continuity and parsimony

Note: this is a series of excerpts from an essay I wrote a couple of years ago, as part of a discussion of two competing theories of how children acquire phonology. In the interests of avoiding plagiarism, if anyone wants to cite it or borrow it in an academic context, please contact me first so that I can give you all the relevant information (leave a comment and I’ll get back to you) .

The crucial point of division between constraint-based models of development and word-based models is whether they admit development of mental representations as well as development of production.

In the word-based account, in short, what stays constant is cognition, rather than grammar. This is how Tomasello (2000) characterises development:

“there is continuity not of structures – adults control a more diverse and abstract set of constructions than do children – but there is continuity of process in the sense that the processes of learning and abstraction are the same wherever and whenever they are applicable” (2000: 237, original italics).

Others emphasise the continuity that is to be found in the sociability of the child, their need to interact with people, which just becomes more sophisticated in its realisation over time:

“infants do not really set out to learn language. Instead, they study the movements of faces and voices – the observable displays of talkers – and gradually accommodate to and reproduce these behaviours. They do this not because they know about language and understand its importance, but because they have a deep biological need to interact emotionally with the people that love and take care of them.” (Locke 1993: 7-8).

There is supporting evidence for such continuity in some of the influences on speech processing identified in pyscholinguistic studies that seem to affect people in the same way throughout life. Storkel and Morrisette (2002) review studies that show that there are abilities that are present in infancy, continue through childhood and adolescence, and remain in adulthood – for instance, the ability to differentiate between frequently and infrequently occurring words (Jusczyk & Aslin 1995) as well as between words composed of common versus rare sequences (Jusczyk, Luce, & Charles-Luce 1994), and sensitivity to phonotactic probability (Messer 1976; Vitevitch & Luce 1999). Finding the same facilitatory and inhibitory effects for infants, children, adolescents, and adults is taken as evidence of continuity between these phases of the lifespan.

That the word-based model identifies this as the locus of continuity highlights the major point of disagreement with the constraint-based approach – while both accounts include ‘continuity’, they identify continuity in two separate areas. In their discussion of the word-based account the main reason of the two provided by Bernhardt and Stemberger (1998) for preferring to work in the constraint-based approach is that they want to uphold the continuity hypothesis. Their critique of the word-based model centres on the validity of the motivation for the continuity hypothesis, namely the resulting parsimony of accounts that adopt it.

However, although parsimony is frequently invoked in generative theories of grammar, there are limits to its application. When two models are being compared, and when both account for the data equally well, it is legitimate to prefer the one that posits fewest entities, only if the models are in all other respects the same. That is, parsimony cannot be judged simply by picking a feature that belongs to the first model and not the second, and concluding on that basis that the second is therefore more parsimonious.

This is particularly true of models that do not arise from the same theoretical background assumptions, as is the case for the models Bernhardt and Stemberger discuss – within the generative framework, for example, it may be valid to weigh up alternatives and accept or discard them on the basis of parsimony, but the word-based model they criticise for being non-parsimonious is not one of a kind with the constraint-based model they adopt. Both accounts have ‘representations’ and ‘cognitive development’ – one says the first changes but not the second, and the other says the second changes but not the first – in terms of parsimony it could be argued this is a tie, except that in order for parsimony to be a substantive term, the whole range of auxiliary assumptions and motivations need to be invoked as the basis for comparison.

Because its background assumptions are different, there are other differences between the word-based model and the constraint-based model, in particular significant qualitative differences, over and above the notion of continuity, such that the ‘continuity hypothesis’ outlined by Bernhardt and Stemberger does not apply to the word-based model. Inconsistency with the continuity hypothesis is just not an issue for models that are uninterested in symbolic mental representations.

References

Bernhardt, BH and Stemberger, JP (1998), Handbook of Phonological Development From the Perspective of Constraint-Based Nonlinear Phonology. London: Academic Press
Jusczyk, PW and Aslin, RN (1995), ‘Infants’ detection of the sound patterns of words in fluent speech.’ Cognitive Psychology 29: 1-23. Cited in Storkel & Morrisette (2002)
Jusczyk, PW, Luce, PA, and Charles-Luce, J (1994), ‘Infants’ sensitivity to phonotactic patterns in the native language.’ Journal of Memory and Language 33: 630-645. Cited in Storkel & Morrisette (2002)
Messer, S (1967), ‘Implicit phonology in children.’ Journal of Verbal Learning and Verbal Behaviour 6: 609-613. Cited in Storkel & Morrisette (2002)
Storkel, HL and Morrisette, ML (2002), ‘The lexicon and phonology: interactions in development.’ Language, Speech, and Hearing Services in Schools 33: 24-37
Locke, J (1993), The Child’s Path to Spoken Language. London: Harvard University Press
Tomasello, M (2000), ‘Do young children have adult syntactic competence?’ Cognition 74: 209-253
Vitevitch, MS and Luce, PL (1999), ‘Probabilistic phonotactics and neighbourhood activation in spoken word recognition.’ Journal of Memory and Language 40: 374-408

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s