages. I do not see this happening in the near future. We do have
research on learners from SVO and SOV languages. But as mentioned above,
even when word orders are the same as in the case of English and French
passives, we do not see positive transfer.
In more general terms, it is certainly possible that learners transfer language
specific aspects of their parsers. That is, they may transfer strategies such as the
late/early closure strategy or other co-reference preferences (Fernández, 1999).
If their L1 parsers cannot handle such things as gaps (e.g., Juffs & Harrington,
 1. INPUT PROCESSING 25

1995), they may have difficulty processing sentences with extractions; the
difficulty being that they have to backtrack to reanalyze a sentence and may not
have the capacity to do so. These functional areas of syntax in processing are
certainly worthy of exploration but, again, we must remember the fundamental
question that has driven current input processing research: What form-meaning
connections do learners make and why do they make those and not others at a
given point in time? Input processing may evolve into two subareas, one that
deals with form-meaning connections and another that deals with issues of
syntax or sentence processing.

The Relationship of Input Processing

to Restructuring
Some might consider or conclude that a model of input processing is a model of
acquisition. This is not so. In previous publications I describe acquisition as
something consisting of different processes, each bringing something to bear on
acquisition. In Figure 1.1, I offer a shorthand sketch of these processes.
(Although this figure captures the basic processes in acquisition, it is not meant
to convey that acquisition is linear with no interaction between some of
processes and/or products.) As this shorthand depiction suggests, input
processing is the “first hurdle” a form or structure must jump through on its path
toward acquisition. If a form is processed (there is a connection of form and
meaning, whether right or wrong from a target standpoint), it becomes available
for further processing and may be accommodated into the developing linguistic
system. Accommodation may be complete, partial, or it may not happen at all
(for reasons not well understood). An accommodated form may cause
repercussions in the grammar and trigger some kind of restructuring, such as a
parameter resetting within a UG framework or something like U-shaped
development, documented in cognitive and behavioral research. Thus, input
processing only offers data for the internal mechanisms that store and organize
language in the brain; it does not do the organization and storage itself. As one
example, English-speaking learners of Spanish get unequivocal data early on
that Spanish has verb movement. The data come in the form of VSX questions
such as ¿Almorzó María con Juan? (Did Mary eat lunch with John?). The verb
appears before the subject and can only have done so if the language allows verb
movement. Input processing tags the content lexical item in initial position,
almorzó, with the meaning ate lunch and assigns it the status of VERB. Input
processing tags the content lexical item María with its meaning and assigns it
the status of NOUN and SUBJECT. Thus, the input processing mechanisms
make available the VS word order to the processors responsible for parameter
resetting (if one believes in a UG account of things) and it is there where the
restructuring begins. Spanish speaking learners of English get consistent SVX
26 PROCESSING INSTRUCTION

word order and so there are no data disconfirming verb movement delivered to
the internal processors given that SVX is possible in verb movement languages.
Output is the result of the acquisition and development of separate processes
such as access and production procedures (e.g., Pienemann, 1998). Again, input
processing is the starting point for acquisition; it is not the end point nor can it
be equated with acquisition itself. Figure 1.2 captures in shorthand what I have
discussed regarding input processing in this chapter. Note that Figure 1.2 zeroes
in on that part of acquisition that involves the derivation of intake from input.

FIG. 1.1. A Sketch of Basic Processes in Acquisition.

FIG. 1.2. A Detail of Process I from Figure 1.1.

The Role of Output in Acquisition, Again

Because of the fundamental role I place on input in making form-meaning
connections for the development of an underlying system, I have often been
juxtaposed to interactionist positions or the position taken by Swain on the role
of output in SLA. The suggestion is that I dismiss output or do not think it is
important simply because it does not figure into my research. As my discussion
regarding P1e above clearly shows, one role for output in acquisition may be the
 1. INPUT PROCESSING 27

effect it has on the task demands. Interaction may make input more manageable
by creating shorter sentences for learners to process, by repeating information so
that the demands to get meaning are lessened, by moving elements into more
salient positions, and so on. Greater manageability can lead to increased
resources for processing.
What I disagree with in terms of a possible role for output is that using a form
in one’s output is a direct path to acquisition. In some approaches to SLA, such
as skill acquisition theory (e.g., DeKeyser, 1997, 1998), the claim is that SLA
follows the same rules that skill learning in general follow, namely, learners
begin with some kind of declarative knowledge that is proceduralized and
automatized via output (e.g., Anderson, 2000). Learners acquire forms and
structures by using them repeatedly in communicative situations. My belief is
that skill development is a separable phenomenon from the creation of an
implicit linguistic system (see VanPatten, 2003). I believe that learners do create
linguistic systems based on input processing, accommodation, restructuring,
interaction, and the mechanisms involved in these processes. I also believe that
learners have to acquire appropriate procedures for accessing forms and phrases
and also for stringing them together to make utterances (Pienemann, 1998;
Terrell, 1986). In short, the creation of a linguistic system does not in and of
itself guarantee fluency and accuracy in production. At the same time practicing
forms and structures in output does not result in acquisition as I (and probably
most others) define it. Thus, I distinguish between output as interaction with
others and output as production of forms and structures. Although learners may
need to produce output to develop the types of procedures Pienemann describes
and only through output can they automate these procedures, making output
does not create an implicit system. (For more detailed discussion, see VanPatten,
forthcoming).

CONCLUSION

In this chapter I have updated and clarified various aspects of input processing
from previous publications (e.g., VanPatten, 1996, 2002a and b). We are far
from understanding everything we need to know about L2 input processing. My
hope is that psycholinguists in SLA continue to work in this fertile area of
investigation. Understanding how learners deal with initial and subsequent
encounters with formal features in the input can offer great insights into the
creation of an L2 linguistic system. As the rest of this book suggests, there are
insights to be gained for L2 pedagogical intervention as well.
To be absolutely clear, my hope for continued work in input processing is not
a call for researchers to abandon other areas of SLA. As the opening sentence to
this chapter says, SLA is a complex process. It cannot be reduced to one simple
theory or one simple mechanism. In a sense, understanding SLA is like
28 PROCESSING INSTRUCTION

understanding how a building works. There is the electrical system, the

plumbing, the foundation, the frame, the heat and air system, and so on. All are
necessary; one alone is insufficient. But like those who work in house
construction and are electrical contractors or plumbing contractors, in SLA some
of us are interested in matters dealing with input. Others are interested in output.
Others are interested in universal aspects of language. Others have focused on
interaction. This is a good thing.

REFERENCES

Allen, L.Q. (2000). Form-meaning connections and the French causative: An

experiment in processing instruction. Studies in Second Language
Acquisition, 22, 69–84.
Anderson, J.R. (2000). Learning and memory (2nd ed.). New York: John Wiley
& Sons.
Barcroft, J., & VanPatten, B. (1997). Acoustic salience of grammatical forms:
The effect of location, stress, and boundedness on Spanish L2 input
processing. In A.T.Pérez-Leroux & W.R.Glass (Eds.), Contemporary
perspectives on the acquisition of Spanish: Vol. 2, Production, processing,
and comprehension (pp. 109–121). Somerville, MA: Cascadilla Press.
Blau, E.K. (1990). The effect of syntax, speed and pauses on listening
comprehension. TESOL Quarterly, 24, 746–753.
Chaudron, C. (1985). Intake: On methods and models for discovering learners’
processing of input. Studies in Second Language Acquisition, 7, 1–14.
Clifton, C.Jr., Frazier, L., & Rayner, K. (Eds.). (1994). Perspectives on sentence
processing. Hillsdale, NJ: Lawrence Elrbaum Associates.
Grain, S., Ni, W., & Conway, L. (1994). Learning, parsing and modularity. In C.
Clifton, Jr., L.Frazier & K.Rayner (Eds.), Perspectives on sentence
processing (pp. 443–467). Hillsdale, NJ: Lawrence Elrbaum Associates.
Cuetos, F., Mitchell, D.C., & Corley, M.M.B. (1996). Parsing in different
languages. In M.Carreiras, J.E.García-Albea & N.Sebastián-Gallés (Eds.),
Language processing in Spanish (pp. 145–187). Mahwah, NJ: Lawrence
Elrbaum Associates.
DeKeyser, R.M. (1997). Beyond explicit rule learning: Automatizing second
language morphosyntax. Studies in Second Language Acquisition, 19, 195–
221.
DeKeyser, R.M. (1998). Beyond focus on form: Cognitive perspectives on
learning and practicing second language grammar. In C.Doughty & J.
Williams (Eds.), Focus on form in classroom second language acquisition
(pp. 42–63). Cambridge: Cambridge University Press.
 Chapter 2
The Nature of Processing instruction

Wynne Wong
The Ohio State University

Processing Instruction (PI) is a type of focus on form instruction that is

predicated on a model of input processing (see VanPatten, chap. 1, this volume,
and elsewhere). The goal of PI is to help L2 learners derive richer intake from
input by having them engage in structured input activities that push them away
from the strategies they normally use to make form-meaning connections. This
chapter describes in detail the nature of PI. I describe the three major
characteristics of PI: (1) explicit information about the target structure, (2)
explicit information about processing strategies, and (3) structured input
activities. I pay particular attention to the nature of structured input activities
because, as is clear from attempted replication research by others (e.g., Allen,
2000; DeKeyser & Sokalski, 1996; Salaberry, 1997), the purpose and nature of
structured input activities is not always clear to readers. As part of my
discussion, I also examine the guidelines suggested for developing structured
input activities as first outlined in VanPatten (1993) and then expanded on in
Lee and VanPatten (1995), VanPatten (1996) and Wong (2002a). Of particular
interest and critical to PI is the sixth guideline, “Keep learners’ psycholinguistic
processing strategies in mind.”

INPUT PROCESSING AND PROCESSING INSTRUCTION

PI is a type of explicit grammar instruction that is informed by a model of how

L2 learners initially process L2 input to make form-meaning connections, that is
to say, VanPatten’s model of input processing (1993, 1996, 2002, this volume).

FIG. 2.1. Three Sets of Processes in Second Language Acquisition

(based on VanPatten, 1996).
34 PROCESSING INSTRUCTION

VanPatten conceptualizes second language acquisition (SLA) as the result of

internal mechanisms acting on meaning-bearing input. These mechanisms are
conceptualized as consisting of sets of processes (Fig. 2.1.). The first process,
called input processing, involves the conversion of input into intake. Intake is
defined as input that learners pay attention to and from which form-meaning
connections have been made (VanPatten, 1996). This subsequent processing
involves the actual incorporation of the data into the system (called
“accommodation”), which may be partial or complete. Depending on the nature
of the data, accommodation may have an effect on the developing system such
that some kind of restructuring may occur. Finally, linguistic data that have been
incorporated into the developing system may be eventually accessed by the
learner as output or production. This process is called “access.”
VanPatten’s model of input processing focuses on the first process, that is,
input processing. The model provides various principles with corollaries to
explain (1) what learners attend to in the input and why, (2) what strategies
direct how they make form-meaning connections, and (3) why they make some
form-meaning connections before others.
The model postulates that L2 learners will initially pay attention to items in
the input that are crucial for understanding referential meaning before items that
are less important for doing so (Principle 1).1 This means that more meaningful
items in the input will initially get processed before less meaningful ones and
implies that content words are probably the first things that learners process
(e.g., see research by Musumeci, 1989; Lee, Cadierno, Glass, & VanPatten.
1997). Principle 1e explains that learners will also be able to attend to forms that
are less meaningful if they do not have to struggle with understanding the
meaning of the message.
Principle 2, also known as the first noun principle, deals with how word order
affects processing strategies, particularly when the L2 does not follow a strict
SVO order. Research (e.g., Lee, 1987; VanPatten, 1984) has documented that
learners tend to assign the role of agent to the first noun that they encounter even
when the first noun is not the subject. Thus, the model predicts that when
learners initially attempt to parse sentences, they have a tendency to assume that
the first noun is always the subject.
Principle 1f concerns how the position of the form in question may have an
effect on whether or not it is likely to get processed. Supported by research by
Barcroft and VanPatten (1997) and Rosa and O’Neill (1998), this principle
postulates that forms that are in initial position in a sentence get processed first
followed by those in sentence final position. Forms in medial position tend to be
the most difficult to process.

1
See VanPatten (this volume, chap. 1) for a complete list and description of the
principles.
 2. THE NATURE OF PI 35

As can be seen from this model, the strategies that L2 learners use to process
input are not always efficient and may sometimes be wrong. The goal of PI is to
push learners away from these processing strategies toward more optimal ones.
What makes PI unique compared to other focus on form techniques is that it first
identifies the processing strategy that hinders learners from processing a
particular form or structure correctly. Once the strategy has been identified,
activities are created to help learners process input more efficiently. For
example, according to Principle 1, learners tend to process input for meaning
before they do so for form and will consequently pay less or no attention to
forms that are not critical to the propositional content of utterances or sentences.
How can we manipulate the input so that they will pay attention to these items?
According to Principle 2, learners tend to rely on a first noun strategy and assign
the role of agent/subject to the first noun in a sentence. What forms and
structures does this strategy affect and in what ways can we structure the input
to ensure that learners do not rely on the strategy?
To summarize, PI is a pedagogical tool that is informed by a model of input
processing. Input processing is a model of how L2 learners initially parse L2
input to make form-meaning connections. Based on this information about how
L2 learners process input on their own, PI pushes learners to abandon their
inefficient processing strategies for more optimal ones so that better form-
meaning connections are made.

THREE CHARACTERISTICS OF PI

PI has three basic characteristics. First, learners are given information about how
the linguistic form or structure works, focusing on one form or use at a time.
This explicit information (EI) also informs learners about a particular IP strategy
that may lead them to process the input incorrectly. This is the second
characteristic of PI. An example of what the El may look like in PI is found in
Figure 2.2. The target of instruction here is the French causative and the
processing problem is a word order problem. Note that in order to alert learners
of their processing problem, this explanation includes information about
learners’ processing strategies (i.e., “don’t use a first noun strategy”).
The third characteristic of PI involves giving learners “structured input” (SI)
activities. They are termed structured input activities because the input has been
manipulated so that learners are pushed away from the less-than-optimal
strategies described earlier. (An incidental byproduct is that learners’ attention is
36 PROCESSING INSTRUCTION

We often ask or get people to do things for us by telling them to do something.

Paul says, “John, would you mind doing the dishes?”
If you and I were to describe what is happening we might say:
We say, “Paul gets John to do the dishes.”
or
“Paul makes John do the dishes.”
This is called a causative construction (because someone is causing a behavior in
someone else). French has a similar structure using the verb faire. Let’s repeat our
examples from above.
Paul says, “Jean, pourrais-tu faire la vaisselle?”
We say, “Paul fait faire la vaisselle à Jean.”
How would we describe the following scenario?
Wynne says, “Sara, pourrais-tu promener le chien?”
We would describe Wynne getting Sara to do it like this.
We say, “Wynne fait promener le chien à Sara.”
Often we don’t mention who we get to do something; we might simply say we have
something done.
“Paul fait nettoyer la chambre.”
In this case, Paul has the room cleaned, but we don’t know who or how.
One of the problems the faire causatif presents is in listening comprehension. Second
language learners of French often misinterpret what they hear because the word order is
different from English. For example, it is not uncommon for learners of French to make
the following mistake:
They hear: “Jean fait faire la vaisselle à Paul.”
They incorrectly think: John is doing the dishes for Paul.
or
They hear: “Marc fait couper les cheveux.
The incorrectly think: Marc cuts hair.
In the activities that follow, we will practice hearing and interpreting the faire causatif.

FIG. 2.2. Example of Explicit Information in PI.

 2. THE NATURE OF PI 37

drawn to the relevant form-meaning connection.) Furthermore, because the goal

of these activities is to help learners create intake from input, they do not
produce the target structure. Instead, learners are pushed to make form-meaning
connections by requiring them to rely on form or sentence structure to interpret
meaning. The fact that learners do not produce the target structure in SI
activities, however, should not be taken to mean that output has no role to play
in SLA (e.g., see Swain, 1997, VanPatten, 2003, and Wong & VanPatten, in
press) for discussion on the roles of output). Output practice is not a component
of PI simply because PI is concerned with input processing, the process
responsible for converting input to intake. Output practice is related to the
process of accessing internalized language for production (see VanPatten 2002,
2003, and this volume for discussions and Lee & VanPatten, 1995 and in press,
for what they refer to as “structured output” activities and their complementary
role to structured input).

DEVELOPING SI ACTIVITIES

To develop appropriate and effective SI activities, certain procedures should be

kept in mind. It is important to point out that SI activities cannot be equated with
just any kind of input-based activity. In other words, the fact that an activity
does not require the learner to produce the target form does not automatically
make that activity an SI activity. For an activity to be an SI activity, that activity
must somehow push learners to circumvent an inefficient processing strategy. In
the next sections, I give a detailed description of procedures to follow in
developing SI activities.

Step 1: Identify the Processing Problem or Strategy

The first and perhaps most important step in developing SI activities is to
identify and understand what the processing problem is for the form in question.
Why are learners having problems processing a particular form? What strategies
are they using that is causing them to process this form inefficiently or
incorrectly? Is it due to a tendency to rely on lexical items (i.e., Principle 1)? Is
it due to a word order problem (i.e., Principle 2)? Is location a problem (P1f)?
Or is some combination of factors involved? Remember that the goal of PI is to
push learners away from their less than optimal strategies for processing input.
If the processing problem or strategy is not identified, we will not be able to
create SI activities to help reach this goal.
Once the processing problem has been identified, then the development of SI
activities can begin. The input in these activities should be structured so that
learners cannot rely on inefficient strategies to successfully complete the
activities. The activities should force them to use more optimal strategies to
38 PROCESSING INSTRUCTION

process the form in question. Guidelines for developing SI activities are

described in the next section.

Step 2: Follow Guidelines for Developing SI Activities

Guidelines for developing SI activities were first outlined in VanPatten (1993)
and then expanded on in Lee and VanPatten (1995), VanPatten (1996) and
Wong (in press). We will discuss these in detail in the following sections.
1. Present one thing at a time. In order to maximize intake efficiency, only
one function and/or one form should be the focus of instruction at any given
time. As explained in Lee and VanPatten (1995), this means that only one rule
of usage and/or one form of a paradigm should be presented at a time. The
reason for this guideline is simply this: when there is less to pay attention to, it is
easier to pay attention. Because learners have a limited capacity to process
information, we want to make sure that we do not overtax their processing
resources. Presenting one rule at a time and breaking up paradigms help
maximize the chances that learners will pay more focused attention to the forms
that they need to for intake. Furthermore, presenting one thing at a time helps
ensure that learners will not be bogged down with too much description and
explanation about rules (Lee & VanPatten, 1995).
2. Keep meaning in focus. Acquisition cannot happen without exposure to
meaning-bearing input, that is to say, input that contains some kind of referential
meaning or communicative intent. Therefore, meaning must be kept in focus at
all times. This means that meaning-bearing input must be used in SI activities
and learners must successfully process the propositional content of the input in
order to successfully complete the activities. If the activity can be completed
without attention to the referential meaning of the input (as in the case of
mechanical drills), then it is not an SI activity.
3. Move from sentences to connected discourse. When grammar is taught via
SI activities, it is preferable to begin with sentences first. The rational for this
goes back to the fact that learners have a limited capacity to process input,
especially at the beginning stages. As outlined in Principle 2 of VanPatten’s
model of input processing, learners process input for form only if their
processing resources have not been depleted after they have processed the input
for meaning. Because short sentences are easier to process than connected
discourse, learners will be more likely to pay attention to the relevant
grammatical information that is the target of instruction (see, for example,
Wong, 2002b). This guideline is exemplified in the following activities (from
Lee & VanPatten, 1995, pp. 106–107). The target structure in these activities is
third person present tense verbs and is a translation of the original Spanish
version. Note how the activities move from sentences to connected discourse
(short narration).
 2. THE NATURE OF PI 39

Activity A. Alice and Ray. Look at the drawings of events from

a typical day in the lives of Alice and Ray. Listen as your
instructor reads a sentence. Say whether that activity is part of
Alice’s routine or Ray’s.

MODEL: (you hear) This person eats lunch with friends,

(you say) That is Ray.

Activity B. In What Order? Without referring to the drawings

about Ray’s day, put the following activities in the correct order
in which he does them.

____ a. He goes to bed late.

____ b. He sleeps in his math class.
____ c. He works at the pizzeria.
____ d. He goes to music class.
____ e. He gets up late.
____ f. He watches some TV.
____ g. He eats lunch with friends.
____ h. He tries to study.

Now compare with the drawings. Did you get them all in the
right order?

Activity C. The Typical Student. Read the following sentences.

Are they true for a typical student at your school?
The typical student… True False
1. gets up at 6:30 a.m. ____ ____
2. skips breakfast. ____ ____
3. drives to school. ____ ____
4. sleeps in at least one class. ____ ____
5. studies in the library, not at home. ____ ____
6. works part time. ____ ___
7. eats a microwaved dinner. ____ ____
8. watches David Letterman at night. ____ ____
9. goes to bed after midnight. ____ ____
40 PROCESSING INSTRUCTION

Your instructor will now read each statement and then ask you
to raise your hand if you marked it as true. Someone should keep
track of the responses on the board. In the end, how did the class
respond to each statement?

Activity D. John’s Day.

Step 1. Break into groups of three and listen as your instructor
reads a short narration.
Step 2. With your group members, give as many details as
you can remember by completing the following sentences. The
group with the most details wins. You have three minutes.
1. John gets up at______.
2. He requires at least______to wake up fully.
3. He prefers not to______in the morning. [the
list continues]

Step 3. Look over the details that you have recalled. Read a
sentence to the class and then say whether or not you do the same
thing.

MODEL: John gets up at 8:00, and so do I.

John gets up at 8:00, but I don’t.

[Part of narration read by instructor. “John is a student at X

university. On most days, he gets up at 8:00 but the mornings are
very difficult for him because he just isn’t a morning person. He
needs to drink at least three cups of strong coffee to wake up.
And more often than not he reads the newspaper in silence since
he prefers not talking to anyone until he is fully awake…”]

4. Use Both Oral and Written Input. Both oral and written input should be used
in SI activities because learners should have opportunities to receive input in
both modalities. Although all learners need oral input, more visual learners
would benefit from “seeing” the input as well. Not giving these learners
exposure to written input could put them at a disadvantage in learning situations
(Lee & VanPatten, 1995), Note how both oral and written input was used in
Activities A through D.
5. Have Learners Do Something With the Input. In line with major tenets of
communicative language teaching, the activities should not only be meaningful,
they should also be purposeful. This means that learners must have a reason for
attending to the input. Therefore, the activities should have learners responding
 2. THE NATURE OF PI 41

to the input in some way to ensure that they are actively processing. This
guideline is illustrated in Activity E.
Activity E was constructed to help learners of French process de with the
verb avoir (to have) correctly in order to derive the meaning of negation in the
sentences. In French, indefinite articles (un/une/des) must change to de or d’ (if
in front of a vowel) if they are used in negative sentences with the verb avoir.
Thus we have Marie a une voiture (Marie has a car) in the affirmative but Marie
n’a pas de voiture (Marie does not have a car) in the negative. The processing
problem here is that learners will skip over the de vs. une distinction and will
rely instead on the ne…pas to get negation.
Activity E was structured so that learners had to rely on the de vs une
distinction to get meaning.

Activity E. Chez les LeBlanc

Étape 1. Pierre and Lise LeBlanc are talking about things they have and
don’t have in their house. Pay attention to the articles to determine
whether they have or do not have the things mentioned. Complete each
sentence with either “Nous avons…” or “Nous n’avons pas…”

______ une salle de séjour.

______ de télévision.
______ de lit.
______ un fauteuil.
______ une cuisinière.
______ de réfrigérateur.
______ une table.
______ une toilette.
______ une douche.
______ de baignoire.
______ de lampes.
______ de chaises.
______des souris (mice).

Étape 2. Based on these descriptions, decide with a partner how rich or

poor this couple is and explain why.

Pierre et Lise sont…très riches/riches/assez riches/assez

pauvres/pauvres/très pauvres (circle one) parce que….

INSTRUCTOR’S SCRIPT
Give students a few minutes to work on step one on their own.
Then go over each answer with them. Go on to step two. Have
42 PROCESSING INSTRUCTION

them discuss their answers with a partner and then share with
whole class to see if everyone wrote the same thing.

In this activity, the learners’ task is to determine whether the couple, the
LeBlancs, has the household items listed by filling in the blanks with either
“Nous avons” (we have) or “Nous n’avons pas” (we do not have). The
guideline of having learners do something with the input is observed by
requiring learners to draw a conclusion about the couple’s economic status at the
end of the activity. In other words, learners have to share with their classmates
their thoughts about how rich or poor they think the LeBlancs are based on the
input they just processed. The answers shared may vary. The LeBlancs don’t
have a bathtub (baignoire) or a bed (lit) but they do have a shower and mice.
Would that make them poor? Somewhat poor? Very poor? Learners may have
different opinions. The point here is that learners were given a reason for
attending to the input, that is, they had to report to their classmates about the
economic status of the LeBlanc’s. And to be able to provide this information,
they had to have correctly processed the target forms in the first part of the
activity.
6. Keep the learner’s processing strategies in mind. The goal of PI is to help
learners move away from inefficient processing strategies so that they adopt
more optimal ones. Therefore, the processing strategies that learners use to
process a particular form must be kept in mind at all times in developing SI
activities. If the activity is not constructed to preempt an inefficient processing
strategy, then it is not an SI activity. This is why it is critical that the processing
problem or the processing strategies that learners use for a given form be clearly
identified before SI activities are constructed. For example, if learners are
relying on lexical items to interpret tense (Principle 1), then we may want to
structure the activities so that learners are pushed to rely on grammatical
morphemes instead of lexical adverbs to get tense.
Recall that the processing problem in Activity E is that learners tend to rely
on ne…pas to get negation and will skip over the de/d’ vs. un/une/des
distinction. Keeping this processing strategy in mind, we would want to push
learners to make the connection that de/d’ with avoir denotes negation. Notice
that in Activity E, the ne…pas was removed from t he phrases so that learners
had to rely on the articles to determine whether the sentence should be
affirmative or negative. By removing ne…pas, learners were forced to pay
attention to de/d’ vs. un/une/des to get meaning.
Another difficulty with the target structure in Activity E is that it occurs in
medial position, the least salient position so learners are even more likely to
miss it (as explained by Principle 4). Notice that in Activity E, the input is
structured so that the articles are in initial position (the most salient position)
and are bolded to increase their perceptual salience. The first item learners see in
each phrase is either un/une/des or de/d’. It is through successfully processing
 2. THE NATURE OF PI 43

these forms that they are able to fill in the beginning part of each phrase (Nous
avons or Nous n’avons pas) and determine whether the couple has or does not
have each of the household items mentioned.

TWO TYPES OF STRUCTURED INPUT ACTIVITIES

Two types of SI activities are used in PI: referential and affective. Referential
activities require learners to pay attention to form in order to get meaning and
have a right or wrong answer so the instructor can check whether or not the
learner has actually made the proper form-meaning connection. Affective
activities, on the other hand, do not have right or wrong answers. Instead, they
require learners to express an opinion, belief or some other affective response as
they are engaged in processing information about the real world.
Activity E discussed above is an example of a written referential activity.
Learners had to process form correctly in order to get meaning and there was
only one correct answer in the first part of the activity. Activity F is an example
of an oral referential activity for the same target form.

Activity F. In the Classroom. The following sentences describe

objects that may or may not be in a classroom. You will hear the
last part of each sentence. Listen carefully to the article in order
to determine if each object mentioned is in the classroom or not.
Circle the correct response. Then based on the answers, comment
on whether you think this is a well-equipped classroom,
1. La salle a… / La salle n’a pas…
2. La salle a… / La salle n’a pas…
3. La salle a… / La salle n’a pas…
4. La salle a… / La salle n’a pas…
5. La salle a… / La salle n’a pas…
6. La salle a… / La salle n’a pas…
7. La salle a… / La salle n’a pas…
8. La salle a… / La salle n’a pas…

Would you say that this is a well-equipped classroom? Share your

response with a classmate.
44 PROCESSING INSTRUCTION

INSTRUCTOR’S SCRIPT
Read each sentence once. After each sentence, ask for an answer.
Do not wait until the end to review answers. Students do not
repeat or otherwise produce the structure.
1. …un tableau.
2. …un ordinateur.
3. …de fenêtre.
4. …des affiches.
5. …de téléphone.
6. …une television
7. …des chaises.
8. …de magnétoscope

As in Activity F, learners are required to process form correctly in order to get

the correct meaning and must form a conclusion about how well-equipped this
classroom is (i.e., do something with the input).
Activity G is an example of a written affective activity for the same target
form. In this activity, there is no right or wrong answer. Learners are instead
asked to process sentences about housing in Columbus, Ohio and Beverly Hills,
California to give a personal reaction and opinion. Notice, however, that the
input is once again structured so that attention to form is privileged: The target
form is in a salient initial position and is visually enhanced through bolding. The
activity makes learners do something with the input by requiring them to form
their opinions and share them with classmates.
Activity G. The following sentences describe what a typical student
apartment in Columbus may have or not have. Read each sentence and
indicate whether you think each statement about student apartments in
Columbus is true or false.
Un appartement d’ étudiant typique à Columbus a…
une cuisine (kitchen). vrai / faux
une salle de séjour (living room). vrai / faux
un grand jardin (garden). vrai / faux
des chambres (bedrooms). vrai / faux

Un appartement d’ étudiant typique à Columbus n’a pas…

de garage. vrai / faux
de salle à manger (dining room). vrai / faux
 2. THE NATURE OF PI 45

de balcon. vrai / faux

de piscine (swimming pool). vrai / faux

Now, repeat the above activity but this time imagine it is a

house in Beverly Hills. Do any of your answers change?

Une maison à Beverly Hills a….

une cuisine (kitchen). vrai / faux
une salle de séjour (living room). vrai / faux
un grand jardin (garden). vrai / faux
des chambres (bedrooms). vrai / faux

Une maison à Beverly Hills n’a pas…

de garage. vrai / faux
de salle à manger (dining room). vrai / faux
de balcon. vrai / faux
de piscine (swimming pool). vrai / faux

Did you and your classmates write similar or different things?

Because referential activities allow instructors to make sure that learners are
focusing on the relevant grammatical information to derive meaning, instruction
should begin with these activities. The purpose of affective activities is to
reinforce those connections by providing them with more opportunities to see or
hear the form used in a meaningful context. Furthermore, by requiring learners
to express an opinion or some other kind of personal response, we can keep
instruction in line with an important tenet of communicative language teaching:
a focus on the learner.

PI VS. TRADITIONAL INSTRUCTION

There is a substantial amount of empirical evidence to suggest that PI is an

effective pedagogical tool. In a series of studies that have compared PI to
traditional instruction (TI), overall results show that PI is superior to TI.
TI typically involves giving learners explicit explanation of a form followed
by controlled output practice. The practice activities usually begin with
mechanical drills followed by meaningful and communicative drills (Paulston,
46 PROCESSING INSTRUCTION

1972).2 TI has been criticized in VanPatten (1996) and elsewhere for being an
approach that does not take into consideration the crucial role that meaning-
bearing input plays in SLA. Because TI involves immediate production practice
of forms, learners do not get the input that they need to construct mental
representations of the structure (p. 6).
The first study on PI was done on Spanish object pronouns and word order by
VanPatten and Cadierno (1993). This study set out to compare PI to the
traditional approach to instruction described above. Subjects were randomly
assigned to three groups (1) a PI group, (2) a TI group and (3) a control group
that received no instruction. In the PI group, learners first received El about how
object pronouns work in Spanish. They were also told that learners of Spanish
have a tendency to think that the first noun they encounter is the subject. They
were told that this is not an effective strategy because Spanish has a more
flexible word order and the first noun is not always the subject. After receiving
this El, learners engaged in a series of SI activities that pushed them to interpret
word order and object pronouns correctly. The following are SI activities from
VanPatten and Cadierno’s study.

Actividad A. Select the picture that best corresponds to the sentence.

(Keep in mind that Spanish does not follow a rigid subject-verb-object
word order and that object pronouns may go before a conjugated verb or
at the end of an infinitive.)
[Learners see a set of two pictures for each item]
[picture 1: parents calling son]
[picture 2: son calling parents]
1. Sus padres lo llaman por teléfono.
[picture 1: Manuel inviting two girls to the movies]
[picture 2: two girls inviting Manuel to the movies]
2. Las invita Manuel al cine.
[picture 1: nephew listening to grandmother]
[picture 2: grandmother listening to nephew]
3. La abuela lo escucha.
[picture 1: little boy waves at little girl]
[picture 2: little girl waves at little boy]
4. Lo saluda la niña.

2
See Lee and VanPatten (in press) and Wong and VanPatten (in press) for
discussions on how TI is manifested in current L2 textbooks and classrooms.
 2. THE NATURE OF PI 47

[picture 1: little boy looks for little girl]

[picture 2: little girl looks for little boy]
5. El chico la busca.
Actividad B. Indicate whether each statement about your parents applies
to you. Share your responses with a classmate.

Sí, me aplica No, no me aplica

______ ______ 1. Los Ilamo con frecuencia por teléfono.
______ ______ 2. Los visito los fines de semana.
______ ______ 3. Los visito una vez al mes.
______ ______ 4. Los abrazo cuando los veo.
(abrazar=to hug)
______ ______ 5. Los comprendo muy bien.
______ ______ 6. Los ignore completamente.
______ ______ 7. Los aprecio mucho.

Did you notice that there are no explicit subject nouns or subject
pronouns in each sentence? Because the yo form of the verb can only
refer to yo, no subject pronoun is needed. All of the sentence simple
word order object pronoun-verb.

Actividad C. Select a female relative of yours (madre, hermana, tía,

abuela, prima, etc.) and write her name below. Which of the statements
describes how you feel about her?
Pariente:_______________ Nombre:____________
______ 1. La admiro.
______ 2. La respeto.
______ 3. La quiero mucho.
______ 4. Trato de imitarla.
______ 5. La detesto.
______ 6. La ______? ______.
Now select a male relative and do the same.
______ 1. Lo admiro.
______ 2. Lo respeto.
______ 3. Lo quiero mucho,
______ 4. Trato de imitarlo.
48 PROCESSING INSTRUCTION

______ 5. Lo detesto.
______ 6. Lo ______? _______.

Compare with two other people. Did you select the same relative(s)? Did
you mark the same items?

Actividad D. Listen to each statement and select the appropriate picture.

1.
[picture a: Juan is calling some other guy]
[picture b: Some guy is calling Juan]
2.
[picture a: man listens to woman]
[picture b: woman listens to man]
3.
[picture a: mother hugs daughter]
[picture b: daughter hugs mother]
4.
[picture a: two men greet woman]
[picture b: woman greets two men]
5.
[picture a: girl looks at boy]
[picture b: boy looks at girl]
For activity D, the instructor reads the following statements.
1. Lo llama Juan por teléfono.
2. La escucha el señor.
3. La abraza la mamá.
4. Los saluda la mujer.
5. El niño la mira.

Actividad E. For each paso of this activity, work in pairs.

Paso 1. Look over the following reading then answer the questions
afterward. The following vocabulary may help you.
seguir (to follow)
asustarse (to become frightened)
aconsejarse (to advise)
jurar (to swear)
el colmo (the last straw)
paz (peace)
 2. THE NATURE OF PI 49

DEBATE ALICIA CUENTA SU PARTE

“Manuel es muy posesivo. No me
deja respirar. Cada vez que voy a
salir, me pregunta con quién, a
dónde voy, qué vamos a hacer….
A veces me sigue. Lo juro.
Un drama familiar muy
Cuando un chico viene a
común: ella se siente
visitarme Manny lo interroga y él
dominada, perseguida por su
se asusta. El colmo: mis padres
hermano (que puede ser
me dejaron ir con unas amigas a
menor, pero con aires de
un concierto de Bon Jovi… Y
grandote) y no sabe cómo
Manuel les aconsejó que uno de
zafarse de él. Alicia y Manuel
ellos fuera con nosotras, para
son el caso típico.
supervisarnos. Por eso peleamos
mucho, le he dicho más de mil
veces que él no es mi papá y que
me deje en paz. Pero Manuel no
me suelta.”

1. The main problem here is Manny. Alicia thinks he is…

a. uncaring. b. domineering.

2. Which of the following words summarizes a major theme

in Alicia’s comments?
a. love b. friendship c. trust

Paso 2. Find the following in the reading:

me dejaron ir
me pregunta
me sigue
no me suelta
viene a vistarme

In each instance, Alicia is saying that someone is doing something to her

or for her. Can you identify the subject of the verb?
Paso 3. Find the following in the reading:
lo jur
lo mato
lo interroga

Who is the subject of each verb? Who or what does lo refer to?
50 PROCESSING INSTRUCTION

Actividad F. Manny responde

Paso 1. In the following selection, Manuel responds to his sister’s
claims. Read it now for general meaning. Then do the activities that
follow. Here is some vocabulary to help you.

no queda más remedio (no choice is left)

confiar (to trust)
mentir (to lie)

MANNY HACE UNA

ACLARACIÓN
“No quería decirlo, pero no me queda
más remedio. Si vigilo a mi hermana,
es porque me ha dado motives para
sospechar de ella. En varias
ocasiones la sorprendí con un tal
Sergio, que es uno de reputación por
el suelo. Una vez le dijo a mis padres
que iba al cine con las amigas y
después un buen amigo me contó que
la había visto en el cine…pero con
Sergio. Cómo puedo confiar en mi
hermana si miente a todos en la casa?
Ella no conoce a los chicos. Ese tipo
sólo busca una cosa. Y yo no quiero
que a mi hermana le suceda nada
‘feo.’”
For the next pasos in this activity, you should first work alone and then
share your responses with someone else.
Paso 2. Find all the uses of a third-person object pronoun. To whom or
what do they correspond? What is the subject of each verb next to which
you found each pronoun?
Paso 3. Which of the following best describes Manny’s feelings for his
sister?
______La quiere mucho.
______La admira.
______La detesta.
 2. THE NATURE OF PI 51

Paso 4. Which of the following does Manny probably do on a

Friday night if his sister goes out?
______La sigue para ver lo que hacer y con quién.
______La deja en paz porque es una adulta.
______La espera en casa.

Object Marker a
Recall that Spanish has the object marker a.
Los padres miran a los hijos.
Llamba a mis padres.

This object marker has no equivalent in English but is important in

Spanish since it provides an extra clue as to who did what to whom.
Since Spanish has flexble word order, the a reminds you that even
though a noun appears before the verb it doesn’t have to be the subject!
A María la llama Juan.
A María Juan la llama.
(John calls Mary)
Note that when an object appears before the verb, the corresponding
object pronouns must also be used. If you think that this is redundant, it
is! But redundancy is a natural feature of languages, right? (Hint: Think
about how we put tense endings on verbs when most of the time we also
say “yesterday,” “last night,” and so on.) What does the following
sentence mean? Who is doing what to whom?
A la chica la busca el chico.
Right. The boy is looking for the girl.
Actividad G. Select the English rendition of each sentence.
1. A mi mamá la besa mucho mi papá.
a. My mother kisses my father a lot.
b. My father kisses my mom a lot.
2. A mi papá no lo comprendo yo.
a. I don’t understand my father.
b. My father doesn’t understand me.
52 PROCESSING INSTRUCTION

3. A la señora la saluda el señor.

a. The woman greets the man.
b. The man greets the woman.
4. A los chicos los sorprende la profesora.
a. The professor surprises the boys.
b. The boys surprise the professor.

Activities A, D, E, F, and G are referential activities and Activities B and C are

affective activities. These activities were constructed to help learners abandon
their reliance on the first-noun strategy to process the input. Notice that learners
do not produce the target forms but they must process object pronouns correctly
to get meaning and to successfully do these activities. The input is also
structured so that the target forms are in initial position, the most salient
position, when possible. Both oral and written input are used and learners must
complete some kind of task with the input, that is, form an opinion, discuss
responses with classmates, among other activities. Furthermore, the activities
begin with sentence level input before moving on to discourse level input
(Activities E and F) and the activities frequently remind learners of the
processing strategy that they should use when doing the activities (e.g., “Keep in
mind that Spanish does not follow a rigid subject-verb-object word order and
that object pronouns may go before a conjugated verb or at the end of an
infinitive”).
Subjects in the TI group received an explanation of object pronouns followed
by mechanical, then meaningful, then communicative output drills (based on
Paulston, 1972).3 This group did not engage in any interpretation activities.
The researchers found that on a sentence level test of interpretation that
required subjects to select pictures that best corresponded to what they heard,
the PI group made significant gains whereas the TI and control group did not.
These gains were maintained on a delayed posttest one month later. On a
production test that required learners to complete sentences based on pictures
that they saw, both the TI and PI groups made significant gains on the
immediate and delayed posttest and these gains were not significantly different
from each other. The control group did not make significant gains.

3
TI was operationalized using Paulston’s (1972) classification of drills. The activities
began with mechanical drills, then moved to meaningful drills and finally to
communicative drills. This is important to point out because some researchers have
misinterpreted the TI treatment in VanPatten and Cadierno’s study as comprising strictly
of mechanical activities (see DeKeyser, Salaberry, Robinson, & Harrington, 2002, p.
816).
 2. THE NATURE OF PI 53

The results from VanPatten and Cadiemo (1993) suggest that PI is more
beneficial than TI because not only did subjects in the PI group gain in ability to
interpret object pronouns, their input processing of this structure resulted in
some kind of change in their system that could be accessed for production. This
is an important finding because at no time during treatment did subjects in the PI
group ever practice producing object pronouns. Yet on the production task, they
were able to perform as well as subjects in the TI group who received lots of
practice in producing this structure. The subjects in the TI group on the other
hand, could not do the interpretation task. Their performance on this task was no
better than those in the control group who received no instruction. Subjects in
the TI group were only good at doing what they practiced doing during
treatment. Other studies that follow the same research design and have reported
similar superior results for PI include Benati (2001) for the Italian future tense,
Cadierno (1995) for Spanish preterite tense, Cheng (1995) for Spanish ser vs.
estar, and VanPatten and Wong (this volume) for the French causative.
VanPatten and Sanz (1995) demonstrated that the effects of PI can also be
generalized to assessment measurements that involve more complex cognitive
processing such as a video narration task. In another study, VanPatten and
Oikkenon (1996) separated out the effects of SI activities and El from PI and
found that the positive effects of PI were due to the SI activities. A recent
longitudinal study by VanPatten and Fernández (this volume) shows that the
effects of PI are durable for at least eight months.
Recently, some SLA researchers have remarked that of the focus on form
techniques that are in the literature, PI appears to yield some of the most
promising results (e.g., Carroll, 2001; Doughty, 2002; N. Ellis, 2002; Norris &
Ortega, 2000). These researchers attribute the positive results of this focus on
form technique to the fact that PI was designed with the goal of altering
learners’ processing strategies. According to these researchers, in L2 learning
situations where input alone may not be enough, the best kind of intervention
appears to be one in which input is structured so that learners can perceive and
process/parse L2 stimuli more effectively and accurately (Doughty, 2002; N.
Ellis, 2002).

PI AND REPLICATION STUDIES

We cannot overemphasize the nature and purpose of SI activities. It is tempting

to see the activities in PI as simply being “more input” or “embedding a
structure in input” because the underlying psycholinguistic strategies of learners
may not be thought about by many researchers (and certainly not by a lot of
instructors). Thus, to illustrate what constitutes appropriate SI activities, the next
section will present activities from certain replication studies that purport to
contradict VanPatten and Cadierno’s findings. My point in reviewing these is
54 PROCESSING INSTRUCTION

not to criticize the authors for their research, but rather to contrast their activities
with those used by VanPatten and his colleagues to demonstrate how easy it is to
misinterpret what SI activities are and what they are supposed to do. Those who
wish to implement or even test the effects of SI activities with their own learners
or under different contexts will benefit by such a presentation.

DeKeyser and Sokalski (1996)

DeKeyser and Sokalski (1996) state that they set out to replicate VanPatten and
Cadierno’s study. They compared two treatment groups, an input group and an
output group, to a control group. The target items were Spanish object pronouns
and the conditional tense in Spanish. The El for the two experimental groups
was the same. This information was adapted from sections of Dos Mundos
(Terrell, Andgrade, Egasse, & Munoz, 1994), the text used by the participants’
classes. The practice exercises in both groups began with a few mechanical
activities and then progressed to meaningful and communicative ones (p. 626).
Sentence length and vocabulary were held constant for both groups. Essentially,
the only difference between the practice exercises for the two groups was that it
was input-based in one version and production-based in the other version. The
following is an example of one of the exercises that the input group and output
group received (from DeKeyser & Sokalski, 1996, pp. 640–642).

Input Group.
Read the following exercises. Circle all possible things that
the direct object can refer to.

Example: Yo lo tengo
me you(sg.) him her it you (p1.)/them
1. La escuchas conmigo
me you(sg.) him her it you(p1.)/them
2. Nosotros te vemos
me you(sg.) him her it you (p1.)/them
3. Ellos los lavan en la cocina
me you(sg.) him her it you (p1.)/them
4. Yo lo prefiero de color azul
me you (sg.) him her it you (p1.)/them
5. Lo tocan ellas muy bien
me you (sg.) him her it you (p1.)/them
 2. THE NATURE OF PI 55

For the next 5, listen to the sentences your teacher reads and
circle all possible things the direct object can refer to.

6. me you (sg.) him her it you(p1.)/them

7. me you (sg.) him her it you(p1.)/them
8. me you(sg.) him her it you(p1.)/them
9. me you (sg.) him her it you (p1.)/them
10. me you (sg.) him her it you (p1.)/them

Output Group
Fill in the missing blanks with the corresponding Spanish direct
object pronoun for the pronoun given in parentheses.

Example: Yo lo tengo. (him)

1._____visitas conmigo (her)
2. Nosotros_____llevamos. (you, sg.)
3. Ellos_____reciben. (them, masc.)
4. Yo_____lavo. (it, masc)
5._____necesitan ellas. (you, p1. masc)
6. Nunca_____llamas tu. (me)
7. Ellas_____buscan en la cocina. (them, masc)
8. Nostros no_____invitamos. (them, masc)
9. El nunca_____visita. (you, sg.)
10._____escuchas tu. (us)

Note how the sample input activity differs from the activities presented earlier
from VanPatten and Cadierno’s study. Learners are not pushed to process form
correctly to get meaning here. In fact, this activity does not even require any
attention to meaning.
The assessment tasks for this study were a comprehension task and a
translation/fill in the blank production task. DeKeyser and Sokalski found that
for object pronouns, the input group was better than the output group on the
comprehension task but that the output group was better than the input group on
the production task. These results were not maintained on a delayed posttest. For
the conditional, there was an overall advantage for the output group in both tasks
but results were not maintained on delayed posttests. Based on these results,
DeKeyser and Sokalski concluded that PI is not superior to TI.
56 PROCESSING INSTRUCTION

This conclusion, however, is not sustainable. As discussed in VanPatten

(2002), it must be underscored that the input-based treatment group in DeKeyser
and Sokalski was not PI. This becomes evident when we examine the
instructional treatment used in their input-based group. Recall that the goal of PI
is to alter learners’ processing strategies so that they process input better.
Therefore, for instruction to be PI, the processing problem or inefficient strategy
that learners use to process the target form must first be identified. In DeKeyser
and Sokalski’s study, there was no mention of any strategy that the researchers
were trying to circumvent through their instructional treatment. Given that there
was no strategy to circumvent, the practice activities were consequently not
designed to push learners away from their natural processing strategy.
Therefore, the practice activities in the input-based group cannot be classified as
SI activities and would not necessarily have the same effect that SI activities
would if they did what they were supposed to do. Furthermore, the sample
activity taken from DeKeyser and Sokalski shows that another of the essential
guidelines for creating SI activities was not adhered to: “Keep meaning in
focus.” For an activity to be an SI activity, that activity must require that
learners process the input for both meaning and form because learners should be
pushed to rely on sentence structure or form to derive meaning. However, as
DeKeyser and Sokalski pointed out themselves, not all the activities in their
input-based instructional treatment required learners to attend to meaning.
Therefore their study did not replicate VanPatten and Cadierno’s study. They
were essentially comparing their version of an input-based instruction to their
version of output-based instruction.

Salaberry(1997)
Salaberry (1997) also claimed to have evidence to refute the findings of
VanPatten and Cadierno (1993). His target form was also Spanish object
pronouns. Salaberry compared two experimental groups, that is, input
processing and output processing, to a control group and found no differences
between the two treatment groups. Therefore, he concluded that PI is not better
than TI. However, an examination of the treatment groups reveals that the
instruction the input processing group received was input-based but did not
contain SI activities that attempted to push learners away from faulty processing
strategies. It is clear that the researcher misinterpreted the guidelines in Lee and
VanPatten (1995) because he incorrectly cited the guidelines as saying that the
SI activities should move from mechanical to communicative language:

For example, Lee and VanPatten (1995) argue that in IP

instruction teachers should (a) present one thing at a time
(sequencing) and (b) move from sentences to connected
 2. THE NATURE OF PI 57

discourse (from mechanical to communicative language).

(Salaberry, 1997, p. 426)

What the guidelines actually say is that because meaning must be kept in focus
at all times, “learners should not engage in the mechanical input activities of
traditional grammar instruction” (Lee & VanPatten, 1995, p. 104). In short, the
instructional treatment in Salaberry (1997) was not PI and his activities were
clearly not trying to circumvent a processing strategy because the only
difference between his two treatment groups was that one set of activities
required participants to produce the target form while the other set did not. It is
no surprise that Salaberry obtained results different from those of VanPatten and
Cadierno. If the reader compares the VanPatten and Cadierno activities with
those of Salaberry (and those of DeKeyser and Sokalski), it is clear that we are
dealing with quite different treatments in terms of their underlying purpose.

Allen (2000)
Allen (2000) claimed to have evidence that conflicts with the findings of
VanPatten and Cadierno (1993). Her target structure was the French causative
with faire. The processing problem associated with this structure is a word order
problem. The verb faire means “to do” or “to make” in French but is also used in
causative constructions to say that someone is causing a behavior in someone
else:

(1) Jean fait laver la voiture à Marie,

Jean makes to wash the car to Mary.
John makes Mary wash the car.
(2) Mes professeurs me font travailler beaucoup
My professors me make work a lot.
My professors make me work a lot.

In (1), the subject of the first verb fait is Jean. The subject of the second verb
laver is Marie, obligatorily placed postverbally and marked by the preposition à.
It is the subject of the second verb that poses problems for learners of French.
When learners are asked to process sentences such as (1) and are asked “Who
washes the car?” they tend to say “Jean.” They would translate the sentence into
something like “John washes the car for Mary” (based on pilot data from Allen,
2000).
In (2) the subject of the second verb appears preverbally. However, in this
case, it is not a subject pronoun but an object pronoun. When learners are asked
to provide a rough translation of (2), they tend to say something “My professors
work hard for me.”
58 PROCESSING INSTRUCTION

These examples illustrate learners’ reliance on the first noun principle. In (1),
learners tend to think that Jean washes the car because he is the first noun that
appears before the verb. In example (2), they incorrectly assume the professors
work a lot because professeurs is the first noun before the verb.
Allen (2000) compared a control group to two experimental groups: a PI
group and a TI group. A sentence-level interpretation task revealed that both the
PI and the TI groups made gains and they were not different from each other. On
a sentence-level production task, results showed that both groups made gains but
the gains made by the TI group were higher. Therefore, Allen concluded that the
results of VanPatten and Cadierno are not generalizable to the causative
structure.
Both the results and conclusions in Allen’s study are questionable vis a vis
what PI is and what it means to assess the effects of PI. Unli ke DeKey ser
Sokalski (1996) and Salaberry (1997), Allen does identify a processing strategy
that must be overcome, that is, the first-noun strategy, and she does attempt to
help learners move away from this strategy through her activities (i.e., she
reminds learners in her activities not to rely on the first-noun strategy).
However, her SI activities are problematic because they do not force learners to
rely on sentence structure to get meaning. First, as exemplified in Activity A
below, only causative sentences were used in her activities. Thus, her activities
did not push learners to make a distinction between causative and non-causative
sentences with faire (see the discussionin VanPatten and Wong, this volume).
Therefore, we cannot be sure what learners were actually learning. It is possible
that they were being led to memorize a pattern with these activities rather than
rely on sentence structure for meaning.

Activité A. Select the correct interpretation of the sentences. Circle a or

b. Keep in mind that the word order in French is not exactly the same as
English.

1. Tom fait faire les valises à

Marc.
a. Tom packs the bags. b. Tom gets Marc to pack the
bags.
2. Les enfants font faire du bateau
à leurs parents.
a. The children go boating. b. The children get their parents
to go boating.
3. Je fais faire du vélo à ma soeur.
a. I get my sister to go bike b. I go bike riding.
riding,
 2. THE NATURE OF PI 59

4. Tu fais faire de l’alpinisme à ton

ami.
a. You get your friend to go b.You go mountain climbing.
climbing.
5. Jason fait faire de la
gymnastique à son frère.
a. Jason does gymnastics. b. Jason gets his brother to do
gymnastics.
6. Le professeur fait faire des
devoirs aux élèves.
a. The teacher does homework. b. The teacher makes the
students do homework.
7. Sara fait faire un voyage à ses
parents.
a. Sara gets her parents to take a b. Sara takes a trip.
trip.

A second problem has to do with the response choices for each practice item. In
Activity A, learners must select the correct interpretation for each sentence.
Note that there are two possible choices: a or b. One choice mentions only one
person and that person is the one performing the action. The other choice
mentions two people; one of the people mentioned is having the other person do
the action. These choices are problematic because they allow learners to
successfully complete the activity simply by matching names in the response
choices to the sentence. For example, in item l, Tom fait faire les valises à Marc,
we see two names: Tom and Mark. The correct answer is b: Tom gets Marc to
pack the bags. Note that the learner does not even have to know any French to
successfully do this item; response b has to be the correct answer here because
only b contains both names, Tom and Mark. The same goes for the rest of the
items. In item 2, we see Les enfants font faire du bateau a leurs parents. The
correct response is b: The children get their parents to go boating. Again, b has
to be the correct choice because only b mentions both the children and the
parents. Thus, we cannot tell if the learners selected this response because they
successfully processed this sentence or simply because only b mentions the two
people in the sentence. A third problem of this activity has to do with event
probabilities. For example, in item 6, learners see Le professeur fait faire des
devoirs aux élèves. The correct response is b: The teacher makes the students do
homework. In the real world, it is almost always professors who make students
do homework and not the other way around. Therefore, learners may select b as
the response because it is more likely that professors get students to do
homework and not because they have made any kind of form-meaning
60 PROCESSING INSTRUCTION

connection from the structure (not to mention the fact that only response b
mentions both professors and students). Therefore, it should be clear that this
activity cannot be considered an SI activity because learners are not pushed to
rely on sentence structure to get meaning. Because the goal of SI activities is to
get learners to process sentence structure for meaning, we must remove things
from the activities that may allow learners to give correct responses without
relying on sentence structure. Similar problems can be observed in subsequent
referential activities in Allen’s study. Other problems include requiring learners
to process the full verb paradigm rather than on focusing on one form at a time
(Guideline 1). Learners were required to pay attention to all conjugations of the
present tense of the verb faire rather than focusing on one form (e.g., Je fais, tu
fais, Paul fait, nous faisons, etc).
In VanPatten and Wong (this volume) we replicated Allen’s study with SI
activities that did push learners to process sentence structure in order to get
meaning and obtained different results. In fact, our results were similar to those
of the original VanPatten and Cadierno (1993) study. Results revealed that on
the interpretation task, the PI group improved significantly more that the TI
group from the pre-to posttest The no-instruction control group did not improve.
On the production task, both the PI and TI groups made gains from pre- to
posttest and there was no difference between the two groups. The control group
showed no improvement. All results were maintained on the delayed posttests
(see VanPatten & Wong, this volume). The following is a referential SI activity
taken from VanPatten and Wong’s study:

Listen to each sentence, then indicate who is performing the action by

answering each question.
1. Who cleans the room?___________
2. Who packs the bags?____________
3. Who watches the movie?_________
4. Who plays the flute?____________
5. Who does the dishes? ___________
6. Who buys wine? _______________
7. Who watches the show? _________
8. Who reads the instructions? ______

INSTRUCTOR’S SCRIPT
Read each sentence once. After each sentence, ask for an answer. Do not
wait until the end to review answers. Students do not repeat or otherwise
produce the structure.
1. Claude fait nettoyer la chambre à Richard.
2. Marc fait les valises pour Jean.
3. Sandra fait voir le film à Pierre.
 2. THE NATURE OF PI 61

4. Louis fait jouer de la flute à Suzanne.

5. Georges fait la vaisselle pour Louis.
6. Louise fait acheter du vin à Diane.
7. Ma mère fait regarder le spectacle à mon père.
8. Sally fait lire les instructions à Jean Luc,

Notice that in this SI activity, both causative and non-causative sentences with
faire are used so learners have to pay attention to respond correctly (item 2 and
item 5 are not causative). Furthermore, unlike the translation task used in
Allen’s activity presented above, this activity requires learners to indicate who is
performing the action so learners cannot simply match names to get the correct
response. Rather than present the whole verb paradigm, only one form, the third
person singular, was presented. Finally, event probabilities were controlled for.
In each item, either person mentioned could logically be performing the action.
In this way, students are required to rely on sentence structure in order to
determine who is doing each action.
The following example is an example of a written affective activity taken
from VanPatten and Wong.

In this activity you will compare and contrast what someone gets
a child to do with what someone gets a dog to do. For each item,
indicate whether it refers to the small child (à l’enfant), the dog
(au chien), or possibly both (à tous les deux).
Un adulte…
1. fait chercher I’os (bone) à/au_______.
2. fait faire la vaisselle à/au_______.
3. fait manger les restes à/au_______.
4. fait jouer du piano à/au_______.
5. fait prendre un bain à/au_______.
6. fait dormir au plancher (floor) à/au_______.
7. fait se comporter bien (behave) à/au_______quand il y a des
invites.
8. fait boire du lait à/au_______.

Does everyone in class agree?

INSTRUCTOR’S SCRIPT
“Ok. On va voir. Numero 1. Un adulte fait chercher 1’osse à
qui? Sharon?” “Au chien.” “OK. Tout le monde est d’accord?”
(students respond) “Et qu’est-ce qu’un adulte fait chercher a son
enfant?” Students may volunteer things like their books, a toy,
62 PROCESSING INSTRUCTION

etc.; students do not say or repeat the verbs. They just provide
nouns and so on.

VanPatten and Wong concluded from their study that there is evidence to show
that PI is generalizable to the causative structure. When activities are structured
to push learners to make the necessary form-meaning connections, they do
successfully alter learners’ inefficient processing strategies.

SUMMARY AND CONCLUSION

The purpose of this chapter was to describe in detail the nature of PI, the only
type of focus on form instruction to date that is informed by the strategies that
learners use to initially parse input to make form-meaning connections.
Research has shown that this type of grammar instruction is effective in helping
L2 learners adopt better processing strategies so that they can derive richer
intake from input. I described the three characteristics of PI as well as the
procedures and guidelines to follow in the development of SI activities.
Particular attention was paid to Guideline 6: “Keep learners’ psycholinguisti
processing strategies in mind.” This point is underscored because it is evident
from some attempted replication studies that PI has sometimes been mistakenly
reduced to be any kind of input practice. To reiterate, not all input-based
activities are SI activities and not all input-based instruction is PI. For
instruction to be PI, the processing strategy that learners use to process a
particular form must be identified. In order for an activity to be an SI activity,
that activity must be designed with that ineffective strategy in mind so that the
activity can help learners use more efficient strategies to process the input.
Furthermore, an SI activity must make learners process form correctly to get
meaning. If learners do not need to pay attention to meaning or if they do not
need to rely on form to get meaning, the activity is not an SI activity (e.g., see
the previous discussion regarding Allen, 2000).
To be clear, I am not saying that other activities that are not designed to
preempt an ineffective processing strategy cannot be useful. However, it should
be evident that the more we know about what learners do with input, the better
we will be at helping them process input better. Because SI activities are
designed with learners’ processing strategies in mind, they probably stand the
most chance at altering learners’ inefficient strategies so that optimal input
processing can take place. Creating activities without first identifying a
processing strategy is like a doctor passing out medication without knowing
what is wrong with the patient. Sometimes it may work and sometimes it won’t
but we won’t know why. With PI, the doctor (researcher) always knows why.
 Chapter 3
Commentary: What to Teach? How to Teach?

Patsy M. Lightbown
Concordia University, Montreal

Some years ago, I argued that SLA research could not tell teachers what to teach
or how (Lightbown, 1985a). In the 1970s and early 1980s, SLA research had
touched on only a few isolated language features, and there was no general
framework for deciding which of the nearly innumerable details of a language
could or should be the focus of instruction. Furthermore, I suggested that,
although there was much in SLA research to support the movement away from
rigid audio lingual instruction to more communicative language teaching, it
seemed that such trends regarding how to teach were confirmed—not caused—
by SLA research. Over the years, many SLA researchers have emphasized their
separateness from language pedagogy, insisting, with considerable justification,
that SLA was a field in its own right, and that it was appropriate for SLA
researchers to seek to understand how language is acquired, without always
having to answer the question, “but what does that say about teaching?” They
left the pedagogy to others. Bill VanPatten was an exception to this. His
research and that of a circle of colleagues and graduate students always had one
foot firmly planted in the classroom. That commitment to improving second
language teaching has not changed and the chapters that are the subject of my
commentary provide a rich resource for teachers, researchers, and
teacher/researchers whose goal is to find ways to help classroom language
learners get beyond roadblocks that limit their progress in second language
development.
The questions of what input is available to LI and L2 learners, as well as how
they perceive and process that input, have been a focus of my research for a long
time (Lightbown, 1980, 1983, 1984, 1985b, 1987, 1991, 1992a, b; 2001;
Lightbown & d’Anglejan; 1985, Lightbown et al, 2002). In 1991, my colleagues
Nina Spada and Lydia White and I organized a colloquium in Montreal on the
role of instruction in SLA. We brought together researchers who agreed that
instruction could play a role in second language learning, but who had quite
different interpretations of that role. The study that Bill VanPatten presented
appeared as VanPatten and Cadierno (1993) in the issue of Studies in Second
Language Acquisition (Volume 15, No. 2) devoted to papers from that
colloquium. The pedagogical implications of his ideas about focussed input
became an important component of every SLA course I taught thereafter—to
undergraduates in teacher training courses and to post-graduate students with a
68 PROCESSING INSTRUCTION

research focus. So, for years, I have reviewed and discussed these ideas
(VanPatten shows up in the reference lists of nearly everything I’ve published
since that time). I have found his ideas to be both useful and provocative, and
students have always found the work fascinating and intuitively appealing. We
often say that the best way to learn something is to teach it. But preparing this
commentary on chapters 1 and 2 has led me to new appreciation of what seemed
so familiar.
I found that, in reading and rereading both previously published work and the
chapters I will comment on here, my understanding of IP and PI has deepened.
With that deeper understanding of both the theoretical background and the
details of the research methods and findings have come a greater awareness of
the specificity of the pedagogical proposals as well as new insights into the
possible limitations of PI—limitations that have always been acknowledged by
VanPatten. In this commentary, I will reflect on some aspects of the potential
value of PI for second language pedagogy, especially in “the acquisition-rich
classroom” (VanPatten, 1993).
All theories of second language acquisition recognize a crucial role for input
(Gass, 1997). There are disagreements, however, about a number of issues,
including frequency, salience, and comprehensibility of the input. The
acquisition of some language features is undoubtedly affected by their frequency
and salience in the input that learners are exposed to. While there is much
evidence that low-frequency features are difficult to acquire (Ellis, 2002), there
is evidence that some high-frequency features with low salience and/or high
redundancy (VanPatten, this volume, chap. 1) are also late and/or difficult
acquisitions. The salience of features is affected by a number of things—
whether they are full syllables or bound morphemes (Wode, 1981), whether they
are in initial, medial, or final position in sentences (Meisel, Clahsen, &
Pienemarm, 1981; Pienemann, 1999; VanPatten, this volume). First language
patterns also affect learners’ ability to perceive patterns in the L2 lexicon and
morphosyntax (e.g., Kellerman, 1983; Schachter, 1974; Zobl, 1980).
With regard to comprehensibility, Krashen (1985, 1994) argues that input
that is comprehensible and that contains language features just beyond the
learner’s current interlanguage will, if affective conditions are right, provide the
necessary and sufficient conditions for acquisition to proceed. Sharwood Smith
(1986) distinguished between input for comprehension and input for acquisition,
suggesting that not everything that is available in the input becomes intake for
acquisition or leads to changes in a learner’s interlanguage. White (1987)
suggested that comprehensible input might lead learners to conclude that they
already know what they need to know. She argued that it is when input is
incomprehensible that learners realize they need to pay attention to formal
aspects of the language (see also Carroll, 1999). VanPatten (1990) showed that
learners had difficulty focussing on certain types of language form while their
attention was oriented primarily toward meaning.
 3. WHAT TO TEACH? HOW TO TEACH? 69

In second language classrooms where language is seen primarily as a carrier

of information rather than as the object of instruction in itself (content-based
instruction, immersion instruction, certain types of “communicative language
teaching”), there is ample evidence that learners can successfully understand
much of what they hear and read, without acquiring all the language features
that are present in the input. Some of the features that learners fail to acquire
through activities that are primarily meaning-based prove to have low frequency
in the classroom input (Lyster, 1994; Swain, 1988). Others, however, may have
high frequency, but high redundancy and low salience, allowing learners to
ignore them as long as contextual cues or event probabilities provide enough
information to permit accurate guessing. This suggests that, in much language
interaction outside the classroom and in meaning-based instruction,
comprehension may interfere with acquisition in the sense that learners are
misled into thinking that they have processed and understood language form,
when in fact, they have relied on contextual or cotextual cues (see Batstone,
2002). As VanPatten points out (2002; this volume, chap. 1), it is only when
learners are confronted with the necessity of using those non-salient features to
grasp meaning that they come to realize that they have either misunderstood
what they heard or read or that they have understood the general idea without
knowing how language encoded the meaning. VanPatten has proposed
processing instruction as one way to increase learners’ chances of making
correct form/meaning connections.
For nearly twenty years, most of my SLA research has been done in language
teaching/learning contexts where teachers were committed to a type of
communicative language teaching in which any focus on language itself other
than a gentle “recast” was considered counterproductive (e.g., Lightbown &
Spada, 1990). The instructional environment was one in which francophone
children, 11–12 years old, were enrolled in “intensive ESL classes” where, for a
period of five months of one academic year, they spent most of each school day
participating in activities in English. The emphasis was always on “meaning”
and it was expected that, with time and motivation, learners would eventually
acquire both fluency and accuracy in their use of the target language. Anything
that might be construed as “teaching grammar” was seen as potentially de-
motivating and, in any case, something that would take time away from the
more important activity—using the language in communicative interaction. It
would be hard to overstate the extent to which students in these classes (with the
encouragement of their teachers) confirmed VanPatten’s Principle 1 and its
subprmciples: Learners process input for meaning before they process it for
form; they process content words before anything else, and they tend to rely on
lexical items rather than redundant, non-salient grammatical morphemes and
function words (my paraphrasing). Activities were designed in such a way that
students were not only enabled, but actively encouraged to “go for meaning”,
70 PROCESSING INSTRUCTION

not worrying about the details, that is, not worrying about function words or
grammatical morphology.1
Our research in these classes showed that, after several months of intensive
exposure to English (3–5 hours a day), most learners had either not noticed or, in
any case, had not incorporated into their developing interlanguage a number of
high frequency language forms. Two examples are inversion in question forms
(Spada & Lightbown, 1993) and the use of possessive determiners his and her
(Lightbown & Spada, 1990; White, 1998). We recently reported on a study in
which we introduced form-focussed instruction on these two language features
(Spada, Lightbown, & White, in press), and it is useful to review both the
experimental methodology and the findings of that study in light of VanPatten’s
principles of input processing and Wong’s (this volume, chap. 2) elaboration of
the “structured input” activities that are at the heart of processing instruction.
Before describing the research procedures and the findings of our study, I will
briefly describe these language features, suggesting the processing problems that
might explain students’ difficulty in acquiring them.
For French-speaking learners, the choice of the correct third person
possessive determiner is complicated by the fact that in French, as in Spanish
and many other languages, the determiner agrees with the grammatical gender of
the possessed entity, not with the natural gender of the possessor.2 Thus, the
sentence J’ai posé sa règle sur son bureau could have four possible translation
equivalents in English:
I put his ruler on his desk.
I put her ruler on his desk.
I put his ruler on her desk.
I put her ruler on her desk.
Only contextual information tells us whether the possessor of the ruler and the
desk is the same person and whether his/her/their gender is masculine or
feminine. Even after French-speaking learners of English have started to use
both his and her (after an initial period of using the definite article or an
undifferentiated possessive determiner—often “your”), they run into trouble
when the possessed entity has natural gender. That is, they may produce
sentences such as The little girl talk to his father, choosing his to agree with the
natural gender of the father (Lightbown & Spada, 1990; White, 1998; Zobl,
1985). White and Ranta (in press) show that students’ explanations for their
choices confirm that they are operating on the basis of the French rule for
choosing the gender marking of the possessive determiner.

1
Similar findings have been reported in other instructional environments, including
French immersion (Lyster, 1994, 1998).
2
In fact, there is a further complication in the fact that, in referring to body parts,
French often uses the definite article rather than a possessive form. (See Spada,
Lightbown, & White, in press, and White, 1998, for further treatment of this aspect of the
learners’ development.)
 3. WHAT TO TEACH? HOW TO TEACH? 71

With regard to inversion in questions, we had found in previous studies

(Spada & Lightbown, 1999) that learners had relatively little difficulty in
learning to recognize the ungrammatically of questions without subject-
[auxiliary] verb inversion when the subject was a pronoun. However, they
continued to accept questions without inversion, if the subject was a full lexical
noun. For example, most accepted Can we watch television? and Why do you
want to go outside?, but at the same time, they accepted Why fish can live in
water? and *Why children like McDonald’s? as grammatical. We hypothesized
that the problem lay in learners’ assumption that English, like French, does not
allow inversion with noun subjects. That is, Peut-il jouer dehors? (literally,
Wants-he to-play outside?) is grammatical, but *Peut-Jean jouer dehors?
(literally, Can John to-play outside?) is not.3
There are several reasons why learners might “hear” English as allowing
pronouns but not nouns in inversion. They may have filtered the input on the
basis of a pattern that they know from their first language (Lightbown, 2001).
VanPatten points out, quite correctly, that the existence of a pattern in L1 is not
a sufficient basis for predicting L2 developmental patterns. For one thing,
learners need to reach a certain level of development before they can even
recognize similarities and differences between the target language and
previously learned languages. Nevertheless, it has been shown that L1 interacts
with developmental sequences in a variety of ways—speeding or slowing
progress, constraining generalization, adding sub-stages (Zobl, 1980). In the
case of these students, they had reached a stage where they recognized the
pattern of inversion in English questions that matched the French pattern, but
could not see that it extended to nouns as well as pronouns.
Another possible reason for learners’ failure to recognize the grammaticality
of inversion with noun subjects is that, in informal speech, the auxiliary verb
often has very little salience. Having had little or no form-focussed instruction,
the students did not realize that in both yes/no and wh-questions in English, a
verbal element must precede the subject, although the lexical verb remains in
its post-subject position (Indeed, I assume this is what VanPatten is referring to
when he says that English has predominantly SVO order, even for questions.)
We can see evidence that students simply do not hear the auxiliary preceding the

3
It has long been noted that inversion is rare in informal spoken French (Lightbown,
1980; Picard, 2002). However, students of this age will have had considerable exposure
to written French and to more formal varieties of spoken French. They will also have
been taught to use inversion in their own writing. Therefore, when they are asked to
judge the grammaticality of sentences, we may assume that their judgements are based on
their knowledge of more formal varieties as well as their own informal spoken French.
72 PROCESSING INSTRUCTION

subject in the changes that some students made to sentences such as *Why
children like McDonald’s? They “corrected” the question by inserting the,
yielding *Why the children like McDonald’s? This seemed to suggest that if
they heard the do at all in the many correct questions present in the classroom
input, they heard it as a plausible pronunciation of the definite article, a required
element in the corresponding question in informal spoken French Pourquoi les
enfants aiment McDonald’s? In normal language, delivered at a normal rate, the
difference between “Why do children…” and “Why the children…” is minimal,
and no doubt extremely difficult for low proficiency learners to hear.
Furthermore, failure to recognize or produce the grammatical auxiliary in pre-
subject position is very unlikely to interfere with communicative effectiveness.
Not surprisingly, this type of error persists in the English of students even after
long periods of exposure to the language.
Thus, in this study, we had identified two problems that we knew to be long-
term challenges for francophones learning English, and we designed an
instructional intervention that was intended to help them improve their ability to
recognize these features and to use the correct forms in their own production.
Looking back on this study in terms of processing instruction, several things are
worthy of note. We did follow Step 1 in Wong’s list of the steps in designing
structured input activities by identifying a plausible strategy to explain how
learners were misinterpreting the input and using their incorrect interpretation to
shape their output. We also followed the first guideline in Step 2, providing the
teachers with activities that focussed on “one thing at a time.” Students in both
groups (those being taught question inversion and those being taught possessive
determiners) were told what the problem was. That is, they were shown how-
French and English differ with regard to the feature they were learning, and they
engaged in activities that kept the focus on these features.4
To a certain extent, some of the remaining guidelines in Step 2 were also met
by the instructional activities. Both oral and written input were provided, and
students performed a variety of tasks that required them to “do something with”
the features in focus. The learners’ processing strategies, as we understood them,
were the basis of all the materials, and teachers were asked to provide feedback
when students showed that they were relying on the incorrect strategies.
Nevertheless, the study was not planned in terms of the guidelines for the
structured input required for processing instruction, and the fit is not perfect.
4
The study had been planned as an opportunity to compare the learning success of
students who were told about the French/English contrasts and those who were simply
taught the English pattern. After the study was under way, we realized that the teachers in
the intact classes participating in the study had, either before or during the experimental
treatment, provided their students with information about the contrast. Thus, all students
had at least some direct instruction that drew their attention to the processing problem
they needed to overcome.
 3. WHAT TO TEACH? HOW TO TEACH? 73

There was not strict adherence to the guideline to move from single-sentence
focus to discourse focus, and activities included opportunities for students to
produce as well as hear and read the target features. The most significant point
where our instructional intervention differed from processing instruction was
with regard to the guideline “Keep meaning in focus.” The overall emphasis of
the classroom activities (including those that were part of the experimental
intervention) was almost always on meaning, and using language in context.
However, there was a difference in meaning focus for the two language features.
For the activities in which possessive determiners were the target feature,
success in carrying out the various tasks, answering questions, and playing
games usually required the correct use and interpretation of his and her.
However, while a learner’s misinterpretation of a possessive determiner or use
of the wrong one in production can lead to misunderstanding, there is little
likelihood that failure to invert in questions led to communication difficulty or to
an inability to continue with the tasks. Thus, in the activities focused on
inversion, the instructional intervention targeted correct form rather than crucial
form/meaning relations.
In pretest/posttest comparisons, we found that the experimental intervention
was very effective in changing learners’ ability to interpret and use possessive
determiners. Changes in their recognition of grammaticality in questions and
their use of inversion in questions were also affected by the instruction, but the
effect was less substantial. We offered several possible interpretations for this
finding (see Spada, Lightbown, & White, in press). The one that is relevant here
is that the instruction in question inversion was not based on a fundamental
form/meaning connection, while instruction on possessive determiners was.
Although it is important to emphasize again that this study was not designed to
replicate or mimic PI instruction, the differences in outcomes seem to be quite
consistent with the predictions that VanPatten and Wong would make regarding
the effectiveness of the instructional intervention. The inference that I draw from
this is that one of the limitations of PI will be in its relative effectiveness in
helping learners overcome problems with language features that lead to errors
that interfere with meaning, compared to those that do not. In previous research,
we have seen learners improve their accuracy in recognizing and using
questions, but these studies have involved relatively long periods of
intervention, explicit focus on accuracy, and sustained teacher feedback (Spada
& Lightbown, 1993; White, Spada, Lightbown, & Ranta, 1991).
VanPatten has been clear that PI is aimed at bringing learners to process a
form/meaning relationship that they have either not previously processed or that
they have processed incorrectly. Not only is there no claim that PI leads to
flawless production in spontaneous communication, there is not even a claim
that the intake that results from PI leads to a change in the structure of a
learner’s interlanguage, PI is proposed as a first step, a way to start something
74 PROCESSING INSTRUCTION

that needs many more components before the feature in focus becomes a natural
part of the learner’s use of the second language.
Most experimental PI instruction (like the instruction in most other
experimental SLA studies) has taken place over a very short period of time—
minutes, hours, days. And, by definition, instruction and exposure to the target
features have been carefully controlled. It may be that longer periods of PI—or
frequent refresher lessons that remind learners of their processing problem—
would eventually lead to correct, fluent use. However, it seems more likely that
a great deal more time and a great many more opportunities to retrieve and use
the feature in communicative settings are likely to be necessary.
Processing instruction is a valuable first step in helping learners to make
connections between language forms and their meanings or their uses. Once this
step has taken place, learners need to encounter the language features again and
again in discourse-rich contexts where their use is appropriate. Psychological
research on transfer-appropriate processing suggests that we recall something
most easily when we are in the same situation in which we learned it (see
Segalowitz & Lightbown, 1999). VanPatten repeatedly emphasizes that the
processing that takes place in the narrow scope of PI is not, in itself, learning
(nor, to acknowledge Krashen’s contrast, is it acquisition). Rather, this initial
processing permits recognition, practice, and learning to take place subsequently
in “transfer-appropriate” contexts (see also Schmidt, 1992).
Throughout the learning process, classroom learners may also benefit from
sustained feedback on the errors in their use of many language features, even
after they have a good foundation—whether that includes formal grammatical
explanation or only structured input without explanation (see Sanz, this volume;
Wong, chap. 2, this volume). Sustained feedback is especially likely to be
important if the linguistic feature is one of those that does not affect meaning—
either in general, or an environment where students (and teachers) share the
same LI and many interlanguage patterns. This brings me to a brief comment
about the example that VanPatten gives for the effectiveness of a certain type of
feedback. In the “Bob and Tom” example, a clarification request (in the form of
a recast with rising intonation) seems to lead the second language speaker to
notice something about the way the target language works. There’s plenty of
evidence to show although such “conversational” recasts sometimes alert
learners to differences between their original version of a sentence and the one
offered in the recast, they may also pass unnoticed outside the classroom or in
classrooms where the learners (and the teacher) have the habit of focusing
primarily on meaning (Lyster, 1998). To date, the research suggests that recasts
are most effective when they include some signal to make it clear that the recast
focuses on form rather than (or in addition to) meaning (Doughty, 1999).
Recasts have been shown to work in laboratory dyads where interaction is very
focused or in classrooms with small numbers of adult learners where the
instruction includes a substantial metalinguistic component (see Nicholas,
 3. WHAT TO TEACH? HOW TO TEACH? 75

Lightbown, & Spada, 2001 for review). Even in these contexts, learners may
misread recasts, especially if they are offered by classmates (Morris, 2002).
A question that inevitably comes to mind regarding not only PI research but
also any research on assisting language learning is the extent to which it can be
adopted or adapted for use by classroom teachers. The research findings are
sufficiently clear to persuade me that PI is an effective tool for helping learners
zero in on form/meaning relationships that they have previously overlooked or
misinterpreted. In translating the research findings to classroom application, I
am aware of how wishful thinking leads some educators to hope that there is a
magic formula that will work across the board for solving pedagogical problems.
I am reminded of the strong reservations I have expressed elsewhere about the
possibility of planning teaching sequences to match developmental sequences
that have been observed in second language acquisition (Lightbown, 1985c;
1998). VanPatten and Wong have been very clear that they do not see PI as the
best or only approach to teaching all language features. They have also left no
doubt that PI is not proposed as the basis for taking learners all the way to
spontaneous, accurate, automatized production of any language features. In
reflecting on the role of PI in “ordinary classrooms”, I’d like to focus on two
concerns. First, how can we identify language features that are (or are not) good
candidates for the PI approach? Second, are teachers prepared to use this tool in
their classrooms?
There are probably many language features that are not good candidates for
PI instruction. This would include those that learners acquire without apparent
difficulty while they engage in interactive communicative language. At the other
end of the continuum are those that learners continue to have difficulty with,
because of their inability to distinguish between correct and incorrect language
forms. These are features that do not ordinarily lead learners to misinterpret
what they hear and read, and accuracy in producing these features leads to a
more polished performance rather than to changes in the meaning. Although
some of the components of PI may be effective in helping students notice the
difference between correct and incorrect versions of a language feature, a full-
scale application of the guidelines that Wong proposes is either impossible or
requires an unreasonable amount of effort at creating structured input. One
example of this may be the inserted auxiliary do in questions.5 Another would be
the kinds of complex syntactic features that are often the focus of SLA research
with a Universal Grammar orientation. The finding that structured input works
as well as structured input plus explanation (VanPatten & Oikkenon, 1996;
the various papers in Part 3 of this volume) may be the key to whether PI can be

5
VanPatten (1996) has suggested, however, that PI might be used to help learners
process do as a carrier of tense and number information, thereby making the form more
salient and possibly available for further processing and learning. (See VanPatten, this
volume, chap. 17 as well.)
76 PROCESSING INSTRUCTION

used to guide learners’ acquisition of more complex linguistic features as well as

features in which errors do not lead to problems with interpreting meaning. It
also shows how PI can be effective with learners whose lack of metalinguistic
awareness or training (including young learners and those with low levels of
literacy) limits the usefulness of “explanations” that are given using unfamiliar
terminology.
Are teachers prepared to use the insights and procedures of PI in their
classrooms? Wong’s carefully detailed guidelines certainly improve the chances
that in future “replications” of PI, researchers can scrutinize the design of their
intervention studies in terms of their conformity to that approach. But, even with
these guidelines, is it reasonable to expect teachers to be up to the challenge?
Although university foreign language instructors may continue to receive
extensive training in both the grammar of the language(s) they are to teach and
SLA research findings, teachers working with younger learners as well as many
teachers of English as a second language to both children and adults have far
less knowledge of the language itself and of the language acquisition research.
Their backgrounds often include more training for teaching in general or for
foreign/second language teaching methods in particular (both of which are often
sadly lacking in the training of university foreign language instructors). My
experience with teachers working in second and foreign language instruction in
primary and secondary schools is that they are often unaware of the reasons
students make certain errors or have difficulty with particular language features.
Indeed, in some cases, they do not even notice the errors—either because they
make them themselves (in the case of teachers who have less than excellent
command of the second language) or because they look “through” the error to
the meaning that is their focus. This is not, of course, a problem with PI. Rather,
it is a potential limitation on its application in situations where it could be an
effective pedagogical tool if only teachers were equipped to use it.
For years, I told students that my SLA course was not a “teaching methods”
course and that I would not be telling them “what to teach or how to teach.” I
told them, however, that I hoped they would find the course “useful” in helping
them to set appropriate expectations for what their students could learn and what
they could teach within the context of classroom instruction. Over time, the
course content changed as more and more SLA researchers were doing research
with a pedagogical focus (see Lightbown, 2000, for a review of some of this
work). Some research that was not classroom-based also contributed
significantly to our understanding of why teaching didn’t always have the effect
teachers hoped it would have. Gradually, some of the work students read did
begin to point to answers to questions about what to teach and how. The work of
VanPatten and his colleagues has evolved to include some of the most explicit
responses to both these questions. The clear and detailed principles and
guidelines that are available in the VanPatten and Wong chapters will make a
further contribution to answering them.
 3. WHAT TO TEACH? HOW TO TEACH? 77

In recent years, I’v e beco me bol de r in cla imin g t hat the SLA course
makes a “practical” contribution to teacher education. Now I almost always end
the term by giving students a set of guidelines for classroom practice that I think
are supported by the research they have read about during the course. Knowing
that most of them will be teaching in contexts where the emphasis is on
meaningful interaction rather than form-focussed instruction, I suggest the
following guidelines.
1. Some linguistic features are acquired “incidentally”—without intentional
effort, conscious awareness or teachers’ guidance. The great news for
teachers: You don’t have to teach everything.
2. Some features may even be harder to acquire when learners are given
metalinguistic rules than when they are encountered (with sufficient
frequency) in meaningful interaction.
3. Some linguistic features are acquired according to developmental stages that
are not altered by form focused instruction. However, learners’ passage
through the stages may be speeded up by some form-focused instruction.
4. Some linguistic features are easier to acquire when learners’ attention is
focused on them (and explanations are provided).
5. Some linguistic features are difficult or even impossible to acquire without
focused attention (and corrective feedback).
6. Progress in language acquisition is not always marked by increases in
accuracy. Some other changes may also indicate that learners have begun to
move toward higher levels of proficiency.
7. It takes a great deal of time and effort for learners to reach a stage at which
they can use many language features accurately in spontaneous
communication. Along the way, they may require frequent reminders of what
they know but don’t always do.
8. Learners may not recognize or benefit from “corrective feedback” presented
in “conversational” style. In some cases, more explicit instruction and
feedback on error may be required.
I believe all of these guidelines are compatible with the principles and
applications of processing instruction and that PI can be an effective foundation
for the acquisition of many language features.

REFERENCES

Batstone, R. (2002). Making sense of new language: A discourse perspective.

Language Awareness, 11, 14–29.
Carroll, S. (1999). Putting “input” in its proper place. Second Language
Research, 15, 337–388,