Introduction My research study is concerned with lexical semantics and the association approach to meaning- the idea that the brain connects associated concepts to create lexical schemas for particular words we are exposed to. The meanings of words can be described through their dictionary definitions; however, this is a rather linear approach to meaning as dictionary definitions fail to acknowledge the broader spectrum of ideas that build the semantic understanding of a word (Traxler, 2011, p.83). Semantic associations are described as connections formed between words “based on meaning and co-occurrence within a semantic field” (Brooks & Kempe, 2012, p.74). The association approach recognises these clusters of connected concepts existent in our mental lexicon and understands word meanings as “whatever comes to mind when someone says the word” (Traxler, 2011, p.83).
Previous research into word association has looked at cross-cultural variations, for example, differences between the experience of depression has been studied using word association tests with Japanese-national, Japanese-American and Caucasian-American college students, investigating associations to the word equivalents of depression (Tanaka-Matsumi & Marsella, 1976). Word Association tests have also been used to study the developing mental lexicon of second language learners. Post found context to be the strongest influence over word associations, and that for both L1 and L2, the mental lexicon is highly complex containing “overlapping features” (Post, 2007). Further studies have employed word association tests for clinical examinations, helping to reveal “problematic complexes” in patients (Carducci, 2016). Word association was used specifically to identify words that signaled an anxiety response in patients, causing them to exhibit nervous behaviour (Carducci, 2016). The effects of both word imageability and frequency in word associations have also been tested. Results of discrete and continued association task experiments demonstrated that word imageability strongly affects association responses, however, word frequency was found to hardly affect association (de Groot, 1989).
Considering these previous studies, in particular the latter carried out by de Groot, my research seeks to challenge the conclusion that word frequency has little effect on word association responses. Brooks and Kemp discuss the activation of associated concepts, explaining how words heard, read or spoken automatically signal their associate words, increasing the potential for these associated concepts to become conscious thoughts (Brooks & Kempe, 2012, p.74). Therefore, my research questions whether words we hear and use more frequently have a greater number of associated concepts connected with them in the brain compared to words we hear and use less frequently. This theory will be tested by analysing the performance of participants in a word association task. I hypothesise that higher frequency words will prompt a greater number of associated concepts than lower frequency words.
Methodology The dependent variable in the study is the number of associated words participants provide for each stimulus word. The primary independent variable is the student/recent student participant characteristic. Participation from twenty university students (including recent graduates within six months post-graduation) was obtained using a judgement sampling method. Certain people representative of the student population were asked to take part, as well as recruitment through a social media post advertisement to students following the researcher’s social media account. The exact participant gender ratio is unknown, as well as the exact age of participants (although age is likely between 18-25 years by estimated general age of participants). The secondary independent variable manipulated in the research is the level of frequency of the stimulus words (either high or low) used in the test. A factorial, within-subjects design was used- each participant responded to each condition of the secondary independent variable.
Ethics Considering ethical issues, I created an information sheet and consent form that clearly outlines what it is the participants are invited to do and exactly what the task entails, asks permission for their responses to be analysed and used in the research report for university coursework, explains their right to withdraw from the study at any stage, highlights any potential risks from participation (low chance of any form of harm), states how long the task will take to complete and who will have access to the responses (myself and the project supervisor), what they are consenting to and how to demonstrate their consent. I made sure to withhold specifics that might have revealed the true aim of the research and influenced the participants’ responses, therefore affecting the validity of the research findings. I created a debriefing section to appear after completion of the task which informed participants of exactly what the research set out to investigate. I asked the participants to provide a nickname to ensure I would be able to identify their responses in case of deciding to withdraw their participation. However, I have ensured anonymity in the documented results by removing the nicknames and randomly alphabetising the participants. I created a task that would not elicit any seeable potential threat to participants’ emotional wellbeing- the words used in the task were general and no extra context was provided that could evoke sensitive topics. There was no concern for physical harm as the test was distributed online, allowing participants to complete it their own space. The potential issue of researcher safety does not apply for the same reason- I did not come into any physical or face-to-face contact with participants, therefore causing no potential threat to my own safety. I gained ethical approval from two project supervisors before carrying out this research.
Materials, Measures and Procedure The research was carried out and data collected through an online form. The total amount of time the form would take each participant to complete was estimated at approximately 15 minutes. Six stimulus words deemed as high-frequency, as typically related to the life of students, were chosen- these words would theoretically produce stronger responses due to stronger lexical schemas for these words and, therefore, a greater number of connections to these concepts able to be recalled. Six lower frequency words were chosen that were deemed to be generally unrelated to the everyday life of students, which would theoretically produce weaker responses for associations. These words were provided in a random order (to avoid awareness of categories and potential guessing of the research question) as separate headings, one below the other, with a text box space below each one for participants to type their responses. Participants were asked to give themselves approximately 60 seconds to respond to each word provided. They were asked to type as many words as they could think of associated with the different stimulus words and encouraged not to overthink, relying on their automatic, natural responses.
Data Responses were collated in a spreadsheet and each individual response was given a numerical value equivalent of the number of concepts provided for each stimulus word. These quantitative values were then added up and divided by the number of participants to find the average number of words/concepts recalled for each stimulus word. The total average for both categories was also calculated and the values inputted into tables and displayed in a bar chart. Word clouds were created and included as a visualisation of the collective schemas for one word of each category.
Results
The results of the low-frequency category show a total of 685 associated words, giving a total average of 34.25 words. The results of the high-frequency category show a total of 764 associated words, giving a total average of 38.2 words. The high-frequency word category scored a higher average number of associated words than the low-frequency word category.The results also indicate that 75% of participants scored highest in the high-frequency word category, with 25% scoring highest in the low-frequency word category.
The lowest scoring word in the low-frequency category (and the lowest scoring word overall) was enterprise with an average of 4.3 words, where the highest scoring word was surfing with an average of 6.75 words. The lowest scoring word in the high-frequency category was seminar with an average of 5.5 associated words. The highest scoring word in the high-frequency category, and the highest scoring word overall, was university with an average of 7.45 words.
Discussion From first glance at these findings, it appears that the original hypothesis can be accepted. The average number of associations provided for words in the high-frequency category is greater than the average for the low-frequency word category. However, the difference between the average number of associated words for the low-frequency category and the high-frequency category was 3.95 words, which is a rather low number. Similarly the difference between the highest scoring high-frequency word and the lowest scoring low-frequency word was still only 3.15 words. Furthermore, two low-frequency words scored a higher average than the lowest scoring high-frequency word, which demonstrates that not every high-frequency word produced more associations than the low-frequency words did. This seems to support de Groot’s conclusion that word frequency has hardly any affect association (de Groot, 1989).
Considering aspects of the methodology of this study, e.g., the sampling method, the method of selecting the low- and high- frequency stimulus words and the general nature of the experiment design, there are a few potential reasons for the notable differences in individual participant performances. Firstly, background information about the participants was unknown, therefore it could be that some of the lower frequency words selected for the test are actually words participants are exposed to more often than expected, having based the choice of stimulus words solely on their student identity. Therefore, the greater number of concepts they recalled for words in the ‘low-frequency’ category may reflect an agreement with the idea that words we are exposed to more frequently evoke a greater number of associated words. This might also support Post’s previous research that found context to be the strongest influence over word associations, with no context having been provided for these stimulus words. In future, research into the participants’ backgrounds and selecting words based around their individual contexts might provide more accurate results.
Furthermore, the experiment being conducted online without the presence of the researcher, meaning an inability for the researcher to reiterate the instructions for the task and be available for questions related to what is being asked of the participants, is arguably a weakness to the design. It was evident from the results that a couple of participants responded to each word with only one associated word, and it also appeared that a couple of participants had tried to keep the number of associated words equal for each stimulus word. If this was the case and participants did not respond exactly how they were asked to, this pattern would not reflect an accurate response to the task and would therefore question the validity of the research. This might have been a result of the instructions not being perfectly clear; however, the majority of participants seemed to respond in the expected way. Possibly the fact that participants were not aware of the true aim of the research meant that they did not realise the importance of the number of associations, focusing instead on the quality of the word they provided.
Based on these findings, a definite conclusion as to whether frequency has an affect on associations cannot be made. Therefore, the original hypothesis that higher frequency words will prompt a greater number of associated concepts than lower frequency words cannot confidently be accepted.
References:
Brooks, P. and Kempe, V., 2012. Language Development. Chicester: John Wiley & Sons, Inc., p.74.
Carducci, B., 2016. ‘Jungian And Adlerian Therapy’ [online] Neuroscience and Biobehavioral Psychology Encyclopedia of Mental Health. Indiana University: Elsevier. Available at: https://www.sciencedirect.com/topics/nursing-and-health-professions/controlled-oral-word-association-test [Accessed 14 January 2021].
de Groot, A. M., 1989. ‘Representational Aspects of Word Imageability and Word Frequency as Assessed through Word Association’. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15(5), pp.824–845. Available at: https://doi.org/10.1037/0278-7393.15.5.824 [Accessed 14 January 2021]
Post, M., 2007. Word Association Responses, Lexical Development and The Relationship Within the Mental Lexicon of Second Language Learners. [online] Birmingham University. Available at: https://www.birmingham.ac.uk/Documents/college-artslaw/cels/essays/lexis/MPostLexisRelationshipsWithintheMentalLexicon.pdf [Accessed 14 January 2021]
Tanaka-Matsumi, J. and Marsella, A. J. 1976. ‘Cross-Cultural Variations in the Phenomenological Experience of Depression: I. Word Association Studies’, Journal of Cross-Cultural Psychology, 7(4), pp. 379–396. Available at: https://journals.sagepub.com/doi/abs/10.1177/002202217674001#articleCitationDownloadContainer [Accessed 14 January 2021]
Traxler, M. J., 2011. Introduction to Psycholinguistics: Understanding Language Science. John Wiley & Sons, Inc., p.83.
Commentaires