7.

Theories in second language acquisition

INTRODUCTION

 There are many different responses that word ‘theoretical’ evokes in language teaching.

 One of the important ones is that ‘work in SLA has not advanced far enough yet for
theorizing to be productive.’

 In short, it means that we know so little about SLA that every theory we come up with will be
wrong.

 Authors of this chapter argue that more theorizing would speed up a process of SLA teaching
and would protect language teachers from inadequate theorizing in the classroom

Theory construction and social science

 Although there has been an explosion of data-based research activity in the field of SLA,
many people are unsatisfied and they want to know where that is getting us.

 The main problem is that data-based work is not theoretically motivated and that some
studies are ‘still only descriptive with the issue addressed appearing to have occurred to the
investigators after data were collected’.

 What is theory?

 ‘Theory refers to what has been discovered through empirical observation’

 Sometimes theory aims at an understanding of phenomena.

 There are two types of theories described in this chapter:

 1) The first one is the set-of-law forms where statements have a status of generalizations or
of laws

 The difference between these two is that unlike laws, generalizations allow for exceptions.

 For example, if we say that ‘all cows have horns’ and consider it a law, one cow with horns
would disprove it or it would not be considered a cow.

 2) The second one is the causal process form which tires to explain the phenomena.

 It tries to explain how and why process is going to occur.

 Statements that try to prove answer those two questions are inter-related, which means that
every statement that cannot be proven is connected to the statement that can be proven.

 ‘A theory remains falsifiable as long as parts of it are testable and all untestable parts are
related to testable ones.’
 Nativist theories of SLA

 Nativist theories are those which intend to explain acquisition by positing an innate biological
endowment that makes learning possible.

 While Chomsky posits knowledge of substantive universals (syntactic categories, distinctive

phonological features, formal universals) to be innate, in other nativist theories what is held
to be innate consists of general cognitive notions like dependency, adjacency, continuity etc.

 Chomsky and those working in his framework support the idea that humans are innately
endowed with universal language -specific knowledge.

 The main argument, called the ‘logical problem’ of language acquisition is that without such
endowment language learning would be impossible because input data are to ‘poor’ to allow
acquisition ever to occur.

 The input is poor in two ways:

 1) It is claimed to be ‘degenerate’ – marred by performance features and ungrammaticality,

therefore an inadequate data base for language learning.

 2) The input is ‘degenerate’ in the sense that is inadequate in various ways.

 It does not usually contain ‘negative evidence’

 Teachers rarely correct ungrammatical speech of their students, they react to the truth
value.

 Research has shown that the corrective feedback is apparently ambiguous, ill- timed and
ineffective.

 White gives some examples of the problem created by hearing only ‘positive’ evidence, e. g.
How do learners find out which sentences are ungrammatical?

 The fact that children do not make errors, confirms Chomskyan claim, that they already know
that rules are structure dependent.

 Similar to that Lightbown and White ask how, without negative evidence, learner finds out
what is ungrammatical in WH- sentences.

 Chomskyan answer to that problem is to posit innate knowledge of constraints on WH-

movement, and WH- word may only be moved out of one bounding category ( S and NP in
English language).

 Universal Grammar consists of a set of such innate, abstract, linguistic principles, which
govern what is possible in human languages, helping to facilitate the learning problem
created by ‘poverty of stimulus’.

 There are some general ways in which languages differ from one another, the principles are
held to be able to vary in certain restricted ways, along so called ‘parameters’.
 Learnability by positive input alone is illustrated by the differential distribution of pied –
piping and preposition – stranding in languages, some allowing only one and some both, i.e.
English both – Spanish only one

 Learnability by positive input alone concerns the supposed ‘pro – drop’, or ‘null – subject’
parameter.

 Some +PD languages (Italian and Spanish) allow empty subjects, while other –PD languages
(English, French) do not.

 Chomsky assumes that the PD parameter is initially set neutrally, and claims that the noticing
relevant features will trigger the correct setting for the language concerned for all structures
governed by it.

 Besides Chomsky and White, there were a lot of studies made by scientists like Hyams and
Hills claiming that learning language is innate endowment or not.

A critique of language- specific nativist theories

 while valuable in their own right, studies have a potentially serious interpretation problem ;

 problems concerning empirical falsifiability;

 Chomsky’s three questionable assumptions :

 1st – learning occurs quickly and is mostly complete by age five;

 2nd – certain syntactic principles are unlearnable, and therefore innate;

 3rd – input available to learners is inadequate; it lacks essential negative evidence with which
to remedy excessive complexity and overgeneralization.

 UG implicitly recognizes that lack of negative evidence does not necessarily prevent
acquisition

 Since lexical rules are language and culture specific, and so cannot be regulated innately,
lexical overgeneralization is a problem which has to be dealt with through linguistic
experience in any theory, including a nativist theory. (Bowerman)

 The cure for lexical overgeneralization, which all learners of first or second language make,
has to be a theory of learning which constrains the initial hypotheses in some way, and then
modifies them via positive evidence. ( Pinker )

 UG’s recognition that overgeneralization of lexical rules is remediable by experience of

positive evidence alone weakens its claim that the basic syntactic rules cannot be learned
this way, but only by use of a heavy language- specific innate endowment.

 The existence of well- attested developmental sequences in first and second languages lends
further support to the idea that the lack of negative evidence is less significant for learning
than Chomskyans claim.
 While attractive because they offer precise, testable, theoretically motivated predictions ,
UG- inspired analyses suffer from the same static, target- language orientation that had
plagued other approaches to Il development based upon theories of language rather then
theories of language learning.

KRASHEN'S MONITOR THEORY

 One of the best known and most influential theories of SLA in the 1970s and 1980s was
Krashen’s Monitor Theory (MT). MT began as a model of SL performance.

 It was an attempt to capture and reconcile two phenomena: first, there existed significant
association between the orders of appearance of certain English grammatical morphemes in
the speech and writing of SLA learners of different ages, L1 backgrounds and conditions of
exposure; second, disturbances in the natural order on certain performance tasks,
specifically the reading and writing tasks.

 Krashen claimed that 2 separate knowledge systems underlay SL performance: the acquired
system and the learned system.

 The acquired system was typically the only knowledge source speakers could use in the real-
time communication, when they were attending to meaning, not form.

 The learned system served only as a planner and editor with which to inspect, or monitor,
the output of the acquired system.

 The learned system was only accessible when 3 conditions were met: time, the learner’s
focus on form and learner’s knowledge of the rules.

 Krashen stated that monitoring (small ‘m’), could be done by “feel” judgments, using the
acquired system.

 Learners might pause to see whether a word or construction, looked or sounded right
(monitoring with the learned system on the other hand, would be equivalent to accessing a
rule of English spelling the NS happens to know).

 The two systems, like the processes which produce them, acquisition and learning, operated
separately, there was no cross-over or interface.

5 major hypothesis of the Krashen’s monitor theory

 Acquisition-Learning Hypothesis (2 independent ways of learning SL: acquisition and

learning. Acquisition refers to the subconscious process used by children developing their
first language; learning is conscious process, which results in a separate system of simple
grammar rules, or knowledge about SL.)

 The Natural Order Hypothesis says that SL rules are acquired in a predictable order , one
apparently not determined solely by linguistic complexity, and certainly not by the order in
which items appear in the teaching syllabuses.
 The Monitor Hypothesis encapsulates the relationship posited between the acquired and
learned systems during SL performance.

 The Input Hypothesis attempts to explain how a learner acquires an SL (It maintains that an
SL is acquired through processing comprehensible input CI – the language that is read or
heard and understood.

 Progress along the ‘natural order’ is achieved when a learner at some stage of IL
development receives CI that contains structures one step beyond the current stage.).

 Speaking is a result, not a cause of acquisition; it emerges as a product of growth in

competence, achieved through understanding target-language samples.

 The Affective Filter Hypothesis: motivation, self-confidence and anxiety play facilitative but
non-casual role in SLA.

 Krashen states explicitly that the five hypotheses can be summarized as follows: “People
acquire second language only if they obtain comprehensible input and if their affective filters
are low enough to allow the input ‘in’…”

 Krashen was instrumental in pioneering so-called ‘sheltered subject matter’ courses (the
instructors in courses for foreign students can adjust the English they use for the advanced
but still non-native language ability of the students).

 Characteristics:

 1. Focus on meaning, not form,

 2. Proscription of structural grading and error correction,

 3. Creation of positive affective classroom climate.

A CRITIQUE OF MONITOR THEORY

 The monitor theory had the side-effect of creating instant experts who had not actually read
the related research, much less conducted any themselves, but who now asserted their own
views.

 McLaughlin expressed serious concerns about learning/acquisition distinction. Krashen

claimed to distinguish between conscious and subconscious processes via introspection, i.e.
by whether subjects reported making grammatically judgments based on ‘rule’ or ‘feel’,
respectively. McLaughlin questioned the reliability of this methodology.

 McLaughlin made two other criticisms of Krashen’s explanation of variation in morpheme

accuracy orders. First, he claimed it was ‘circular’ to say that a ‘natural order’ was obtained
under Monitor-free conditions, the product of ‘acquisition’ but that when the Monitor
operated, the order was disrupted, the product of learning. Second, he objected to Krashen
modifying his initial claim as to when the Monitor could be assessed, with resulting
disturbances in the ‘natural order’ on speeded composition after viewing the results of
studies.
 The Input Hypothesis and Affective Filter Hypothesis were also untestable.

 Some additional problems for MT are:

 1. There is no explanation for why the ‘filter’ does not exist in children, and only comes into
play at puberty, thereby, according to Krashen accounting for the low success of adults.

 2. The limited content and domain of the learned system, cannot account for the data on the
effects of instruction on IL development

 3. MT offers no explanation for the morpheme orders on which many of its claims are based
and tested, nor for any other developmental sequences. Etc.

 In conclusion Krashen’s ideas initially stimulated a good deal of data based research, and
forced some fresh thinking in language teaching circles (although some of the original claims
no longer excite much interest among researchers) .

Environmentalist Theories of SLA

 -an organism's experience is more important than its nature

 -Behaviorists' and neo-behaviorists' theories

 (Skinner, Chomsky...)

 -Connectionist models

 -PDP (Parallel Distributed processing)- learning is based on the processing of input

Schumann's Pidginization Hypothesis and Acculturation Model

 on the example of Alberto, Schuman tended to explain that his acquisition of ESL was limited
because of his social and psychological distance from speakers of the target language

 pidginization very similar to early naturalistic SLA

 -pre-verbal negation system, failure to invert questions, omission of the most of the
auxiliaries, subject pronouns, verbal and possessive inflections are some of the common
features

 the same simplification processes are at work in producing each

A critique of the Pidginization Hypothesis and Acculturation model

 The analogy between pidginization and early naturalistic SLA has been very productive, and it
generated a good deal of empirical research and theoretical debate.

 The following are some of the main objections that have been raised to analogy:

 Pidgins develop when groups of speakers are in contact, whereas SLA is a situation of
bilingual contact (Flick and Gilbert 1976.)
 SLA generally involves monolinguals, whereas speakers of pidgins generally know several
other languages. (Bickerton 1975., Flick and Gilbert, 1976.)

 Pidginization is a group phenomenon (closed community with great stability), and SLA is an
individual phenomenon.

 Pidginization is not targeted language change, but more independent linguistic development.
Pidgins often show incorporation of features from diverse source languages by way of
compensation.

 Schumann’s Acculturation Model has served to turn what have otherwise often been rather
vague notions about the role of social and psychological factors in SLA into coherent
predictions. The Model suffers from problems:

 It is empirically unstable (H.D.Brown)- no reliable and valid measures of psychological and

social distance exist. The Model may be testable in theory but not in fact.

 Second, if concerns about instrumentation are temporally aside, then empirical findings have
been mixed. The study of a group of scientists (Stauble and Kelley) showed that there was no
linear relationship between acculturation and the ESL development. The most acculturated
subject was actually the last proficient, and one of the least acculturated the most proficient
in English. The examiners were Spanish and Japanese speakers.

 A study which does report finding evidence consistent with Schumann’s predictions is that of
Maple. His research is interesting because his sample was of respectable size, he attempted
to validate his social distance measure(a questionnaire) by interviews with a sub—sample of
30 subjects, and he used more sophisticated statistical procedures to assess relationships
between the various independent variables and SL proficiency.

 Above mentioned researchers draw attention to the lack of any principled means of
weighting the various sub-components of acculturation, i.e. the numerous social and
psychological variables that go to make up the construct, recognizing the possibility that the
influence of each factor may vary both from individual to individual and over time- and, one
might add, at different proficiency levels and for different types of linguistic abilities. It is
important to mention that Acculturation Model is intended to apply only to immigrant
groups acquiring a SL naturalistically in the target-language environment.

 The other study which concerns examining SLA by members of groups whose social distance
does not appear to be either positive or negative, which is when Schumann claims
psychological factors will determine SL attainment a number of researchers have found little
or no relationship between those psychological factors and achievement.

 The third problem with Acculturation model concerns two conceptual aspects of the Model
which make the falsification impossible:

 Since there is no a priori specification of the conditions necessary for

psychological factors to override the social one, every possible combination
 of (positive, neutral and negative) social factors, psychological factors and
level of SL attainment is “predicted” by the model.

 Schumann provides no indication of the combination and/ or levels of social

and psychological factors he clams predict learning. If a direct connection
between social and psychological factors and SLA is what is being claimed,
the Model says nothing about how the effect is achieved, or even about why.

 In the last Schumann’s paper, written in 1986, he writes that: “Acculturation

as a remote brings the learner into contact with TL-speakers. Verbal
interaction with those speakers as a proximate cause brings about the
negotiation of appropriate input which then operates as the immediate
cause of language acquisition. Acculturation then initiates the chain of
causality.

 In conclusion, Schumann’s Pidginization Hypothesis and Acculturation model has been useful
in that they focused researches’ attention on linguistic simplification processes in early IL
development and on a possible causal role for a large body of social and psychological factors
in SLA.

 Each has been criticized on a variety of formal, conceptual and empirical grounds, but each
has made a contribution.

 Schumann has identified several of the contextual factors relevant in a predicting group-level
success in acculturating to the new society; the social-psychological variables he discusses
may constitute the raw material for a viable model of acculturation.

 The last thesis says that, on the other hand, it should come as no surprise if a mental process,
(second) language learning, is not successfully explicable by any theory which ignores
linguistic and cognitive variables.

Interactionist theories of SLA

 These theories are considered to be more powerful than either nativist or environmentalist
theories.

 When talking about theories, “power” is a negative characteristic meaning that more
variables, factors, causes and processes are needed to conduct the data of interest.

 Interactionist theories invoke both innate and environmentalist factors to explain language
learning.

 Interactionist theories share general characteristics, but they also differ from one another.

Givon’s Functional-Typological Theory and SLA

 Thomas Givon developed this theory from his work in functional-typological syntax and
diachronic language change.
 His goal was to crate unified theory of all kinds of language change, including (second)
language acquisition.

 His approach FTSA (functional-typological syntactic analysis) “shows” that syntax derives
from properties of human discourse and that we need to consider not one language, but
diverse body of languages.

 Givon claims that his theory can be applied to development of creoles and pidgins, SLA, child
LA etc.

 He also claims that speakers and linguistic systems move from pragmatic mode to syntactic
mode of communication.

A critique of Givon’s theory in SLA research

 FTSA is not considered to be capable of capturing accurately the differences among types of
language-change situations, which is shown in the mixed results of the research conducted
on this topic.

 Givon’s theory and his claims regarding pragmatic and syntactic modes of communication
were tested and each time defined as incomplete.

The ZISA group’s Multidimensional Model

 Zweitsprachenwerb Italianicher und Spanicher Arbeiter project was the beginning of one of
the most important bodies of SLA research.

 ZISA’s project started as a research on native speakers of Italian and Spanish and their
acquisition of German as SL and developed into the theory of SLA.

 This new theory was then used as a motivation for new studies of SLA (GSL, ESL and JSL).

 Original research was focused on GSL word-word order rules in five stages.

 Learners accumulated the rules, adding new ones to the ones they’ve already learned.

 This scale shows the five stages in learning German as SL:

 SVO < ADV < SEP < INV < V-END

 ZISA’s work is interesting because they included explanation of the GSL word-order data
which had potential to be used for other languages.

 There are three speech-processing strategies used by learners when dealing with those five
stages mentioned before:

 a) Canonical Order Strategy

 b) Initialization-Finalization Strategy

 c) Subordinate Clause Strategy

 Using five stages of German as second language and these 3 strategies ZISA created a new
model for SLA:

 1. The first, X stage represents sequence of words and phrases created by learners using only
the meaning or information focus, without any knowledge of the grammatical status of the
elements.

 2. The next stage, X+1 is the one where learners move elements from final to initial position
(and vice versa), leaving the canonical order intact. The learner still does not need any
knowledge of the grammatical status of the elements.

 3. The third stage, X+2, involves disruption of the elements, separating auxiliary verb from
the main one (used for example in questions) etc. This is the stage where knowledge of the
grammatical status of the elements is important (i.e. to know the word category of the
element which should be moved).

 4. The fourth stage, also known as X+3 also involves knowledge of the grammatical status of
the elements; the learners have to know the grammatical category of the elements in the
string.

 5. The final stage, X+4, is the most complicated and involves the greatest amount of
knowledge. Learners have to be able to recognize sub-strings within a string and to move
elements to other positions.

 These rules, stages and strategies are supposed to be working with every language system.
But, the model showed flaws when it was applied to English language because of the
differences between German and English word-order rules, and grammar features in general.

A critique of the Multidemensional Model

 Multidimensional model has made several valuable theoretical and methodological

contributions to SLA.

 One of the greatest strengths qua theory is its predictive power, a quality which also
promises several potentially exciting practical applications.

 One is the use of the developmental stages as a principled means of sequencing items in a
syllabus, given that the stages are by definition statements about what a learner can be
taught at a given stage of development.

 This is another way of saying that the data are made interesting by their combination with a
causal-process theory, an explanation with predictive power, and so with generalizability.

 Pienemann has illustrated this with respect to both structural and notional-functional
syllabus design. Another application is to teaching methology, where knowledge of
developmental stages and variational features can serve as a useful diagnostic resource for
teachers , allowing them to identify different kinds of errors and to assess their remediability.

 A third application is to placement and achievement testing. Language testing in general has
too long focused on accuracy – What has clearly been needed for a long time is IL-sensitive
 tests-instruments or procedures which capitalize on research findings about developmental
sequences in SLA, e.g. utilizing the stages defined in the predictive framework.

 Since the early 1980s several research projects have begun exploring relationships among
sets of IL features and between traditional proficiency measures and developmental features
.For an early application of the developmental stages in the predictive framework to a
teacher-rating procedure for assessing spoken language.

 The multidimensional model is not without problems. There are some current problems with
the falsifiability of the predictive framework, and a major question looming concerning just
what the Model will explain about acquisition, as opposed to constraints on acquisition.
These are rather mostly limitations rather than flaws.

 The most serious is the possibility that the Model may turn out to be quite revealing about
constraints on acquisition without saying much about how learners actually learn whatever
they do, constrained as they are.

 Each of three problems could be rectified: the first by some laborious empirical work and/or
careful searches for chunked items , the second by defining variational features a priori or, at
least, specifying the status of particular items before the study is undertaken, and the third
by setting limits on 'gaps'.

Conclusion
Comparing and evaluating theories

 Knowing when to modify and when to abandon a leaky theory is a pervasive problem in the
sciences. It seems to be particularly acute in SLA research for several reasons. As noted
earlier, there are at least forty theories, models, perspectives, metaphors, theoretical claims
in the SLA literature. There is often overlap among them, but equally often , areas and
uniqueness.

 What makes them difficult to evaluate is the fact that they sometimes differ greatly in scope,
or the range of SLA phenomena they treat; the type of data to which they are held
accountable; and the degree of abstraction oft he statements they contain.

 Differences in scope are easily documented. Krashen's Monitor theory, for example, has
always been supposed to account for all SL acquisitional types, as have Wode's linguo-
cognitive strategies, whereas Schumann's Acculturation Model is explicitly limited to
naturalistic SLA, and the Multidimensional Model just to morphology and syntax.

 Differences in the types of data to which theories are held accountable often result in one
theory being tested more precisely and so more stringently than another. Supporting or
disconfirmatory data usually take the form of global test scores, reaction-time measures, not
linguistic features in ILs.

 Lastly, with regard to differences in degree of abstraction, compare two apparently similar
claims concerning the potential for acquisition of new forms in a learner's input.
 These differences present serious problems when one attempts to evaluate the relative
merits of such position.

 Theory of a wider scope is better than one of narrower scope because it can be used to
address a wider range of problems. If two theories are equally good explanations of existing
data, the better theory is the simpler one, the one that makes fewer distinctions and uses
fewer distinctions and uses less power to handle the data.

A note of caution

 We alluded earlier to the fact that not all SLA researchers share our view that SLA research
will best be served by advancing a theory of theories to account for what is know about SLA.
The reservations those researches have are essentially two. The first; a theory at this point in
the nascency of SLA research is likely to be an oversimplification, if not a downright
erroneous description of the actual SLA process, so the theory may mislead, rather than be
helpful.

 The second reservation stems from the observation of the effect of a monolithic theory in
related fields, e.g. linguistics and psychology. It can sometimes be the case that a single
theory can be so politically powerful that it has the effect of stifling alternative points of
view.

 We must guard against overzealousness on the part of theorists or their devotees who feel
that they have a monopoly on the truth.

 What is our own view of the role of theory in SLA? We see SLA research as anything but an
'ivory tower' activity. Its findings, e.g. often affect the life chances of 'at risk' populations,
such as adult refugees needing an SL in order to hold down a job in a new country, minority-
language children seeking an education through an SL, and so on. These people will
appreciate the clear direction a theory can give them when designing and executing studies,
the interim solution it provides practitioners until SLA is better understood.