What is the Nature of Lexical Processing? An Investigation of Gender Effects on Categorisation

Abstract

An investigation of how words are recognised and processed in the mind, reviewing the influences on lexical processing and reporting a small experiment testing whether gender has an effect on how we categorise words.

Introduction

This experiment is set up for the module Mind and Language at the University of Brighton. In order to answer the question: "What is the nature of lexical processing? An investigation of gender effects on categorisation”, an empirical investigation will be carried out. The following chapters will reveal how this experiment is structured with the outcome to answer this complex question. It is of importance to understand how word recognition works, what lexical processing means, are there other influences that might create a difference in the way we process words, and could gender have an effect on how we process and categorise words? In order to understand the results of this essay, it is important to fully understand the term “lexical processing”. Lexical processing refers to things regarding the words or lexicon of a language in the mind. The mental lexicon is not literally similar to a dictionary as we know it, it seems more complex with associations between phonologically and semantically related lexical items (Aitchison, 2012). The key difference between auditory and visual lexical processing will be discussed in more detail in chapter 2.

Supporting research

This chapter will demonstrate how the idea for this experiment was formed. It will investigate a variety of theories: those with ideas that concur with one another as well as those with opposing ideas. 2.1 What is word recognition? There are two types of word recognition. One is visual, the ability of a human being to recognise written words correctly and practically without any effort (Harley, 2014). Whereas, the second one, acoustic word recognition, requires a higher knowledge, such as, lexical, syntactic, semantic and pragmatic information (Harley, 2014). While written words are there to be read repeatedly, spoken words are not and due to the higher knowledge mentioned earlier, humans are experts in recognising speech. Despite the invariance and segmentation problem, speech recognition is done automatically and therefore is effortless (figure 1) (Harley, 2014). Figure 1 Source: Adapted from Harley 2014 There is a reasonable answer to why humans are experts in speech recognition. Language has existed for perhaps 150,000 years, at least 80,000 years, and initially arose as speech. That is what humans are probably genetically specified for. Reading and writing is something that came along much later, there is a little bit of controversy as to exactly when that happened, but according to traditional estimates, if humanity had existed for 24 hours, then writing only came along at about 11:07 p.m.(McWorther, 2011). Nonetheless, Jarema and Libben (2007) argue that the starting point of any investigation is the role of phonological and phonetic properties is through a consideration of the individuality and variability related with spoken words. The written word is, by design and context, substantially more constant synchronically and more traditional diachronically than the spoken word. 2.2 What influences our lexical process? According to Jarema and Libben (2007, p. 2) the mental lexicon can be defined as: The cognitive system that constitutes the capacity for conscious and unconscious lexical activity. This suggests, the term “system” is the effective fact that is the characteristic of lexical activity. Most importantly, the core element of the mental lexicon is the manner in which lexical representations are fundamentally linked in complex multidimensional networks (Jarema, Libben, 2007). The mental lexicon is not simply the source of representations that enables lexical activity, it relatively grasps the claim that the mental lexicon is that lexical activity. The use of the term “capacity” highlights the importance of the productivity and power in lexical knowledge and ability. Lexical decisions are made easier when the related words become activated in the lexicon. There is another phenomenon that occurs in the lexicon, Hoey (2005) utilises the term “semantic preference”, and in other words, individuals automatically prefer semantically related words. Also mentioned, the interchangeable “semantic association” in which Hoey (2005, p. 63) states: “The formulation of quite complex representations of naturalness without jeopardising our ability to account for creativity in language. We have seen, however, that it may sometimes not be the word or word sequence that is being primed but the semantic set, created by the operation of abstraction from a variety of individual primings.” This suggests, the answers between participants will differ due to their individual priming (association) with certain semantically and/ or phonetically related words. However, semantic priming refers to the observation that a response to a target (e.g. Train) is faster when it is preceded by a semantically related prime (e.g. Track) compared to an unrelated prime (e.g. Apple). Semantic priming may occur because the prime partially activates related words or concepts, simplifying their later processing or recognition (Harley, 2014). Although this process is often automatic, priming can also be guided by the use of specific strategies to achieve a particular task (figure 2). Figure 2

Source: Adapted from Harley 2014 All theories above support the hypothesis that participants should achieve better results in the acoustic part of the experiment in this essay. 2.3 What other influences are there? Humans are genetically similar and different at the same time. Same in matters of belonging to one human race and different because of their parents, their parents-parents and so on. Our personal experiences in life cause changes in the way we think and act. This also has an influence on how we understand words and the associations we create in our own minds. Altmann (1997) takes this thought to another level by raising the question, “What is experience to a bunch of nerve cells?” Altmann claims that there is enough evidence to say that neural activity changes due to the result of accumulated experiences. One of the examples he provides is that of Pavlov’s dog: the dog has learned the relationship between the bell and dinner. Humans are capable of connecting complex and in-depth tasks, unfortunately we are not all capable of achieving the same developments throughout our lives. This does raise the question of how accurate experiments can be if we are all different in the way we think and process information. 2.4 Could gender have an effect on how we categorise words? As far as writing goes back, Plato stated that: “Women are from Venus and Men are from Mars.” The physical differences between men and women place different demands on the brain. This would suggest a different set of neural pathways are in use. These differences may extend to language processing, however, studies remain controversial. On one hand, research done by Ingalhalikar (2013) show how much the findings supported old stereotypes that women are better at multi-tasking, organising, languages etc. The reason for this could be because men’s brains are apparently wired more for perception and co-ordinated actions, and women’s for social skills and memory, making them better equipped for multi-tasking. Brain imaging was used to map the neural connections of 428 males and 521 females. The scans of neural circuitry showed that, on average, women’s brains were highly connected across the left and right hemispheres in contrast to men's brains, where the connections were typically stronger between the front and back regions as seen in figure 3. Figure 3 Source: adapted from Ingalhalikar 2013 This suggests that men focus on the task given to them and on that task only, while women make other connections at the same time, thinking of other options and possibilities. Therefore, I hypothesis that in the experiment females should have better results than men when it comes to categorising semantically and phonetically related words at the same time in the right domain. In addition, if they are making more connections that would suggest that women should respond slower than men. Men, conversely, should focus on one topic and provide answers faster in relation to time, as they focus on one task only. On the other hand, research done by Daphna Joel at Tel-Aviv University (2014), conducted MRI scans of 1400 people, comparing traits, relationships, activities, and attitudes. Joel’s results conclude that the differences between males and females are exceedingly small, and cannot be categorised by gender. The two results should not be compared as they investigate different areas. Ingalhalikar researched the ages 8-22, and looked at the development of the brain in gender, whereas, Joel looked at sexual orientation and compared results of adult men, women, and LGBT (lesbian, gay, bisexual, and transgender). Comparing the experiment in this essay with Joel’s should conclude that, all participants should be seen as individuals and take the average of all results at the end, by then splitting it into gender.

The experiment

The experiment used a semantic and phonological relatedness task, in the form of an online –quiz, to investigate the nature of lexical processing, and whether gender has an effect on how we categorise words. Empirical analysis of the task results will be used to test these claims: Participants should respond more efficiently when the pair of words are presented acoustically in all categories, as that is the natural approach. Participants will differ in results due to their individual priming, however, when the focus is set on the words that are semantically and phonetically related the participants will focus on the semantically related words rather than both options, due to the automatic semantic preference. Female participants are expected to score higher when categorising semantically and phonetically related words.

3.1 Participants, Design and Procedure

Data was gathered at the University of Brighton, participants are all students of the English Language, Linguistics, or Literature department. The experiment conducted for this essay took place in two parts. The first part of the experiment was conducted by eighteen females and thirteen males, they were presented with pairs of written words and asked if the words were semantically related (e.g. apple-orange), phonetically related (e.g. blue-black), semantically and phonetically related (e.g. train-track), or unrelated words (e.g. type-nose). The second part of the experiment was presented a week later, the participants (eight females and ten males) were asked to make the exact same decisions, this time the words were presented acoustically. Figure 4. Introduction page of the test

Source: LimeSurvey The participants were presented 40 sets of words (see appendix 1): Ten pairs of Semantically related words Ten pairs of Phonetically related words Ten pairs of Semantically and Phonetically (S+P) words Ten pairs of Unrelated words In the visual part of the test the participants were shown pairs of words and were asked to answer by simply choosing the correct options provided within a certain amount of time on their computers or mobile phones. In the acoustic part of the test, the participants were asked to begin an audio clip, after hearing the word-pair, a timer appeared and the participant then had 5 seconds to provide an answer.

3.2 Data Analysis

To simplify data analysis, the results displayed are analysed on a per-category basis, not per question. Full question-level results are available in the appendix. Figure 5 presents an overall view of the results (full discussion can be found in Chapter 4), namely: The phonetically related words were the most accurately categorised group. The visual part of the phonetically given words are more accurately performed by women (93.1%) than men (84.5%). Figure 4 shows that, for the acoustic portion of the phonetically given words, men (92.3%) outperform women (90.2%). In addition, male performance increased dramatically by 8.2% whereas female performance reduce by 3.1%. This can be explained by the fact that the number of female participants were lower than in the visual part of the test (chapter 3.1). Overall, both sexes score highest on the phonetic part of the test in both visual and acoustic. The unrelated words were the second most accurately categorised group by both sexes and women were better than men in this category. It is also interesting to note that the response time was faster in this part of the test. For the semantically related words: participants have shown that the differences are exceedingly small in both visual and acoustical parts of the test. All participants had a better result in the visual part (men 83%, women 81.5%) and men answered more accurately than women in both areas. The semantically and phonetically related words were the most challenging part of the test and as a result both sexes scored exceedingly low on this part (Visual: men 39.9%, women 31.8%. Acoustic: men 44.3% women 35.1%). One way or another, it appears that both sexes halved their scores as they had to click both options to have a positive score. As the hypothesis suggested, both sexes were more focused on the semantic relationship of the word-pairs rather than the phonetic relationship (even “Touch-Tickle” was overwhelmingly categorised as semantically related (see figure 6 and/ or appendix 5). Nevertheless, it also shows that both sexes score higher in the acoustic version of the test. In this part of the test, the hypothesis that women would score higher as they can categorise better than men, has turned out not to be true as men scored higher, even if the results were below 45%. Averaging all male and female scores from figure 5, the overall result for women is 74.2% and for men also 74.2%. This suggests that all participants show perfectly equal results.

Figure 5. average correct answers in percentages (%)

Source: appendix

Figure 6 presents an overall view of the results in time per second (full discussion can be found in Chapter 4), namely: Men respond faster in all categories (as provided in chapter 2.4), therefore it is of importance to look at gender separately. Reaction time performed by men concludes: they respond fastest to semantically related words 5.85 sec., 5.91 sec. to the phonetically related words, 5.92 sec. to unrelated words, and 6.09 sec. to S+P related words. Reaction time performed by women concludes: they respond fastest to unrelated words 6.12 sec., 6.15 sec. to semantically related words, 6.16 sec. to phonetically related words, and 6.32 sec. to S+P related words.

Figure 6. average time in seconds

Source: appendix

Figure 7. Difference in the visual and acoustic part of the both semantically and phonetically answered words

Source: appendix

Figure 7. Image taken from the visual test

Source: LimeSurvey

Discussion and Development

The phonetic results complement the hypothesis that participants would score higher when they hear the words as that is a natural behaviour in contrast to reading. What also should be highlighted in this part is Hoey’s theory on “semantic association” as six out of the 31 visual, and seven out of seventeen acoustic participants chose both options (S+P) when they heard the word-pair “Ocean-Motion”. The unrelated word pairs have shown a controversy in theory provided in chapter 2 that individuals respond faster to related words than unrelated words. Women, however, have shown that they respond faster to unrelated words than any other category. It could be because of the click option was closer to the ‘next’ button (see figure 7). If that were the case then men would have shown a similar result. By contrast (in figure 6), men responded faster to the semantically related words, the furthest option from the “next” button. However, it could also be considerable that the focus was set on three different possibilities and it could have been that women started detecting unrelated words easier as they were the “unimportant” category. Ingalhalikar showed in her research that women do make other connections and think of other options and possibilities. Figure 6 shows that more time was needed for the semantically and phonetically related words, all participants needed extra time (women 0.20 sec. extra, men 0.24 sec extra compared to their fastest time). That could relate to: Participants had to press two options and therefore more time was needed An unconscious connection was made by the brain. It recognised both options, and therefore needed more time (as it can be seen in appendix 5 all answers are in the order in which it was presented in the test. Participants seem to score better towards the end, which could suggest that the brain started to understand and it became an automatic response in recognition).

5 Conclusion

To summarise, the findings indicate that further research and thought is needed to have more accurate results. Firstly, the reason I wanted the participants to have the acoustic part presented a week later was the worry that they might memorise the words. After writing the essay I have realised that individual semantic association will not interfere with the results, even if the participant could memorise the words. After all, this suggests it stays the same in the lexicon. It is of importance that the experiment takes place on the same day, since fourteen participants were not able to be present the following week. This will have a huge impact on the results when the number of participants is already low. Secondly, in order to produce a more accurate reaction time result and to exclude the onscreen position of the answers as a timing factor, the survey could be identically recreated whilst only changing the answer positions from vertical to horizontal. The computer and internet connection is also highly relevant and could cause inaccuracies. Therefore, ensure that all participants are using identical devices and a shared internet connection. Finally, researching hypothesis one: Due to the natural language processing approach discussed in chapter two, it was expected that the accuracy of the acoustic results would be higher than the results for the visual-only data. The prediction was upheld by the phonetically, semantically and phonetically, and unrelated words. The semantically related words contradicted this theory, suggesting that the design of the acoustic survey had some defects. Four of the semantically related words were poorly pronounced in the test.

Hypothesis two: it was expected that both sexes would be more lenient towards the semantically related words, which they were as predicted. However, all participants were expected to score at least a minimum of 55% in the test, which they both failed to do. Hypothesis three: Female participants were expected to score higher when categorising semantically and phonetically related words. Overall, the findings clearly indicate equality in percentages, and that one is better in one area than the other and vice versa. The test should be reconsidered in individuality rather than sex. All chapters considered, it can be said that the nature of lexical processing is not gender related but it is an individual phenomenon which should be further explored. In the preceding chapters I have provided ideas and a framework with which to do so.

References

Altmann, G. T. M. (1997) The Ascent of Babel. Oxford: Oxford University Press. Aitchison, J. (2012) Words in the Mind. Hoboken: John Wiley & Sons Press. Harley, T. (2014) The Psychology of Language. Hove, East Sussex, UK: Erlbaum (UK) Taylor & Francis Press. Hoey, M. (2005) Lexical Priming. London: Routledge Press. Ingalhalikar, M. et al. (2013) "Sex Differences In The Structural Connectome Of The Human Brain". Proceedings of the National Academy of Sciences 111.2: 823-828. Jarema, G. and Libben, J. (2007) Phonological and Phonetic Considerations 4of Lexical Processing. Joel, Daphna et al. (2014) "Queering Gender: Studying Gender Identity in ‘Normative’ Individuals". Psychology & Sexuality 5.4: 291-321. McWhorter, John H. (2011) What Language Is. New York: Gotham Books Press.

← Back to Papers