Hipótesis 2: el problema metodológico
4. Transformaciones del paradigma de la planeación urbana
4.2. Crisis del planeamiento
4.2.2. La perspectiva posmoderna y el cambio de paradigma
Table 17 summarizes the significant ERP effects and trends for all word types across tasks. As illustrated in Experiment 1, the meaning acquisition effects for the semantic judgment task are referred to as N400-related, and for the sentence judgment task, N400-congruent. The effects for the FN400, N400-related, and N400-congruent refer to the reduction of the N400 negative amplitude from 300 to 500ms either for familiarity or meaning congruence. The P600 effect refers to greater positivity from 500 to 700ms. Boxes that are marked with a ‘+’ indicate a significant effect, boxes that are marked with a ‘-‘ indicate a trend, boxes that are blank indicate neither.
Table 17. Experiment 2: ERP effects summary
Our first hypothesis for this experiment was that the degree of semantic constraint in which a word has been encountered would result in episodic traces that differ in strength, and that these differences would affect familiarity and recognition. Our data did not support a significant effect of familiarity, however, marginal effects of the FN400 indicated that familiarity was modulated by variability rather than by semantic constraint. Reduced FN400s were observed for repeated-hi and repeated-lo words, suggesting that episodic traces that are strengthened by repetition rather than by availability of meaning modulated familiarity for this experiment. As observed by the trends of the data, there was small difference between the FN400 magnitude for repeated-hi words (.87) and repeated-lo words (.78), thus further reinforcing the idea that familiarity was modulated by episodic memory strength. These results do not replicate our findings from Experiment 1 that indicated that familiarity was modulated by the availability of meaning (i.e., repeated D and varied C conditions). These differences in familiarity effects across the two experiments may be due to a learning strategy adopted by the learners during learning in the absence or presence of dictionary definitions. In Experiment 1, the availability of explicit information about the word’s meaning from dictionary definitions may have shifted the learners’ attention to conditions in which they could derive the word’s meaning as a way to remember the words. In Experiment 2, the absence of explicit meaning information
may have shifted the learners’ attention to learning conditions in which the words were repeated as a way to remember the words.
However, the trend observed for the P600 old/new effect suggests that differences between the magnitudes for ‘old’ words are affected by the availability of meaning. Among the learning conditions, most positive amplitudes from 500 to 700ms were observed for repeated-hi and varied-hi words; for these conditions, the single-sentence contexts were supportive of the word’s meaning, whereas varied-lo and repeated-lo conditions were to a lesser extent. Across all word types, varied-lo words showed the most negative amplitudes, suggesting that the absence of meaning information in combination with a larger sum of diffuse contextual memory traces does not facilitate the recollection of a recently learned word. A similar trend was observed in Experiment 1 for repeated-D, varied-D, and varied-C conditions; learning conditions which are supportive of the word’s meaning. Data from across these two experiments suggests that the ‘depth of encoding’ of a word’s meaning from a recent learning episode may modulate the recollection of ‘old’ items (i.e., Mestre-Misse et al., 2007).
Our second hypothesis for this experiment was that semantic constraint and variability will result in differences in meaning acquisition. Specifically, we predicted that varied contexts of high semantic constraint would result in better learning of a word’s meaning than single- repeated contexts and varied contexts of low semantic constraint. Meaning acquisition was measured in two ways; in a new-single word and in new sentence contexts that differed from those that were encountered during learning. In a single-word context, we observed a trend for repeated-hi words with a reduction of the N400 for related words. This suggests that in the repeated-hi conditions, learners were able to acquire some aspects of the word’s meaning to make the relatedness judgment. As predicted, a significant robust reduction of the N400 was
observed for related trials in the familiar words. However, our data does not support our prediction that varied-hi words would lead to better meaning decisions in single-word contexts. No reduction of the N400 was observed at any hemisphere or lobe locations for this condition. Across all four learning trials for the varied-hi conditions, the meaning of the word was supported with different content. Learners may have had difficulty encoding a unified meaning representation of the word from this varied content. The absence of such a representation would hinder a learner’s ability to make a relatedness judgment in a single- word context. Repeated-hi words have an advantage for the semantic-judgment task because a single encoded word is abstracted and strengthened with each encounter, thus yielding a more unified meaning representation of the word. As expected words learned with repeated-lo and varied-lo conditions did not show significant reductions of the N400 for related trials; neither condition supports the abstraction of word meaning from context.
Results from the sentence judgment task support our prediction about contextual variability facilitating more accurate contextual decisions. Our data indicated a reduction of the N400 for congruent trials for the varied-hi and varied-lo conditions, which was lateralized at central-midline locations. A smaller, but a signification reduction of the N400 was also observed for repeated-hi congruent trials. These results indicated the availability of meaning knowledge and the availability of word-use knowledge are important for making contextual decisions. For the varied conditions, varied-lo words generated a larger N400 magnitude (1.60), but not significantly different than for the N400 magnitude for varied-hi words (1.34). This suggests that learners may benefit from variability of content in the absence of definitional meaning information to make accurate contextual decisions. In the case where meaning information is available (i.e., repeated-hi and varied-hi), this contextual information may not be as critical.
5.5.3 Individual Differences
Correlations with behavioral and ERP data indicated that orthographic, phonological, and semantic knowledge is correlated with learning new word meaning from context, and that these skills are especially critical for learning encounters where the context sentence is not supportive of the word’s meaning (i.e., varied-lo condition). Vocabulary scores and scores from the spelling, real word, and PHaT tests correlated with bigger N400 congruency magnitudes for the varied-lo condition. We also observed that the Raven’s measure of non-verbal intelligence was positively related to bigger N400 magnitudes for low constraint contexts. This correlation may indicate that learners with better non-verbal intelligence may employ different strategies to learn meaning from a low constraint context. These correlations suggest that lexical skills, and perhaps non-verbal intelligence, are especially critical for learning new meaning in vague contexts. These results correspond to other findings (e.g., Ouellette, 2006; Cain, Oakhill, Lemmon, 2004) that illustrate that high-skill readers with better lexical and comprehension skills are more successful at learning new word meaning from context.