Professional Documents
Culture Documents
LeGard (2004) OU
Pinker and Prince (1988) proposed dual-route theory in which one route is rule-
governed and enables the formation of past-tense regular verbs. A second route
pertains to a memory system of irregular past-tense forms. The production of an
irregular past-tense form is only possible once the appropriate past-tense form
has been learnt and memorized, and if it can be retrieved before the rule-
governed route − operating in parallel − produces an incorrect regularization.
Dual-route theory maintains that over-regularization errors occur when the
memory system is insufficiently developed to prevent the rule-governed route
from overriding it. With experience, memorization of infrequent inflections is
consolidated and errors are reduced. The inflection that occurs most (/ed/ in
English) is established as the default rule (Marcus et al., 1992).
*
These dual- and single-route explanations have traditionally been associated with nativist and empiricist
accounts of language development. Although dual- and single-route theories are not implicitly related to these
positions, proponents of dual-route theory propose that the rule mechanism is innate while single-route
theorists dispute this view.
Accounting differently for over-regularization errors, single-route theory asserts
that memory resources become congested as numerous inflections are stored.
This leads to ‘interference effects’, that is, similar words are confused and so, for
example, the irregular verb ‘go’ is confused with other regular verbs with a
similar phonological structure (e.g. mow, row, sew) and ‘goed’ is produced. With
experience, the irregular verbs become more securely stored and are able to
resist the interference effects of regular verbs.
Both single- and dual-route theories, then, explain the U-shaped pattern.
However, an examination of the different kinds of mistakes predicted by both
theories appears to support the single-route approach. Although dual-route
theory maintains that particular types of error should not take place, irregular
inflections suffixed to regular verbs occur (although infrequently) in children’s
speech. Indeed, single-route theory predicts this type of error, asserting that
interference effects influence children’s production of regular and irregular
words.
Most nouns and verbs in English are regular and take a single ending. German,
however, has multiple regular endings. Marcus et al. (1995) claim that the
default rule in German causes over-regularization errors and is used for inflecting
novel words. Thus, if add ‘/s/’ is the default, it should feature in children’s over-
regularization errors. A longitudinal study by Szagun (2001), however, revealed
that single-route theory is preferable for acquiring languages with more than one
regular inflection. The participants’ use and frequency of the different plural
inflections was consistent with the occurrence and frequency of those inflections
in their parents’ speech. Errors in the use of plural formation occurred from the
onset of the children’s speech. These findings suggest that − consistent with a
single-route account of inflectional morphology − children initially generalize
from the different regularities of German inflection they are exposed to.
Prior to the vocabulary spurt, young children focus upon a small number of words
(Harris, 2006). However, networks require a large number of input-output
pairings to operate efficiently. Moreover, it is uncertain whether a computer
model actually reflects the operation of a child’s brain. Bruner (1993) asserts
that children encounter and acquire language in familiar social contexts. Indeed,
Harris et al. (1988) discovered that there exists a close relationship between a
child’s use of first words and the mother’s use of those same words. Children
interact with others who assist them. Connectionist models, however, exist in a
pre-programmed, artificial environment.
Nativists claim that, via Universal Grammar, children recognize that languages
can be principally syntactic or morphological. This innate cognitive mechanism
contains a series of constraints on language processing. Through exposure to the
linguistic environment, constraints that relate to particular languages (e.g. those
that exploit syntactic cues) are activated and constraints that do not apply are
deactivated.
Elman’s (1993) neural network was able to calculate the correct sequence of
categories in a sentence. At a later stage, the network improved learning when
initially exposed to simple grammatical sequences. Indeed, child-directed speech
(Snow, 1972) assists babies in making sense of the complex speech stream,
which bears out the notion that syntactic development proceeds in a piecemeal
fashion.
It has been established, then, that the recent challenges to the nativist approach
present a more credible account of language development. Single-route theory
predicts particular types of inflectional errors that dual-route theory cannot
account for. Single-route theory offers a more convincing explanation of how
children generalize inflections to novel words, and proves more advantageous for
learning languages with more than one regular inflection. The
psychopathological evidence, however, is ambiguous. Neural networks can
produce a U-shaped pattern of development. However, they possess several
limitations. While it is cognitively challenging to work out the significance of
syntax and morphology from the linguistic environment, child-directed speech
seems to support the notion that development of syntax occurs in a piecemeal
fashion.
Although nativist accounts have been challenged, it seems that innate abilities
and social experience operate together to promote language development.
Indeed, little progress will be achieved if psychologists accept either extreme
empiricism or pure nativism (Bruner, 1983, p.132).
References
Bellugi, U., Bihrle, A., Jernigan, D., Trauner, D. and Dougherty, S. (1990) cited in
Plunkett, K. and Wood, C. (2006) p.189.
Gopnik, M. and Crago, M. (1991) cited in Plunkett, K. and Wood, C. (2006) p.189.
Harris, M., Barrett, M., Jones, D. and Brookes, S. (1988) cited in Harris, M. (2006)
p.87.
Harris, M. ‘First Words’, in Oates, J. and Grayson, A. (eds) (2006) Cognitive and
Language Development in Children, Oxford, Blackwell/The Open University.
Marcus, G., Brinkmann, U., Clahsen, H., Wiese, R. and Pinker, S. (1995) cited in
Plunkett, K. and Wood, C. (2006) p.188.
Marcus, G., Pinker, S., Ullman, M. et al. (1992) cited in Plunkett, K. and Wood, C.
(2006) p.183.
Pinker, S. and Prince, A. (1988) cited in Plunkett, K. and Wood, C. (2006) p.182.
Plunkett, K. and Juola, P. (1999) cited in Plunkett, K. and Wood, C. (2006) p.184.
Thomas, M. S. C., Grant, L., Barham, Z. et al. (2001) cited in Plunkett, K. and
Wood, C. (2006) p.190.