• No se han encontrado resultados

El catálogo Ikea para Arabia Saudí: un hogar sin mujeres

27 2.1.1.La publicidad “imposible”: algunos ejemplos

2.2. La publicidad: una esperanza para la mujer saudí

2.2.3. El catálogo Ikea para Arabia Saudí: un hogar sin mujeres

Forster (1979) proposed that word processing has afunctional autonomy, and that lexical identification was encapsulated and immune from contaminations by context, and Fodor (1983) went further and proposed a highly modular mental architecture. If one were to try to shoehorn the ideas of n-gram pro- cessing into a series of modular processes, the results would be dismal. There is no discrete set of black boxes operating independently and in a certain order that can explain the massively interactive effects of information and conditional probability in my studies. The only solution is to accept the emer- gentist school of thought and reject this serialized way thinking. By allowing all levels of processing to fully interact, the observed behavior begins to make sense. N-gram recognition processes interact with word recognition processes, transitional probabilities interact with sub-lexical statistics, and new input continuously changes the state of the system.

There are still those who hold tight to the “symbols and rules” approach to language systems. Jackendoff (2002) shrugs off emergentist models as irrel- evant to human language comprehension. His most current model is steadfast in its loyalty to sequential, directed processing (Jackendoff, 2007). Meanwhile SRNs and other types of models are quickly beginning to dominate the field.

These detailed computational models are capable of explaining how meaning is constructed as language unfolds. To add to the already long list of models using SRNs, Crocker, Knoeferle, and Mayberry (2010) have created an emergentist model of language based on Elman’s SRN, that computes the interaction of utterances and visual attention. Misyak, Christiansen, and Tomblin (2010) have used an SRN to predict human performance patterns in a sequential learning task. This is the first study to provide a link between linguistic and non-linguistic learning of probabilistic sequences 2.

Another emergentist model that uses truly sub-symbolic information to explain lexical and supra-lexical phenomena is the Naive Discriminative Reader model developed by Baayen et al. (2011). This model is a two-layer symbolic network model built using the equilibrium equations of the Rescorla-Wagner model (Danks, 2003). Building a complex mapping between letter bigram cues and semantic representations, it is able to simulate frequency effects, morphological family size effects, and some initial n-gram frequency effects (Baayen & Hendrix, 2011). The NDR reader is a truly emergentist model, with all these effects arising simply from the interaction of all the sub-symbolic learning taking place at the letter bigram level.

Are there only two types of models at play, or are there more worth con- sidering? The simpler associationist, network models are more attractive than the complicated, nativist systems. Is there a third way? One family of non- connectionist models are the Bayesian models. The Bayesian Reader (Norris & Kinoshita, 2008) is one such model. Bayesian models represent the world as a set of competing hypotheses, and the task of the models is to choose the most likely hypothesis, given the evidence and the level of uncertainty surrounding it. Tenenbaum, Kemp, Griffiths, and Goodman (2011) argues that Bayesian models are valid and relevant, but Kwisthout, Wareham, and Rooij (2011) have criticized these models on the grounds that they cannot be approximated or computed within a reasonable amount of time.

2Incidentally, Christiansen, Conway, and Onnis (in press) found that linguistic and non- linguistic sequential learning, as modelled by Misyak et al. (2010), produce very similar ERP responses, implying that language shares domain-general processing substrata with other sequence processing systems.

There is still a strong resistance to emergentist models in many quarters. Why is it so hard to shift from sequential viewpoints to emergent viewpoints? Some have proposed that the sequential process schema are an ingrained, intu- itive first attempt at explaining all phenomena (Chi, 2005; Chi, Roscoe, Slotta, Roy, & Chase, 2011). Many phenomena that we observe our whole lives, such as the way that our blood moves around (directed by the heart, moving from point A to point B) are not emergent. When young adults and even some scientists are faced with emergent phenomena, such as the way ants collect food (without the aid of a queen ant telling them where to look or how to co- operate), they are initially resistant to the notion of emergence. In chemistry classes, for example, the way that ink diffuses in water is confusing to students, even after the emergent explanation is taught. This is one of many examples of emergent phenomena that are often misunderstood. Chi et al. (2011) has listed criteria for identifying emergent phenomena, and they fit the language system to a tee. Using their terminology, words would be the “agents”, and they note that (1) in emergent systems the interactions of the entire collection of all the agents together “cause” the observable pattern, not any one special type of agent. (2) All the interactions have equal status with respect to the pattern. (3) Agents’ interactions and the pattern can behave in disjoint or non-matching ways. (4) Interactions are undertaken by the agents with the intention of achieving local goals only, without any intention of causing the (changes) in the pattern. The pattern emerges from the local interactions of all the agents. (5) The pattern is caused by the collective summing or net ef- fect of all the interactions at each point in time. (Chi et al., 2011, p. 10). This difficulty in understanding the difference between emergent and non-emergent phenomena in all the sciences (including physics, chemistry and biology) is a ongoing issue, but psycholinguistics should move forward and accept the fact that the mind is a emergent phenomenon and that there is no other way to explain the combined action of all the brain activity we observe.

Documento similar