Second language learning 43

discover and describe different kinds of errors in an effort to understand

how learners process second language data. Error analysis was based on the
hypothesis that, like child language, second language learner language is a
system in its own rig h t -o n e that is rule-governed and predictable.
Larry Selinker ( 1972) gave the name interlanguage to learners' developing
second language knowledge. Analysis of a learner's interlanguage shows that
it has sorne characteristics influenced by previously learned languages, sorne
characteristics of the second language, and sorne characteristics, such as the
omission of function words and grammatical morphemes, that seem to be
general and to occur in all interlanguage systems. Interlanguages have been
found to be systematic, but they are also dynamic, continually evolving as
learners receive more input and revise their hypotheses about the second
language. The path through language acquisition is not necessarily smooth
and even. Learners have bursts of progress, then reach a plateau for a while
before something stimulates further progress. Selinker also coined the term
fossilization to refer to the fact that sorne features in a learner's language seem
t o stop changing. This may be especially true for learners whose exposure to
the second language does not inelude instruction or the kind of feedback that
would help them to recognize differences between their interlanguage and
the target language.

AC T I V I T Y Analyse learner language

The following texts were written by two learners of English, one a French
speaking secondary school student, the other a Chinese-speaki ng ad ult learner.
Both learners were describing a cartoon film entitled The GreatToy Robbery
(National Film Board of Canada). After viewi ng the film, they were asked to
retell the story in writi ng, as if they were telli ng it to someone who had not seen
the film.
Read the texts and answer the followi ng questions:
Can you u nderstand what each learner is trying to say?
l Exami ne the errors made by each learner. What ki nds of errors interfere
most with you r ability to understand?
3 Do both learners make the same ki nds of errors?
4 In what ways do the two interlanguages differ?
Learner I: French first language, secondary school student
Du ri ng a sun ny day, a cowboy go in the desert with his horse. he has a big hat.
His horse eat a flou r. In the same time, Santa Clause go in a city to give sorne
surprises. He has a red costume and a red packet of surprises.You have th ree
robbers in the mou ntai n who sees Santa Clause with a ki ng of glaces that it
permitted us to see at a long distance. Every robbers have a horse.They go in
44 Second language learning

the way of Santa Clause, not Santa Clause but his pocket of surprises. After
they will go in a city and they go in a saloon. [...]
(Unpu blished data from P. M.Lightbown and B.Barkman)
Learner 2: Chinese first language,adult
This year Christmas comes soon! Santa Claus ride a one horse open sleigh to
sent present for child ren. on the back of his body has big packet. it have a lot of
toys. in the way he meet th ree robbers.They want to take his big packet. Santa
Claus no way and no body help, so only a way give them, then three robbers
ride their horse dashing through the town.There have saloon, they go to d ri nk
sorne beer and open the big packent.They plays toys in the Bar.They meet a
cow boy in the saloon.
(Unpublished data provided by M. J.Martens)

Perhaps the most striking thing here is that many error types are common to
both learners. Furthermore, both make errors of spelling and punctuation
that we might find in the writing of a young first language speaker of English.
Even though French uses grammatical morphemes to indicare person and
number on verbs and Chinese does not, both these learners make errors of
subject-verb agreement-both leaving off the third person -s marker and
overusing it when the subject is plural ('a cowboy go' and 'three robbers in the
mountain who sees' by Learner 1 and 'Santa Claus ride' and 'they plays' by
Learner 2). Such errors reflect learners' understanding of the second language
system itself rather than an attempt to transfer characteristics of their first
language. They are sometimes referred to as 'developmental' errors because
they are similar to those made by children acquiring English as their first lan
guage. Sometimes these are errors of overgeneralization, that is, errors caused
by trying to use a rule in a context where it does not belong, for example,the
-s ending on the verb in 'they plays'. Sometimes the errors are better described
as simplification, where elements of a sentence are left out or where all verbs
have the same form regardless of person, number, or tense.
One can also see, especially in Learner 2's text, the influence of classroom
experience. An example is the use of formulaic expressions such as 'one horse
open sleigh' which is taken verbatim from a well-known Christmas song that
had been taught and sung in his English as a Second Language (ESL) class.
The vivid 'dashing through the town' probably comes from the same source,
with the substitution of 'town' for 'snow'.
For those who are familiar with the English spoken by native speakers of
French, sorne of the errors (for example, preposition choice 'in the same
time') made by the first learner will be seen as probably based on French.
Similarly, those familiar with the English of Chinese speakers may recog
nize sorne word arder patterns (for example, 'on the back of his body has big
packet') as based on Chinese patterns. These may be called transfer or
 Second language learning 45

'interference' errors. What is most clear, however, is that it is often difficult

to determine the source of errors. Thus, while error analysis has the advan
tage of describing what learners actually do rather than what they might do,
it does not always give us clear insights into why they do it. Furthermore, as
Jacquelyn Schachter pointed out in a 1974 article, learners sometimes avoid
using sorne features of language that they perceive to be difficult for them.
This avoidance may lead to the absence of certain errors, leaving the analyst
without information about sorne aspects of the learners' develop ing
interlanguage. The absence of particular errors is difficult to interpret, and
the phenomenon of 'avoidance' may itself be a part of the learner's sys
tematic second language performance.

Developmental sequences
Second language learners, like first language learners, pass through sequences
of development: what is learned early by one is learned early by others.
Among first language learners, the existence of developmental sequences may
not seem surprising because their language learning is partly tied to their cog
nitive development and to their experiences in learning about relationships
between people, events, and objects around them. But the cognitive develop
ment of adult or adolescent second language learners is much more stable,
and their experiences with the language are likely to be quite different, not
only from the experiences of a small child, but also different from each other.
Furthermore, second language learners already know another language that
has different patterns for creating sentences and word forms. In light of this,
it is more remarkable that we find developmental sequences that are similar
in the developing interlanguage of learners from different language back
grounds and also similar to those observed in first language acquisition of the
same language. Moreover, the features of the language that are most frequent
are not always learned first. Far example, virtually every English sentence has
one or more anides ('a' or 'the'), but even advanced learners have difficulty
using these forms correctly in all contexts. Finally, although the learner's first
language does have an influence, many aspects of these developmental stages
are similar among learners from different first language backgrounds.
In Chapter 1 we saw sorne developmental sequences for English first lan
guage acquisition of grammatical morphemes, negation, and questions.
Researchers in second language acquisition have also examined these, as well
as other features. They have found patterns in the development of syntax and
morphology that are similar among learners from different language back
grounds. Evidence for these developmental patterns first carne from studies
oflearners whose primary learning environment was outside the classroom.
Far example, Jürgen Meisel, Harald Clahsen, and Manfred Pienemann
46 Second language learning

( 1981) identified developmental sequences in the acquisition of German by

speakers of several Romance languages who had little or no instruction.
Subsequent research has shown that learners who receive instruction exhibit
similar developmental sequences and error patterns. Inthe interlanguage of
English speakers whose only exposure to German was in university dasses in
Australia, Pienemann (1988) found patterns that were similar to those of the
uninstructed learners. In Chapter 6, we will discuss other studies that have
investigated the influence of instruction on developmental sequences.

Grammatical morphemes
Researchers have examined the development of grammatical morphemes by
learners of English as a second language in a variety of environments, at
different ages, and from different first language backgrounds. In analysing
each learner's speech, researchers identify the obligatory contexts for each
morpheme, that is, the places in a sentence where the morpheme is necessary
to make the sentence grammatically correct. For example, in the sentence
'Yesterday I play baseball for two hours', the adverb 'yesterday' creates an
obligatory context for a past tense, and 'for two hours' tells us that the required
form is a simple past ('played') rather than a past progressive ('was playing').
Similarly, 'two' creates an obligatory context for a plural -s on 'hours'.
For the analysis, obligatory contexts for each grammatical morpheme are
counted separately, that is, one count for simple past, one for plural, one for
third person singular present tense, and so on. After counting the number of
obligatory contexts, the researcher counts the correctly supplied morphemes.
The next step is to divide the number of correctly supplied morphemes by
the total number of obligatory contexts to answer the question 'what is the
percentage accuracy for each morpheme?' An accuracy score is created for
each morpheme, and these can then be ranked from highest to lowest, giving
an accuracy order for the morphemes.
The overall results of the studies suggested an order that was similar but not
identical to the developmental sequence found for first language learners.
However, the order the researchers found was quite similar among second
language learners from different first language backgrounds. For example,
most studies showed a higher degree of accuracy for plural -s than for posses
s,
sive - and for -ing than for regular past (-ed). Stephen Krashen summarized
the order as shown in Figure 2.1. The diagram should be interpreted as
showing that learners will produce the morphemes in higher boxes with
higher accuracy than those in lower boxes, but that within boxes, there is no
dear pattern of difference.
 Second language learning 47

-ing (progressive)
plu ral
copula ('to be')

t
auxiliary (progressive
as in 'He is going')
article

t
irregular past

t
regular past -ed
third person singular -s
possessive 's

Figure 2.1 Krashen's ( 1982) summary of second /anguage grammatical morpheme

acquisition sequence
The similarity among learners suggests that the accuracy arder cannot be
described or explained in terms of transfer from the learners' first language,
and sorne researchers saw this as strong evidence against the CAH. However,
a thorough review of all the 'morpheme acquisition' studies shows that the
learners' first language does have an influence on acquisition sequences. Far
example, learners whose first language has a possessive form that resembles
the English s (such as German and Danish) seem to acquire the English
possessive earlier than those whose first language has a very different way of
forming the possessive (such as French or Spanish). And even though ani
des appear early in the sequence, learners from many language backgrounds
(including Slavic languages, Chinese, and Japanese) continue to struggle
with this aspect of English, even at advanced levels. Learners may do well in
supplying articles in certain obligatory contexts but not others. If the lan
guage sample that is analysed contains only the 'easier' obligatory contexts,
che learner may have a misleadingly high accuracy score.
Another reason why something as difficult as English articles appears to be
acquired early is that the arder in the diagram is based on the analysis of
correct use in obligatory contexts only. Itdoes not take into account uses of
grammatical morphemes in places where they do not belong, for example,
when a learner says, 'The France is in Europe'. These issues led researchers
48 Second language learning

to question the adequacy of obligatory context analyses as the sale basis for
understanding developmental sequences. Teresa Pica (1983) argued that
accuracy seores should take account of overuse and incorrect uses to deter
mine a score for target-like use rather than reflect only use in obligatory
contexts.
The morpheme acquisition literature raises other issues, not least of them the
question of why there should be an arder of acquisition for these language
features. Sorne of the similarities observed in different studies seemed to be
due to the use of particular tasks for collecting the data, and researchers found
that different tasks tended to yield different results. Nevertheless, a number
of studies have revealed similarities that cannot be explained by the data col
lection procedures alone. As with first language acquisition, researchers have
not found a single simple explanation for the arder. Jennifer Goldschneider
and Roben DeKeyser (2001) reviewed this research and identified a number
of variables that contribute to the arder. Salience (how easy it is to notice the
morpheme), linguistic complexity (for example, how many elements you
have to keep track of), semantic transparency (how clear the meaning is),
similarity to a first language form, and frequency in the input all seem to play
a role.

Negation
The acquisition of negative sentences by second language learners follows a
path that looks nearly identical to the stages we saw in Chapter 1 for first
language acquisition. However, second language learners from different first
language backgrounds behave somewhat differendy within those stages. This
was illustrated in John Schumann's (1979) research with Spanish speakers
learning English and Henning Wode's (1978) work on German speakers
learning English.

Stage 1
The negative element (usually 'no' or 'not') is typically placed befare the verb
or the element being negated. Often, it occurs as the first word in the sen
tence because the subject is not there.
No bicycle.
I no like it.
Not my friend.
'No' is preferred by most learners in this early stage, perhaps because it is the
negative form that is easiest to hear and recognize in the speech they are
exposed to. Italian- and Spanish-speaking learners may prefer 'no' because it
corresponds to the negative form in Italian and Spanish (No tienen muchos
libros). They may continue to use Stage 1negation longer than other learners
because of the similarity to a pattern from their first language. Even at more
advanced stages, they may also use Stage 1 negatives in longer sentences or
 Second language learning 49

when they are under pressure. Thus, similarity to a learner's first language
may slow clown a learner's progress through a particular developmental stage.

Stage2
At this stage, 'no' and 'not' may alternate with 'don't'. However, 'don't' is not
marked for person, number, or tense and it may even be used befare modals
like 'can' and 'should'.
He don't likeit.
I don't can sing.

Stage3
Learners begin to place the negative element after auxiliary verbs like 'are',
'is', and 'can'. But at this stage, the 'don't' form is still not fully analysed.
You can not go there.
He was not happy.
She don't likerice.
At this stage, German speakers, whose first language has a structure that
places the negative after the verb may generalize the auxiliary-negative
pattern to verb-negative and produce sentences such as:
They come not [to] home. (Sie kommen nicht nach Hause. )

Stage 4
In this stage, 'do' is marked for tense, person, and number, and most interlan
guage sentences appear to be just like those of the target language.
l t doesn't work. We didn't have supper.
However, sorne learners continue to mark tense, person, and number on
both the auxiliary and the verb.
I didn't went there.

Questions
fanfred Pienemann, Malcolm Johnston, and Geoff Brindley (1988)
described a sequence in the acquisition of questions by learners of English
from a variety of first language backgrounds. An adapted version of the
sequence is shown in Stages 1-6 below. The examples (except those in Scage
6) come from French speakers who were playing a game in which they had
to ask questions in arder to find out which picture the other player
,(the researcher) was holding. As we saw for negation, the overall sequence is
similar to the one observed in first language acquisition. And again, there are
sorne differences that are attributable to first language influence.

Suzge 1
Single words, formulae, or sentence fragments.
50 Second language learning

Dog?
Four children?
What's that?

Stage2
Declarative word order, no inversion, no fronting.
It's a monster in the right comer?
The boys throw the shoes?
Declarative order with rising intonation is common in yes/no questions in
informal spoken French. French speakers may hypothesize that in English,
as in French, inversion is optional.

Stage3
Fronting: do-fronting, wh-fronting without inversion, other fronting.
Do you have a shoes on your picture?
Where the children are playing?
Does in this picture there is four astronauts?
Is the picture has two planets on top?
French has an invariant form est-ce que (literally 'is it that') that can be placed
before a declarative sentence to make a question. For example, ]ean aime le
cinéma becomes Est-ce que]ean aime le cinéma? ('is it that) John likes movies?'
French speakers may think that 'do' or 'does' is such an invariant form and
continue to produce Stage 3 questions for sornetime.

Stage 4
lnversion in wh- + copula; yes/no questions with other auxiliaries.
Where is the sun?
Is there a fish in the water?
At Stage 4, German speakers may infer that if English uses subject-auxiliary
inversion, it may also permit inversion with full verbs, as German does, leading
them to produce questions such as 'Like you baseball?' (Magst du basebalP.)

Stage5
lnversion in wh- questions with both an auxiliary and a main verb.
How do you say 'proche'?
What's the hoydoing?
French-speaking learners may have difficulty using Stage 5 questions in
which the subject is a noun rather than a pronoun. They may say (and
accept as grammatical) 'Why do you like chocolate?' but not 'Why do chil
dren like chocolate?' In this, they are drawing on French, where it is often
 Second language learning 51

ungrammatical to use inversion with a noun subject ( *Pourquoi aiment les

enfants lechocolat?).

Stage 6
Complex questions.
question tag: It's better, isn't it?
negative question: Why can't you go?
embedded question: Can you tell me what the date is today?
Pienemann's developmental sequence for questions has been the basis for a
number of studies, sorne of which will be discussed in Chapter 6. Alison
Mackey and her colleagues have done a number of these studies, and she pro
vided the data in Table 2.2. These examples come from three adult Japanese
learners of English as a second language who were interacting with a native
speaker in a 'spot the differences' task. In this task, learners have similar but
not identical pictures and they have to ask questions until they work out how
the picture they can see is different from the one their interlocutor has. Note
that progress to a higher stage does not always mean that learners produce
fewer errors.

ACTIVITY Analyse learners' questions

Using the information about the developmental sequence for questions, circle
the stage of second language question development that best corresponds to
ea.ch question.
(Hint: Read all of each learner's questions befare you begi n.)
Stage
i Learner 1
1 Where is he going and what is he saying? 1 2 3 4 5 6
1 2 Is the room his room? 1 2 3 4 5 6
3 Is he taking out his skate board? 1 2 3 4 5 6
" 4 What is he thinking?
1
1 2 3 4 5 6
11

5 The gi rl, what do you, what does she do, what is 1 2 3 4 5 6

she doing?
,,,

"r':'
.11 Learner 2
J: 6 Are they buying sorne things? 1 2 3 4 5 6
11111

1
7 Is they bought present? 1 2 3 4 5 6
111 !