Investigating Language on Twitter - A2 Mini Investigation

Investigating Language on Twitter - A2 Mini Investigation

Introduction

I have chosen to focus on how gender affects language as my specialised topic area in this investigation, and in order to study this I put together a hypothesis. My hypothesis for this investigation is 'I think that men are more likely to show dominance than women in their twitter posts.' I have chosen to focus on particular feature areas such as how men and women use language features to show male dominance and female language deficiency in their tweets. I am going to focus on imperatives, multi-modal tweets, uncertainty features, empty adjectives and abbreviations and quantify them to see which sex used which features more. I would also focus on how the male and female participant used co-operative or competitive language however this would be harder to evaluate as they are not communicating to a particular speaker in a conversation thread. I also wanted to see how this would link to Robin Lakoff's deficit model and how the amount of empty adjectives and uncertainty features would show how women's language is deficient and this could link to how male speakers tend to be more dominant.

Methodology

When constructing this investigation I decided to pick two celebrities and systematically select 5 tweets from each. We did this by choosing every fifth tweet on their feed and chose celebrities so it was unbiased rather than using someone we knew personally and this could affect how we interpret the results. We also picked random celebrities so it was not biased and opinions could not affect what we analysed from the data. One benefit of using data from twitter is that we don't have to ask for consent as the celebrities who have posted these tweets are essentially allowing their information to be used as they have posted it publicly. However one limitation with our data that we collected is that it is only a very small pool of data and therefore we are unable to generalise this data to males and females in general as it is not fully representative.

Analysis

We found that Taylor (female) showed a higher amount of empty adjectives and uncertainty features than Zac (male). However Zac used a higher amount of multi-modal tweets (pictures and words), but also imperatives and abbreviations. From these results we could suggest that men are much less descriptive and more concise with their language in their tweets whereas women are much more expressive with words and describe something in more detail. This is proved as 4 out of 5 of Zacs tweets were multi-modal which incorporated pictures and words. We could also suggest that men are more dominant due to their higher amount of imperatives. However, one of Zac's tweets was an anomaly because due to my pragmatic understanding the tweet was song lyrics, however it uses the uncertainty feature 'like', but I don't know whether to consider it as his language use as he didn't put it in quote marks. We also considered the average number of features per tweet and didn't count 'retweets' as they were not written by the celebrity themselves and therefore cannot evaluate their language use. Taylor uses 'SO MUCH' in one of her tweets and this was a dilemma because both of these words are individual uncertainty features, and we can't decide to count them as one uncertainrt feature or two, as they are used in a phrase and work together.

Conclusion and Evaluation

There are massive limitations in this investigation as the data pool is very small and limiting as tghere was only 10 tweets. However, the data proves hypothesis as female uses more uncertainty features and empty adjectives however the male uses more multi-modal tweets, imperatives and abbreviations, which could prove male dominance in language.