Exploration of Different Aspects Related to Statistical Learning - Lab Report Example

Exploration of Different Aspects related to Statistical Learning al affiliation: Statistical learning is considered to be a simple domain as well as a general mechanism which could account for a complex linguistic behavior or speech segmentation in both second language learning and infant language acquisition. The study found that the statistical learning can still be observed in case of different length of words. Also a larger number of words can make statistical learning harder. In addition, it was found that longer exposure can’t compensate for larger number of words. Introduction Natural learning entails one of the major complicated human cognition aspects, yet children are seen to have a good native language grasp prior riding a bicycle or tying their shoes (Jusczyk and Krumhansl, 1993). The acquisition ease indicates that when a guess is made by the children about the language structure based on the apparent evidence that is limited, the child has a tendency of guessing right. There is clear evidence that there should be a very close relationship between the processes where the child processes and acquires language as well as the language structure (Reber, 1993). There is a likelihood for infants to use different types of information in a tandem manner in discovering word boundaries that include silences prosodic clues at the utterances end (Aslin et al., 1996).One clue that could be relevant in word segmentation include statistical information that is derived from the sound’s pattern distribution (Christophe et al., 1994; Catani et al., 2005). The assumption that the word boundaries could be explored through computing the sound sequences in linguistic input is seen to have a long history that dates back to the eminent structural linguist Harris (1955). It was noted by Harris that various sounds might follow the last word sound. For instance, the last sound in antelope could be followed by any word based on the fact that grammar permits it to appear in the next utterance. Contrary to this, sound sequencing that is within a word is considered to be more constrained. There is a higher expectation that the next sound sequence could be vator or phant when given the sequence like ele (Bear et al., 2001). The main reason why this study was undertaken is to be able to replicate the statistical learning phenomenon as well as attempt to answer the three questions as below. Research questions Is statistical learning still seen in case of different length words? Do a larger number of words make statistical learning harder or easier? Is the exposure duration a critical factor or can longer exposure compensate for larger number of words? The study hypothesis Statistical learning is seen to challenge the fact that language learning needs complex and specific mechanisms Method Participants The undergraduate students from the University of York, Department of psychology were considered in the study. The sample size considered was 207 students. Materials and apparatus Description of the stimulus used How the stimulus was presented The repetition of the words was done in a random order in order to create the stream. The pitch for every word was identical with 120Hz.The syllables between and within the words was equally co-articulated. Design Score for group 1: 4-word language, 6 min Score for group 2: 6-word language, 6 min Score for group 3: 6-word language, 9 min Procedure The exposure phase consisted of streams where, listening carefully as well as trying to discover the language words was undertaken. The test phase included 16 pairs of stimuli which was 4 word languages or 24 pairs of stimuli with 6 word language. Each word was paired with a syllables sequences heard less frequently; the pairs consisted of different or same length sequences; half the pairs started with the word while the other half started with a non-word; the pairs were randomized and 500-ms inter-stimulus interval (ISI). For every pair, a decision was made on which stimulus was the language word that was heard and this was based on the confidence scale that include 321 and 123. In regard to this, one of the blue numbers was circled only if the word language was thought to be the first stimulus heard. In case, it was thought to be the second stimulus, the red numbers were circled. The number showed the confidence level of the decision made where, 1 represented somewhat confident; 2 represented moderately confident and 3 represented highly confident. Results Data coding Word: 321(coding: 321) Non-word: 123 (coding:-1-2-3) Every participant ended up with a single score and it was the mean score for all the trials heard. The scores were between -3 and +3 (above 0!) Figure 1.Bar graph with the mean scores of the three groups Figure 1 shows the mean scores of the three groups and it is observed that group 2 has the highest mean scores, Data analysis One sample t test One sample t test was run to determine if the sample (the scores) comes from the normal population with a mean of zero. Score for group 1: 4-word language, 6 min Table 1. One-Sample Test Test Value = 0 t df Sig. (2-tailed) Mean Difference 95% Confidence Interval of the Difference Lower Upper group1 9.011 61 .000 .81629 .6352 .9974 Table 1 shows that the p value is less than 0.05 and it can be concluded that the Score for group 1 is statistically significantly higher than the population normal score. Score for group 2: 6-word language, 6 min Table 2. One-Sample Test Test Value = 0 t df Sig. (2-tailed) Mean Difference 95% Confidence Interval of the Difference Lower Upper group2 5.208 74 .000 .34760 .2146 .4806 Table 2 shows that the p value is less than 0.05 and it can be concluded that the Score for group 2 is statistically significantly higher than the population normal score. Score for group 3: 6-word language, 9 min Table 3. One-Sample Test Test Value = 0 t df Sig. (2-tailed) Mean Difference 95% Confidence Interval of the Difference Lower Upper group3 6.703 68 .000 .45449 .3192 .5898 Table 3 shows that the p value is less than 0.05 and it can be concluded that the Score for group 1 is statistically significantly higher than the population normal score. Independent samples t test Between group 1 and group 2 Table 4. Independent Samples Test Levenes Test for Equality of Variances t-test for Equality of Means F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Score Equal variances assumed 3.135 .079 4.249 135 .000 .46869 .11031 .25053 .68685 Equal variances not assumed 4.165 116.829 .000 .46869 .11252 .24585 .69153 Table 4 shows that the p values is less than 0.05 and it implies that there is a statistically significant difference between group 1 mean scores and group 2 mean scores. Between group 2 and group 3 Table 5.Independent Samples Test Levenes Test for Equality of Variances t-test for Equality of Means F Sig. t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper Score Equal variances assumed .073 .787 -1.122 142 .264 -.10689 .09525 -.29518 .08139 Equal variances not assumed -1.124 141.524 .263 -.10689 .09514 -.29498 .08119 Table 5 shows that the p values is greater than 0.05 and it implies that there is no statistically significant difference between group 2 mean scores and group 3 mean scores. Discussion The study found that the statistical learning can still be observed in case of different length of words. Also a larger number of words can make statistical learning harder. In addition, it was found that longer exposure can’t compensate for larger number of words. In relation to the hypothesis, the phenomenon was replicated and this is supported by the literature where, natural learning entails one of the major complicated human cognition aspects (Bloom, (2001; Carpendale and Lewis, 2004). The limitation of the study was that the confounding factors were not considered and they could have an effect on the results. The strength of the study was that the sample size was the representative of the population. The study gives a foundation of future studies where, the confounding factors should be considered. It can be concluded that Statistical learning can be suggested to challenge the fact that language learning needs complex and specific mechanisms References Aslin, R.N., Woodward, J.Z., LaMendola, N.P., Bever, T.G., (1996). Models of word segmentation in maternal speech to infants. In: Morgan, J.L., Demuth, K. (Eds.), Signal to Syntax. Erlbaum, Hillsdale, NJ, pp. 117–134. Christophe, A., Dupoux, E., Bertoncini, J., Mehler, J., (1994). Do infants perceive word boundaries? An empirical study of the bootstrapping of lexical acquisition. Journal of the Acoustical Society of America 95, 1570–1580. Harris, Z.S., (1955). From phoneme to morpheme. Language 31, 190–222. Bloom, P. (2001) Pre´ cis of How children learn the meanings of words. Behavioral and Brain Sciences 24(6):1095–103. Bear M. F., Connors, B. W. & Paradiso, M. A. (2001) Neuroscience – Exploring the brain, 2nd edition. Lippincott, Williams & Wilkins. Carpendale, J. I. M. & Lewis, C. (2004) Constructing an understanding of mind: The development of children’s social understanding within social interaction. Behavioral and Brain Sciences 27:79–151. Catani, M. C., Jones, D. K. & Affytche, D. H. (2005) Perisylvian language networks of the human brain. Annals of Neurology 57:8–16. Jusczyk, P.W., Krumhansl, C.L., (1993). Pitch and rhythmic patterns affecting infants’ sensitivity to musical phrase structure. Journal of Experimental Psychology: Human Perception and Performance 19, 627–640. Reber, A.S., (1993). Implicit Learning and Tacit Knowledge: an Essay on the Cognitive Unconscious. Oxford University Press, New York. Read More