Testing authentic language:

the problem of meaning

Herbert W. Seliger Queens College, City University of New York

In this paper, of the ways in which meaning is extracted in authentic

some
language contexts discussed. The claim is made that it is always the
are
case that some level of the language we use is imperfect or deviant from
some putative ideal. Finally, two common contexts in which authentic
but deviant language is the medium of communication are examined. In
attempting to understand how participants manage to communicate in
situations where the code is deviant and where they are unequal in their
control of the language, we can gain insights into what contributes to
successful communication under authentic conditions. Some implications
for authentic language testing are considered.

We aretold that Samuel Clemens, better known as Mark Twain, had

a reputation for using four letter words and expletives to pepper his
speech, especially when he was exasperated about something. One
day, his wife, disturbed by that kind of language around her home,
decided to teach him a lesson. As she was passing the bathroom where
Twain was shaving, she heard a string of four letter words resulting
from Twains cutting himself. She opened the door and repeated all
of the expletives she knew to show him how horrible it sounded.
Twain calmly turned to his wife and said, My dear, you know the
words but you lack the melody.
The point of this story is that there is more to language than strings
of words and that authentic language consists of more than either
words or the grammatical rules for arranging these words into more
meaningful relationships. It also consists of, among other things, the
intent of the speaker and the meaning system of the listener who
receives and processes these messages.
One of the illusions that we have outgrown about language is that
if the learner can manipulate the linguistic code, everything else will
fall into place. In recent years, we have seen a developing interest in
how language works in context, coming to realize that the formal
2

linguistic aspectsof language are only part of the picture. We realize

aswell, that when language is embedded in a social context it conveys
meanings other than those present in the sentence in isolation. Look at
the examples below taken from those now classic Pink Panther movies:

la) Police inspector to cab driver as approaches taxi: Follow that cab!
Cab driver drives off leaving inspector standing at the kerb.
lb) Police inspector to passer-by: Do you know the way to the hotel?
Passer-by: Yes (and continues on his way).
The above examples may seem like demonstrations of miscom-
munication of meaning but they demonstrate that the problem does
not lie in the misuse of lexicon or in something ungrammatical in the
exchange. Rather, the problem lies within what might be called the
deviant use of sentences outside of their strict grammatical inter-
pretation. The ability to use language in authentic environments is
predicated on our being able to deal with deviancies from some ex-
pected idealized norm. As will be demonstrated below, such deviant
use of language is common to everyday communication.

I Defining authentic language

A primary characteristic of authentic languageis that it is deviant in
some way from an ideal norm which none of us apparently speaks.
While it is possible to claim that it is normal for authentic language
to deviate from some ideal norm which the speaker/listener has in his
mind, it is necessary to distinguish between authentic language which
is commonly deviant, and authentic but deviant language which is
the result of some pathology. How is the pathological speech of a
schizophrenic or someone suffering from aphasia different from the
deviant speech produced by a baby, a foreigner or a creole speaker?
2) Well, this is
... mother is away here working out ohere to get her better,
but when shes working, the two boys looking in the other part. One their
small tile into her time here. Shes working another time because shes
getting too (Goodglass and Kaplan, 1972).
3) dei wen go ap dea erli in da mawning go plan ai gata go haia wan kapinta
go fiks da fom (Bickerton, 1982).
From examples 2 and 3 we can see that one important difference
between pathological deviance in speech and non-pathological devi-
ance is that the
language conforms to some predictable code or
sequencing rules which are also in the possession of the listener.
While we may not be familiar with creole (3), it is possible to decode
it by reading it aloud and once some minimal rules are understood, it
 3

is possible to understand it fully and to even converse in it. It is a

deviant code in that it consists of an amalgam of other simplified
codes enabling speakers of pidgins or creoles to communicate with
each other requiring a minimum of language learning effort.
This cannot be said for the speech of schizophrenics (2). Here the
breakdown appears to be the result of a combination of syntactic
and semantic problems. Schizophrenic speech is pathological because
it is an idiosyncratic dialect in the possession of only one person and
therefore not useful for communication. It is not clear what causes this
breakdown and even whether the speaker is aware of the breakdown.
Based on the above two examples, the following general principles
may be concluded about the nature of authentic but deviant language:
1) Authentic deviant language is a shared code at some level. In order
for language to function as a communicative tool it must be minimally
shared by other people. What is the minimum number of people
required? The answer would seem to be two as in the case of siblings
or twins who sometimes develop their own code. A more
important
question is what is the critical level of sharing which is necessary for
a code to function as a tool for discourse? At what
point do codes
break down, no longer functioning for communicating information?
As will be shown below, language never conforms to the ideal
described by linguists except perhaps in the laboratory. Authentic
language as has been noted above is naturally deviant. What remains
to be explored is how speakers deal with this deviance both as
speakers and as listeners and what the limits of our ability to deal
with deviancy are. The components of the code we are dealing with
are not only linguistic in the formal sense. Other sources of
knowledge
arc brought to bcar in dealing with authentic
speech and these will be
discussed below. A common example of this problem is the speech of
a non-native speaker which is heavily accented. It is clear that we are
able to understand deviant speech until we become so involved with
attempting to decode the phonological information that the mind
cannot process other information at other levels of the language.
2) Authentic deviant language relates to a shared experience with the
real world and conforms to certain universal expectations about what
communication is supposed to be like. For example, we expect com-
munication to have content that is relevant and truthful to the
listener and to contain information that is not self-evident. This would
go beyond such self-evident statements as We are standing here in
this room and talking or I see that you are wearing shoes or the
common types of sentences found in
language drills such as My shirt
is red. My hat is on my head.~
4

Returning to example 2 above, it is clear that the speech of a

schizophrenic, while authentic, does not conform to the two prin-
ciples discussed above. Pathological language cannot function as a
system of communication because it does not adhere to either lin-
guistic or pragmatic convention nor does it relate to the real world
in the expected way in terms of coherency of expression and it is not
a shared system with anyone outside the producer of the language.
The therapist must work with such output and try to make sense of
it. As will be discussed below, this process is not too different from
that employed by speakers in everyday situations except that our
expectations are that our interlocutors are trying to make sense and
that sense can be made from their utterances.
This does not mean that authentic language always succeeds in its
goal of total communication or that the linguistic and pragmatic
codes and conventions are sufficient to accomplish the task of com-
munication. It will be claimed that because authentic language is
always deviant in some way, it is never the case that the speaker fully
succeeds in communicating what he has in his mind to his listener or
that the listener fully comprehends what is in the mind of the speaker.
These realities of authentic language have important implications for
language testing. Since it is obvious that the very act of testing
removes language from its authentic environment, the description
of conditions and processes within which language users function in
dealing with authentic but deviant language may lead to the develop-
ment of tests focused on these processes rather than on futile attempts
to replicate nature in language.

II Authentic language and meaning

The primary use of language is to communicate meaning. Meaning
can, of course, be communicated at different levels. Unfortunately,
in the act of using language, we often do not communicate the mean-
ing we intend and various levels of breakdown can occur. Therefore,
it may be argued that miscommunication and dealing with the
common phenomenon of incomprehension is part of
functioning in
authentic language contexts. That is, since misunderstanding or not
understanding is common, all speakers must develop from childhood
the necessary mechanisms to compensate for inadequate language
systems which do not process all levels of meaning.
In this regard, Bates (1976) has stated the following:
Meaning is a set of mental operations carried out by the speaker, which the
speaker intends to create in the mind of the listener by using a given sen-
tence. Whether or not the speaker actually succeeds is a separate issue.
 5

It is this separate issue of the relative success of the speaker to

create the same intended meaning in the mind of the listener and the
success of the listener in reconstructing the speakers intended mean-

ing that it is important to consider in language testing. This problem

will be returned to later. Suffice it to say at this point, language test-
ing is often concerned with looking at the product of the learner
from the meaning perspective of the evaluator and not in measuring
the relative success of the learner in creating his intended meaning
nor in the success of the learner in inferring intended meaning from
another speaker. In language testing, especially testing which focuses
on testing communicative or functional
proficiency, the learners
intentions are thought to match the testers conjectures about those
intentions.
From this view of the role of meaning in language use we can con-
clude the following:

1) Meaning has no independent status outside the minds of the speaker

and the listener. That is, while words in a language have dictionary
meaning and may be described in terms of formal semantic features
by the linguist, the language which we use is also derived from a net-
work of associations which each of us has developed through his own
idiosyncratic experience with the language. Additionally, if we are
bilingual or polyglot, the use of words in one language or their intended
use may be affected by the semantic fields of these words or lexemes
in another language.
Austin (1962) has referred to these different levels of meaning as
locutionary, illocutionary, and perlocutionary levels of meaning. The
locutionary level of meaning would indicate some objective meaning
which could be assigned to an utterance; the illocutionary meaning
would indicate the force associated with a particular formulation in a
language such as making a promise or disagreeing by using a par-
ticular linguistic formulation (locutionary level). The deletion of
subject position you would indicate that a command is being given
and the meaning associated with the command itself. The perlocu-
tionary meaning refers to the effect on the audience or listener
such as obedience, resentment, agreement and so on. That is, a
response to the speakers utterance is not entirely overtly expressed
by the listener (see Austin, 1962; and Levinson, 1983 for further
discussion).
2) It followsthen that if meaning has no independent status, the
successof communication can be measured by the degree to which
the meaning intended in the mind of the speaker is generated in the
mind of the listener. However, the success of this generating process
6

may not be dependent on what is present in the code at the time of

the communication but rather on the ability of the listener to identify
what is not present and to reconstruct the missing pieces on the basis
of an assumed shared linguistic and pragmatic knowledge base.
3) It also follows that meanings of utterances may not reside in the
wellformedness of utterances themselves but in shared presupposi-
tions about language which underlie a given code and in the ability
of a speaker/listener to utilize these presuppositions. Example 1
above would be a demonstration of this point.
If the above points are accepted, it follows, then, that the task for
language testing is to find ways of measuring the degree to which
language learners are able to utilize implicit as well as explicit infor-
mation in a given language context and to what degree the illocu-
tionary and the perlocutionary responses of language learners can be
calibrated to some putative illocutionary and perlocutionary response
norms of native speakers.
In the real world there are many versions of authentic language
just as there are many versions of what may be recognized as English
from varieties of the language spoken all over the world. There are
age as well as class and regional varieties of a given language not to
mention those varieties that arise as a result of bilingualism and lan-
guage mixing. There is a continuum of language variation within any
specified variety of a language such that a speakers knowledge is
revealed not by his production of normatively grammatical sentences
but rather by his ability to adapt to the changing demands of the
perceived features of the environment. That is, a speakers ability to
interpret deviant language and to adjust his own grammar concomi-
tantly is more conclusive evidence of language knowledge than
irresolute control over some norm.
Even native speakers apparently do not acquire all of the varietal
forms of the dialect of a particular language which they speak. It
would therefore be unrealistic to expect second-language learners to
reach this functional level. However, what second-language learners
probably do is to employ a universal communication strategy. They
learn to utilize what they have in order to interpret a message formu-
lated in a variety outside their competence. That is, an authentic
language test which takes place continually is one which, based on
the assumptions of incomplete knowledge, requires the listener to
utilize what he knows in order to reconstruct the intended meaning
of the speaker. He does this through the formation of hypotheses on
the basis of some system of probable meanings which he assigns to
utterances.
The paradox is that the absolute intended meaning can only be
 7

known by the speaker himself. The best that the listener can do is
settle on a probable meaning from among several possibilities. This
also means that the more knowledge, linguistic and pragmatic, which
is shared by speaker and listener, the more likely will the listener be
able to reject unlikely possible meanings from consideration in inter-
preting a particular utterance. Given this view, it becomes increasingly
clear why miscommunication is so common in language teacher-
language learner interactions and why language testing so often does
not seem to reflect what learners can or cannot do in natural language
interaction situations.

III Inequality in authentic language interaction

Up to this point we have been concerned with the problem that
authentic language is always deviant from some ideal norm and that
listeners must develop ways of dealing with these deviancies in order
to arrive at approximations of the intended meanings of speakers.
This problem is further complicated by the fact that language inter-
action also takes place under what may be called conditions of
inequality. We have already alluded to this inequality in the discus-
sion above. It exists not only between native speakers but also among
other dyads where the pragmatic and linguistic knowledge is unequal
in some way.
When it is stated that two participants in a language interaction are
unequal, it means that at some level - linguistic, pragmatic, cognitive
-

one of the participants is able to use language in a more elaborate

form than the other. This could express itself in the range of registers,
expressive ability (Weinreich, 1968), semantic nuances attached to
utterances, the use of culturally or ethnically loaded expressions
(Pfaff, 1979), or in the ability to manipulate basic syntactic devices
such as paraphrases (Gleitman and Gleitman, 1970).
While it is not possible to describe all of the parameters of this
inequality, there is sufficient evidence in the literature on adult-
child, foreigner-native speaker, teacher-language learner to assume
that it is present in most of our everyday language exchanges.
The interesting empirical and theoretical question is: given the fact
that speakers of unequal language capability come into regular con-
tact, how is this inequality resolved so that communication of mean-
ing may occur? The theoretical implications of this question will
perhaps lead to a better understanding of the process of language
acquisition as well as the hierarchical importance of various compo-
nents of language in contributing to projecting and interpreting
8

meaning. The empirical importance of this question for language

testing is that it can lead to deeper, subtler and more valid means
of testing how language learners negotiate and obtain meaning which
is perhaps a better indication of language ability than the measure-
ment of some arbitrary set of language items.
Language inequality may be resolved both diachronically and syn-

chronically :
1 ) Diachronically, the question is whether over time, does the more
limited system evolve to resemble the more elaborate one? Does
the speaker of the more elaborate system develop a mediating
code specialized for contact with speakers of the limited system?
In the first instance, we have the cases of first and second language
acquisition. In the second instance, we find the development of
simplified registers (Ferguson, 1982) such as foreigner talk, pidgins
and eventually creoles. In some cases, the mediating system becomes
fixed (pidgins) and in other cases, it changes over time as the lan-
guage learner increases in proficiency (baby talk, foreigner talk,
2
teacher talk).2
2) However, of more concern to the problem of testing authentic
language is the synchronic question of how speakers function in
situations of inequality. What must happen in order for communica-
tion to take place where language does not conform to some Platonic
(or Chomskyean) ideal? That is, the ability to process language when
the speaker is in a disadvantageous position linguistically, is itself

deserving of study.

IV Types of inequality and their resolution

Two types of unequal language relations having special relevance for
language testing will now be examined in more detail in terms of the
characteristics of the inequality as well as how this inequality seems
to be resolved. The two types are: a) adult-child inequality; b)
native speaker-second-language learner inequality.33

a Adult-child inequality: While there is some disagreement among

researchers regarding the function of baby talk (the speech of adults
to babies) - whether it has a teaching function as well as a communi-
cation function - there is consensus that its primary function is to
facilitate communication. Brown ( 19 7 7 ) claims that baby talk evolves
from the desire to communicate and to clarify.
4 Baby: Baby highchair.
Mother: Yes, baby is in the highchair (Brown and Bellugi, 1964).
 9

Example 4 demonstrates one of the most common clarification

devices, the expansion by the mother of the reduced utterance of the
child. The mother seems to be both clarifying for herself as well as
providing a more elaborate form for the child. This response to a
deviant utterance might be regarded as a form of hypothesis testing
on the part of the listener.
The inequality in adult-child language interaction rests with the
obvious facts that the child lacks verbal production and comprehen-
sion ability as well as the conceptual cognitive development of the
adult. We must assume pragmatic inequality as well or else how could
we explain the very different illocutionary forms found in baby talk

compared with adult language such as one word commands, redun-

dancy, repetition of words, what is referred to as nursery tone and
high pitch levels in speech.4
What is interesting for our purposes is that mothers do not always
use baby talk register when conversing with the child. They often

lapse into more adult forms. Snow (1977) found that mothers sen-
tences tended to become more complex when embedded in a dis-
course than when used in isolation.

Diachronically speaking, we know that the language ot babies

evolves into that of the adults in their environment. What is interesting
for Ferguson (1982) is that we apparently maintain this register for
baby talk on a back-burner for use later in life either for talking to
other babies or even to pets.
Synchronically, the question which naturally arises is: what does
the baby do in order to process or comprehend language beyond its
current level of competence? On the other side of the coin, we might
ask: what does the parent do to deal with deviant baby talk? How
does each side get meaning out of the others message? If we assume
that language acquisition progresses through stages whereby com-
prehension informs the child about meaning distinctions associated
with different forms, then answers to these questions may provide
important directions for understanding how language acquisition
actually happens.
Based on what we know about the linguistic and pragmatic charac-
teristics of the language to which the child is exposed as well as the
physical and affective environment in which language interaction
takes place, it is not unreasonable to hypothesize that the child
utilizes all of the special extralinguistic information available to
develop meaning from the linguistic code. Tone of voice, pitch and
stress patterns are obvious cues but so are gestures, smiling, etc. What
is likely is that the child shifts gradually from a dependency on these
extralinguistic cues for meaning as more ability to process information
at the linguistic level
develops.
 10

What the child may be doing, therefore, is developing a set of

probability relationships between linguistic forms and extralinguistic
cues. In short, the child must infer the meaning of utterances by first

developing hypothetical rules relating meaning to pragmatic and

extralinguistic levels of utterances and later relating these to linguistic
meanings. That is, the child progresses from illocutionary/perlocu-
tionary levels of meaning to locutionary levels of meaning. The parent
is in a similar position in interpreting the talk of babies simply be-
cause the linguistic or locutionary form of the childs speech can
have many different meanings. Like the child, the parent must hypo-
thesize about possible interpretations based on the context in which
.
the utterance has taken place.
b Native speaker-second-language learner inequality: The second ex-
ample of deviant and unequal language communication is that between
adult native speakers and adult second-language learners. It may
be claimed that there is a shared knowledge base between the two
participants as well as the non-linguistic cognitive tools necessary to
approach language processing at conceptually mature levels. This is
not the case with adult-child language interaction where the same
word may represent a different concept for the child than that
assumed by the adult (Vygotsky, 1962).
While it may be assumed that both speakers possess universals
about the nature of language systems and language communication
by virtue of the fact that both control a complete language system,
it may also be assumed that both possess a set of pragmatic univer-
sals for the same reason but one lacks the pragmatic conventions
specific to the target language.
At the pragmatic level it may be assumed that both have acquired
the equivalent of what may be called conversational universals or
maxims (Grice, 1975) described in Note 1 and that these too can
act as a basis for developing meaning from a conversation. In native
to non-native speaker conversations, however, these maxims will not
necessarily be fully realized due to the unequal levels of language
knowledge on the part of the participants.
For example, Grices Maxim of Quality states that ones contri-
bution to a conversation should be just as informative as required.
This means that while foreigner talk is often conceived of as a
reduced or simplified register, it must also contain redundancies
sufficient to make the message clear to the interlocutor as in the
following exchange between a native speaker (A) and a non-native
speaker (B):
 11

5) A: Did you see that building in Boston? The one thats all glass like that?
I live there.
B: You live there.
A: Yeah, near there. Near, around the back (Vander Brook et al., 1980).

Sometimes, however, the native speaker seems to break with the

sufficiency maxim when using idiomatic expressions that may re-
quire cultural background order to understand them. For example,
in
while the expression theyll take an arm and a leg refers to some-
thing which the speaker feels he is paying too much for and is
syntactically and lexically simple, a second-language learner will have
trouble with it unless he is already familiar with such expressions.
Native speakers often make conversation richer than necessary by
loading their speech with unintended idioms that they may perceive
to be part of a simplified register. The degree to which native speakers
can perceive such exceptions to the Maxim of Quantity in native-
non-native interactions remains to be investigated.
In addition, while there is still much simplification at the foreigner
talk level, second-language learners are still exposed to much discourse
that cannot be controlled for listener. For example, learners are
exposed to the second language through movies and television where
no adjustments are made (except in foreign countries where subtitles

may be supplied). Exposure to the unadulterated form of the target

language requires the learner to fall back on pragmatic, cultural and
extralinguistic information to form hypotheses about possible mean-
ings in the discourse or text. For this reason, it is difficult for second-
language learners to function on the telephone where voice frequencies
are reduced and the listener has no recourse to gesture, facial expres-
sions and other clues to meaning.

V Conclusions
In the cases that have just been examined, it is clear that conditions
of inequality require the utilization of extralinguistic information in
developing inferences or hypotheses to resolve conditions of incom-
prehension at linguistic and pragmatic levels. Depending on the parti-
cular nature of the language interaction, interpretation will be based
to a greater or lesser extent on linguistic or pragmatic systems.
Both linguistic and pragmatic systems would seem to supplement
each other in helping language acquirers to get meaning depending
on the degree of inequality and deviance in the
language interaction.
It is reasonable to suppose that at initial levels of first or second-
language learning, there is a greater reliance on pragmatic sources
12

of knowledge upon which to develop hypotheses about the meaning

of the message. Second-language learners at beginner level report
listening for particular words in particular positions in sentences and
ignoring everything else. It is suggested by Brown and Bellugi (1964)
that children at beginning stages of sentence production listen for
stressed content words and ignore stress reduced words which are
usually grammatical morphemes.
As more linguistic knowledge is acquired, greater attention is paid
to the linguistic details of the utterance since more specific and im-
portant information is only available through this system. However,
since we know that native speakers and advanced second-language
learners are able to function under conditions of reduced redundancy,
this would seem to indicate not only greater linguistic competence
but also a return to reliance on a different level of pragmatic infor-
mation which by this stage is more language-specific. In this sense,
the role of pragmatic knowledge is cyclical, serving initially as the
source of general inferences which aid in the development of the

linguistic system and appearing again at higher levels as a primary

source for inferences about the meaning of a communication under
conditions of reduced redundancy, deviancy and inequality.

VI The implications for authentic language testing

It has been shown that the only manner in which to explain how
language learners both function and acquire under conditions of
deviancy and inequality is to postulate the ability to infer or develop
and test hypotheses about the possible meaning of utterances beyond
the specific knowledge of the learner. While the study of these pro-
cesses is interesting for its own sake, several important conclusions
are suggested for language testing research:

1) Since the goal of language use in authentic contexts is to com-

municate meaning, and ...

2) Since successful communication may be defined by the extent to

which meaning intended by the speaker is generated in the mind of
the listener, and ....
3) Since inferencing and hypothesis testing are necessary processes by
which to extract meaning at all levels of language proficiency ...
it follows that a possible avenue for language testing would be to
develop tests that focus not on the product but on the successful
implementation of the processes. For example, what is the quality
and degree of dependency of the learners hypotheses on pragmatic
 13

versus linguistic information? If asked to order hierarchically a list of

inferences on a given text, to what extent would they match those
of a native speaker or the author? Is it possible to devise a method
which could measure the degree to which native speakers could infer
correct meaning from non-native speech assuming that success in
communicating intended meaning is also a measure of proficiency
for the learner?
What is clear from these suggestions is that it is not necessary to
throw ones hands in the air and say that by the very act of testing
we are destroying authenticity. The ability to test authentic language

capabilities may rest on our ability to identify, extrapolate and simu-

late these abilities rather than try to emulate nature. It is hoped that
this article has provided some suggestions in this direction.

VII Notes
1 Grice (1975) has proposed a principles which he
set of claims allow
for efficient conversation. In brief, they are:
a) The Maxim of Quality. Contributions to a conversation are ex-
pected to be true.
b) The Maxim of Quantity. Contributions to a conversation are not
expected to be more or less informative than required. In the discus-
sion here, I have interpreted this to also refer to linguistic devices
such as paraphrasing, stress and pitch which are probably redundancy
markers in some registers but not in others.
c) The Maxim of Relevance. Contributions are expected to be relevant
to the conversation. Note how often this maxim is ignored in lan-
guage classes.
d) The Maxim of Manner. Contributions should not be obscure or
ambiguous but should be brief and orderly.

It is not clear that all of these maxims can be applied unambiguously

to either native or non-native authentic language. However, they do
provide an interesting mechanism with which to explore some of the
similarities and differences that can occur or must occur in speech
between native and non-native speakers of a language.
2 It is important to note that while the focus in language testing is
on the learner, it is just as interesting to look at the responses of the

speaker of the target code (adult or native speaker). The talk to

babies, to foreigners and the talk of teachers to language learners at
different levels of proficiency reflect the perceptions of the speakers
of the target language and of the capabilities of their interlocutors -
the child, the foreigner or the language learner.
3 A third kind of language inequality mentioned above will not be
14

discussed here because of the scope of this paper. This is the inequa-
lity which exists between adult native speakers of the same dialect.
The reader is referred to an excellent treatment of this problem from
the psycholinguistic perspective by Gleitman and Gleitman, 1970.
They concluded on the basis of their research that some native
speakers apparently acquire a more elaborate form of the grammar
thus enabling them to perform more creative and sophisticated
operations with abilities such as paraphrasing.
4 As an example of the different illocutionary repertoire of the child,
I recently had the following telephone conversation with a three-year
old who insisted on answering the telephone:
Telephone is picked up. There is silence.
Adult: Hello?
Child: Hello.
Adult: Is your mother there?
Child: Yes.
Adult: Can you bring your mother to the phone?
Child: Yes. (Long pause, breathing through a stuffed up nose is heard.)
Adult: Bring your mother to the phone.
Child: OK.
5 Evidence for this is that as general language proficiency increases in
both L1 and L2 and as sentences become longer and more complex,
there is increasing acquisition of the various categories of the stress-
reduced grammatical morphemes (see Brown, 1973; Bailey et al.,
1974).

VII References
Austin, J.L 1962: How to do things with words. Oxford: Clarendon Press.
Bailey, N., Madden, C. and Krashen, S. 1974: Is there a natural sequence in
adult second language learning? Language Learning 24, 2.
Bates, E. 1976: Language and context : the acquisition of pragmatics. New York:
Academic Press.
Bickerton, D. 1982: Learning without experience the creole way. In Obler,
L.K. and Menn, L., editors, Exceptional language and linguistics, New
York: Academic Press.
Brown, R. 1973: A first language. Cambridge, Massachusetts: Harvard University
Press.
1977: Introduction. In Snow, C. and Ferguson, C., editors, Talking to child-
ren, London: Cambridge University Press.
Brown, R. and Bellugi, U. 1964: Three processes in the childs acquisition of
syntax. Harvard Educational Review 34.
 15

Ferguson, C. 1982: Simplified registers and linguistic theory. In Obler, L.K. and
Menn, L., editors, Exceptional language and linguistics, New York: Academic
Press.
Gleitman, L.R. and Gleitman, H. 1970: Phrase and paraphrase. New York:
W.W. Norton.
Goodglass, H. and Kaplan, E. 1972: The assessment of aphasia and related dis-
orders. Philadelphia: Lea and Febiger.
Grice, H.P. 1975:Logic and conversation. In Cole, P. and Morgan, J.L., editors,
Syntax and semantics: speech acts, New York: Academic Press.
Levinson, S.C. 1983: Pragmatics. Cambridge: Cambridge University Press.
Pfaff, C.W. 1979: Constraints on language mixing. Language 55, 2.
Snow, C. 1977: Mothers speech research: from input to interaction. In Snow, C.
and Ferguson, C., editors, Talking to children, Cambridge: Cambridge
University Press.
Vander Brook, S., Schlue, K. and Campbell, C. 1980: Discourse and second
language acquisition of yes/no questions. In Larsen-Freeman, D., editor,
Discourse analysis in second language acquisition, Rowley, Massachusetts:
Newbury House.
Vygotsky, L.S. 1962: Thought and language. Cambridge, Massachusetts: MIT
Press.
Weinreich, U. 1968: Languages in contact. The Hague: Mouton.