International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org ON THE STATISTICAL LEARNING THEORY, THE
RELATIONAL FRAME THEORY, AND THE EMERGENTIST
THEORY DEALING WITH L1 ACQUISITION PROCESSES
WITHIN THE CONTEXT OF INNATISM
Nima Pouladian
Department of Foreign Languages, Islamic Azad University, Shiraz Branch, Shiraz, Iran
Email:[email protected]
Firooz Sadighi
Department of Foreign Languages, Islamic Azad University, Shiraz Branch, Shiraz, Iran
E-mail: [email protected]
ABSTRACT
There have been numerous theories dealing with the process of first language (L1) acquisition,
three of which are discussed here in detail. These are the Statistical Learning Theory, the
Relational Frame Theory and the Emergentist Theory. This report looks at the mentioned
approaches from an innatist perspective in an attempt to find the possible existing common
grounds amidst the three. In order to realize this goal, the researchers first provided a short
account of each theorem, followed by a comparison of them in regard to the nativist camp’s
claims on first language acquisition process. Lastly, they try to determine whether the
aforementioned theories are inherently similar to innatism and identify fundamental differences
in the way they account for the L1 acquisition process. In the end, the positions of the three
approaches in regard to major learning theories such as nativism were identified.
KEYWORDS: Statistical Learning Theory, Relational Frame Theory, Emergentism, Innatism.
INTRODUCTION
According to Sadighi (2012), nativist theories hold that children are born with an innate
propensity with language acquisition and this ability makes the task of learning a first language
easier than it would otherwise be. These ‘hidden assumptions’ allow children to quickly figure
out what is and is not possible in the grammar of their native language and allow them to master
that grammar by the age of three. Nativists view language as a fundamental part of the human
genome, as the trait that makes humans human, and its acquisition as a natural part of maturation.
They believe that children learn language as normal and natural as dolphins learn to swim. As for
the statistical learning theory and emergentism, the views are somewhat similar to those of
nativist approach with tendencies toward developmental learning theories while the relational
frame theory has its major principles shared with cognitivism and behaviorism and is considered
a mixture of both approaches. This report tries to elaborate on the nature, merits and main tenets
of these approaches trying simultaneously to determine their similarities with nativism.
406
International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org LITERATURE REVIEW
The statistical learning theory
Statistical learning theory was introduced in the late 1960’s. Until the 1990’s it was simply a
theoretical analysis of the problem of function estimation from a given collection of data but now
statistical learning theory is not only a tool for the theoretical analysis but also a tool for creating
practical algorithms for estimating multidimensional functions. The goal of this overview is to
demonstrate how the abstract learning theory established conditions for generalization which are
more general than those discussed in classical statistical paradigms and how the understanding of
these conditions inspired new approaches to function estimation problems.
The main goal of the statistical learning theory is to provide a framework for studying the
problem of inference that is of knowledge gain, making predictions, arriving at decisions or
constructing models from an existing set of data. This is in turn studied in a statistical framework
with assumptions of statistical nature about the underlying phenomena. Statistical learning is a
process through which an individual can track regularities in the input (Graf-Estes et.al., 2006
cited in Sadighi, F. 2012). This can be defined as identifying patterns in the environment. Natural
language contains a great deal of statistical information to be captured by the listener. Statistical
learning can be described as the process of extracting statistical properties of the data input (Kim,
et.al. 2009).
Infants and adults are able to track statistical information in linguistic input. They may also be
able to use that information for phonetic acquisition, speech segmentation, word learning and
acquisition of simple grammatical structures and statistical models of language acquisition claim
that learners are able to identify and track these regularities. Statistical learning can be figured out
as a combination of structural linguistics and nativists perspectives and the distributional
characteristics of natural language reflect underlying linguistic structure and that the development
of language requires learning.
The proponents of statistical theories like Saffron, Aslin and Newport (1996) are in favor of
“innately based statistical learning mechanisms rather than innate knowledge”. This is the
greatest distinction between them and the nativists’ camp since Chomsky and his followers have
underestimated the power of learning and overestimated the need to build the language- specific
knowledge into the organism in advance.
According to this theory, statistical learning of speech and sensitivity to social cues help infants
in the acquisition of their first language. As infants grow up, they become more sensitive to their
particular contrasts and by the end of the first year they are nearly nonresponsive to non-native
contrasts. Speech sounds of a particular language demonstrate significant regularities regarding
the distribution of particular phonemes. These regularities help infants acquire language- specific
consonants and vowels. And as for the segmentation of the spoken language, statistical theorists
say that the key to successful word segmentation is statistical information about words and their
boundaries which exist in the spoken form of natural languages.
407
International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org Yet there are constraints on this approach. Although natural languages are rich with statistical
information that can be tracked by powerful statistical learning mechanisms, such information
may be unrelated for the human language learner in some cases. There are infinite number of
generalizations that are in line with a given linguistic corpus based on which the learner can
compute the related statistical information but some of these generalizations may be irrelevant to
language acquisition. Thus, as Hay (2009, cited in Sadighi, 2012) concludes, statistical learning
must be constrained or biased in order to reduce the acquisition problem.
The Relational Frame Theory
The second theory, or the relational frame theory (RFT), developed by Hayes, Barnes-Holmes
and Roche (2001), provides a wholly selectionist learning account of the origin and development
of language competence. Based on the Skinnerian principles of behaviorism, RFT posits that
children acquire language purely through interacting with the environment. RFT theorists first
introduced the notion of functional contextualism in learning the language, which emphasizes the
significance of predicting and influencing psychological events, such as thoughts, emotions, and
behaviors, by focusing on changable variables in their context. RFT is distinguished from
Skinner's work by identifying and defining a particular type of operant conditioning known as
derived relational responding, a learning process that appears to occur only in humans possessing
a capacity for language. Empirical studies of RFT suggest that children learn language via a
system of inherent reinforcements. This challenges the view that language acquisition is based on
cognitive capacities that are innate and language-specific.
RFT does not believe in specific innateness for language, instead it believes in cognition in that it
claims humans alone are able to deduct relations among objects. Yet sometimes children are not
able to get to this entailment since they are not developed cognitively equally, even at the same
level of age or intelligence.
According to Hayes, Barnes-Holmes and Roche (2001), relational framing is a term for the
process of arbitrarily applicable derived relational responding that is non-arbitrarily applied. This
short definition includes the key concepts of RFT and at the same time distinguishes it from
Behaviorism. RFT is believed to be located, theoretically between empiricism and cognitivism. It
incorporates features from both approaches. It, however, is far from nativism in that it does not
believe in innate abilities helping humans in acquiring language. But the very existence of the
ability to relationally respond to different stimuli and constructing the relations can be considered
as innate abilities, being totally different from the concept of innateness as introduced by
nativists. Hence, the proponents of RFT do not believe in a language device (Sadighi, 2012).
According to Sadighi (2012), The contextual approach of RFT provides a functional account of
the structure of verbal knowledge and cognition, creating an important link between the disparate
perspectives of cognitive and behavioral psychology. Some of the merits of RFT are presented in
table 1.
408
International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org Table 1: Merits of Relational Frame Theory
1. RFT is economic, requiring only the concepts of relational responding, mutual entailment,
combinational entailment and transformation of function to explain the complex processes.
2. RFT is precise, allowing the study of human language to be conducted in accordance with the
carefully-specified definitions of its component processes.
3. RFT has a broad scope, providing plausible explanations and new empirical approaches to a
wide variety of complex human behaviors in basic and applied domains.
4. RFT has depth and analyzes coherence with established treatments at other levels of analysis.
5. RFT is based on empirical research that has supported its tenets and it accounts for the data
observed in hundreds of empirical studies on the concept of stimulus equivalence.
The Emergentist Theory
Emergentist theories posit that language acquisition is a cognitive process that emerges from the
interaction of biological pressures and the surrounding environment. According to these theories,
neither nature nor nurture alone is sufficient to trigger language learning; both of these influences
must work together in order to allow children to acquire a language. The supporters of these
theories argue that general cognitive processes serve language acquisition and that the end result
of these processes is language-specific results, such as word learning and grammar acquisition.
Emergentism is more of an alteration of the nativists’ beliefs. As nativist proponents were unable
to provide accurate accounts of the details of language acquisition, many researchers sought
alternatives to genetically-wired modules (Sadighi, 2012). These tried to explain the formal
structure of language as emerging from the interactions of social patterns that are implicit in the
input.
McWhinney (2005, p.7) maintains that “the emergentist approach to language acquisition views
language as a structure arising from interacting constraints”. Whils many researchers believe that
emergentist approaches to language acquisition stand in direct opposition to theories of language
faculty (innatism or UG), some others claim that the principles of emergentism are not well
defined and there is no general consensus as to how problems of linguistic analysis should be
dealt with (McWhinney, 2005 cited in Sadighi, 2012).
Emergentism, as posited in Sadighi, (2012, p.78), is a reconciliation of empiricism and nativism.
The proponents of emergentism believe that neither empiricism nor nativism is right and both of
them are wrong. Empiricism is wrong since it tries to construct the mind out of nothing and
nativism is wrong for its attempts to make untestable assumptions about genetics and
unreasonable proposals regarding the hard-coding of complex formal rules in neural tissue
(McWhinney, 2002).
DISCUSSION
In brief, Statistical language learning is considered to be located somewhere between pure
innatism and pure developmentalism when it comes to first language acquisition, yet it has a
tendency towards innatism. From an innatist view, statistical language learning posits that human
mind has the potential or natural ability of detecting and acquiring computational data. But rather
than considering language innate, proponents of statistical language learning argue in favor of
409
International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org innately based statistical learning mechanisms. From the developmental end, statistical learning
believes that all human languages have statistical and computational features and that learners
must be exposed to this input so that they can observe those irregularities and acquire them. But,
overall statistical learning theory has only been used to provide an account for the acquisition of
simple aspects of language like phonetics, word recognition and simple syntactic structures.
Additional research is needed to determine whether this approach can provide any account of
more complex language aspects.
However, this approach has been only incorporated to account for the acquisition of simple
aspects of language like phonetics or simple syntactic structures while RFT is claimed to be
closely related to behaviorism. More precisely described, RFT is located somewhere between
empiricism and cognitivism, inheriting features from both approaches. However, it is far from
nativism since its proponents do not believe in innate abilities helping humans in the process of
learning. But it can be related to nativism since the very notion of ability to relate stimuli and
responses can be considered innate. Emergentism, on the other hand, posits that language
acquisition is a cognitive process that emerges from the interaction of biological pressures and
the environment. It is considered to be the reconciliation of empiricism and nativism.
CONCLUSION
This report set out to provide an account of the Statistical Learning Theory, the Relational Frame
Theory and the Emergentist Theory from an Innatist perspective in an attempt to find the existing
common grounds amidst the three to determine whether the aforementioned theories are
inherently similar to Innatism and identify fundamental differences in the way they account for
the L1 acquisition process. The researchers found that it may not be plausible to single out one of
these approaches as being the closest one to the nativists’ views in accounting for the L1
acquisition process, rather it may prove better to identify their claims and fundamental tenets as
matter of their overall position in the continuum of theories in the field since each has their own
contributions as well as criticism.
REFERENCES
Graf-Estes, et.al. (2006). Can infants map meaning to newly segmented words? Statistical
segmentation and word learning. Psychological Science, 18(2), 254-260.
Hayes, S.C., Barnes-Homes, D. & Roche, B. (2001). Relational Frame Theory: a post-Skinnerian
account of human language and cognition. New York: Kluwer Academic Publishers.
Kim, R. et.al. (2009). Testing assumptions of statistical learning: Is it long term and implicit?
Neuroscience Letters, 461, 145-149.
McWhinney, B. (2002). The gradual evolution of language. The evolution of language.
Philadelphia, Benjamins.
McWhinney, B. (2005). A unified model of language acquisition. Handbook of bilingualism:
Psycholinguistic approaches, Oxford: Oxford university press.
410
International Journal of Language Learning and Applied Linguistics World
(IJLLALW)
Volume 7 (3), November 2014; 406-­‐411 Pouladian, N., & Sadighi, F EISSN: 2289-­‐2737 & ISSN: 2289-­‐3245 www.ijllalw.org Sadighi, F. et.al. (2012). A manual of language acquisition approaches and theories. Navid
publications, Shiraz.
Saffron, J. et.al. (1996). Statistical learning by 8-month-old infants. Science, 274, 126-1928.
Woolfolk, A. (2004). Educational Psychology (9th ed.) Boston, MA: Allyn and Bacon.
411