Microaggressions are subtle manifestations of bias (Breitfeller et al., 2019). These demonstrations of bias can often be classified as a subset of abusive language. However, not as much focus has been placed on the recognition of these instances. As a result, limited data is available on the topic, and only in English. Being able to detect microaggressions without the need for labeled data would be advantageous since it would allow content moderation also for languages lacking annotated data. In this study, we introduce an unsupervised method to detect microaggressions in natural language expressions. The algorithm relies on pre-trained wordembeddings, leveraging the bias encoded in the model in order to detect microaggressions in unseen textual instances. We test the method on a dataset of racial and gender-based microaggressions, reporting promising results. We further run the algorithm on out-of-domain unseen data with the purpose of bootstrapping corpora of microaggressions “in the wild”, and discuss the benefits and drawbacks of our proposed method.
The growth of Social Media platforms has been accompanied by an increased visibility of expressions of socially unacceptable language online. In a 2016 Eurobarometer survey, 75% of people who follow or participate in online discussions have witnessed or experienced abuse or hate speech. With this umbrella term, different phenomena can
Copyright © 2021 for this paper by its authors. Use
permitted under Creative Commons License Attribution 4.0
International (CC BY 4.0).
be identified ranging from offensive language to
more complex and dangerous ones, such as hate
speech or doxing. Recently, there has been a
growing interest by the Natural Language Processing
community in the development of language
resources and systems to counteract socially
unacceptable language online. Most previous work has
focused on few, easy to model phenomena,
ignoring more subtle and complex ones, such as
microaggressions
Microaggressions are brief, everyday
exchanges that denigrate stigmatised and culturally
marginalised groups
Previous work on microaggressions with computational methods is quite recent. Breitfeller et al. (2019) is one of the first work to address microaggressions in a systematic way, also introducing a first dataset, SelfMA. A further contribution specifically focused on racial microaggression is Ali et al. (2020), where the authors focus on the development of machine learning systems.
In this study we introduce an unsupervised method for microaggression detection. Our method utilizes the existing bias in wordembeddings to detect words with biased connotations in the message. Although unsupervised approaches tend to be less competitive than their supervised counterparts, our method is languageindependent and thus it can be applied to any language for which embedding representations exist. Furthermore, the reliance of our methods on specific lexical items and their context of occurrence makes transparent the flagging of a message as an instance of a microaggression. In addition to the usefulness of our method in languages with no labeled data, the reliance of our model on words in the sentences would make it interpretable as it allow human moderators to understand what the system has based its decision on.
Our contributions can be summarised as follows: • we introduce a new unsupervised method for the detection of microaggressions which builds on top of pre-trained word embeddings; • we compare the performance of our model using different pre-trained word embeddings (Glove, FastText, and Word2Vec) and discuss the potential reasons behind the differences; • we test the proposed algorithm on unseen data from a different domain (i.e., Twitter), in order to qualitatively evaluate its efficacy in discovering new instances of microaggression.
The rest of this paper is structured as follows: we introduce our method in Section 2. The data and our results are reported in Section 3. We deploy our model and discuss its limitations in Section 4. Finally, we present the conclusion and future work in Section 5. 2
Embedded representations, either from pre-trained
word embeddings or pre-trained language models,
have been shown to contain and amplify the biases
present in the data used to generate them
We devise a simple and effective method that exploits existing bias in word embeddings and identify words in a message that are related to particular and distant semantic areas in the embedding space. Messages are analysed in three steps: first, for each token ti we compute its relatedness to a list of manually curated seed words s = s1, ..., sn denoting potential targets of microaggressions; second, we consider only the similarities of the pairs (ti, sj ) above an empirical similarity threshold ST and compute their variance vi; finally, we classify the token ti as a micro aggression trigger, and consequently the message as a micro aggression, if the vi is above an empirically determined variance threshold V T .
The intuitive idea behind this algorithm is that some lexical elements in a verbal microaggression are often (yet sometimes subtly) hinting at specific features of the recipient of the message, in an otherwise neutral lexical context.
In this work, we choose to focus on microaggressions related to race and gender, therefore the seed words have to be chosen accordingly. The seed word lists for race and gender are, respectively, [white, black, asian, latino, hispanic, arab, african, caucasian] and [girl, boy, man, woman, male, female] for gender. There is also a practical reasons to focus on gender and race, namely the scarcity of data available for other categories of microaggression and other idiosincrasies of the available datasets — the religion class was specific to different religions, therefore hard to generalise, sexuality and gender presented a large overlap, and so on.
An example of how the proposed method works is illustrated in Figure 1. In the example, consider the word ”chopsticks” in the message ”Ford: Built With Tools, Not With Chopsticks” (from the SelfMA dataset, described in Section 3). The target word exhibits a much higher relatedness to the word asian (0.237) than any other seed words. Even just considering the seed words with a similarity above a fixed threshold ( white, asian and, african), the variance of their similarity score with respect to chopsticks is still higher than the variance threshold, and therefore this target word, in this context, triggers a microaggression according to the algorithm. This process is repeated for all the words in the message in order to detect microaggressions. Some categories of words are bound to exhibit a high relatedness to all the seed words, e.g., “people” or “human”. This is the reason to introduce the variance threshold in the final step of our algorithm, to filter out these cases when classifying a given message, and instead focus on words that are related to different races (or genders) unevenly, with a skewed distribution of similarity scores.
An important by-product of this algorithm is that the output is one or more trigger words, in addition to the microaggression label — in the example, the trigger word is indeed chopsticks — therefore enabling a more informative and interpretable decision process.
SelfMA Gender SelfMA Racial Tumblr
1,314
1,278
2,021
To test our method, we use two subsets of the
SelfMA: microaggressions.com dataset
It is important to note that a microaggression can have multiple tags, so there is an overlap of 1Tumblr is a popular American microblogging platform https://www.tumblr.com instances. However, the seed words used to detect microaggression types in the method are different for each target phenomenon (e.g., race, gender).
We ran the algorithm on the SelfMA dataset,
empirically optimising the two thresholds on the
training split, for each word embedding type and
each microaggression category, filtering by the
seed words listed in Section 2. We test the
algorithm with three pre-trained word embedding
models for English, namely FastText
The results, shown in Table 2, indicate that FastText has a better F1 score on Racial microaggressions while word2vec performs better on Gender microaggressions. The difference in performance between FastText and word2vec is not major, and we attribute this to the difference between the corpora on which the two models were trained (i.e., web crawl and Wikipedia for FastText vs. news data for word2vec). The GloVe pretrained model, trained on a combination of newswire texts, encyclopedic entries and texts from the Web, underperforms in both experiments. In general, the absolute figures are encouraging, especially considering the simplicity of this unsupervised approach. 4
To better understand the performance of our unsupervised model, we performed an additional experiment. Our goal is to understand the false positive results and the potential harm the model could cause. To do so, we use our unsupervised model to label unseen instances from another domain (Twitter) than the SelfMA dataset (Tumblr) in order to see how the model would perform in detecting microaggressions.
We begin by performing keyword searches on Twitter (using Twitter’s official API) and collect a new dataset of of 3M tweets with seven keywords potentially containing race and gender expressions.Next, we set the threshold values ST and V T in our model in order to obtain the highest Precision scores, rather than the highest F1 value. This step is performed exactly like the optimization described in Section 2 with the only difference of the target metric. The aim of this step is to only label tweets as microaggressions with the highest possible degree of confidence. We set ST = 0.12 and V T = 0.014 for racial microaggressions leading to Precision of .931 and ST = 0.13 and V T = 0.019 for gender-based microaggressions leading to a Precision of .912. Precision has been measured on the original SelfMA dataset used as a validation set.
We then run the unsupervised model on the new Twitter dataset by automatically labelling 256,843 tweets for gender and 373,631 tweets for race. After the data is labeled, we manually explore the positive instances in order to evaluate the performance of the model. The algorithm tuned for high precision found in this dataset 6,306 genderrelated microaggression candidates, 13,004 racerelated microaggression candidates.
We find that while the model does detect actual instances of microaggression, there is a noticeable amount of false positive instances. These tweets discuss race or gender in some manner. However, they do not necessarily contain microaggressions towards these groups. While the model does learn to detect discussions of these topics, it seems to sometimes confuse these discussions with microaggressions towards the aforementioned groups. Some examples follow, paraphrased to avoid tracking the original messages.
Saying ”Arrested Development isn’t funny” in an office full of women just to feel something “Men have moustaches, women have oversized bracelets”
The humorous attempts in this tweets hinge on gender stereotypes, and therefore in some contexts it could be perceived as offensive by some recipients. The high relatedness in the word embedding space between some words (moustaches and bracelets) and gender-related seed words (men and women) triggers the detection algorithm.
The automatic detection of racial microaggressions “in the wild” is more challenging than gender-based ones, according to our manual exploration of this automatically labeled dataset. This may be due to the difficulty of crafting a list of seed words that is sufficiently race-related, but at the same time avoids generating too many false positives. We indeed found many of them,
GloVe FastText
GloVe word2vec FastText word2vec
not-MA MA macro avg. not-MA MA macro avg. not-MA MA macro avg. not-MA MA macro avg. not-MA MA macro avg. not-MA MA macro avg. .692 .603 mainly due to named entities and multi-word expressions such as “White House”, or simply because of the polysemy of color words, e.g. “black” and “white”. We, however, still found instances of messages containing different extent of racial stereotyping.
“why are you being so dramatic? just say I’m not originally arab, you don’t have to fight about it” “I will need to explain that to the chinese old lady who works at my school’s administrative office”
In summary, running the unsupervised microaggression detection algorithm on unseen data seems to represent a promising intermediate step towards the semi-automatic creation of language resources for this phenomenon. While the accuracy is not ideal, and lists of seed words have to be handcrafted carefully in order to avoid false positives, these drawbacks are balanced by the fairly cheap computational cost and the ease of application in a multilingual scenario. 5
In this paper we introduce a novel algorithm that exploits the existing bias in pre-trained word embeddings to detect subtly abusive language phenomena such as microagressions. While supervised methods of detection in the field of natural language processing are plentiful, these methods are only viable for languages and topics with available labeled datasets. That is however not the case for many languages. As a result, the unsupervised method of detection introduced in this study could help address the need for the moderation of microaggressions in languages other than English. This is further helped by the availability of multilingual word-embeddings as they would allow the method to be used in any of the languages supported by the embedding.
The method is unsupervised and only needs a small list of seed words. Considering its simplicity, the results obtained from an experiment on a dataset of manually annotated microaggressions are very promising. Further, the method is transparent, explicitly identifying the words triggering a microaggression, and thus paving the way for explainable microaggression detection.
Although the preliminary results are promising, an experiment on unseen data from a different domain shows that there is leeway for improvement. Given that we are looking at the explicit words used in each message, our method is not sensitive to implicit expressions like “you people” or “your kind”, often occurring in microaggressions. We would have to add further steps to our algorithm to catch expressions like these.
Polysemy is another known issue, e.g., in words
like “black” and “white” whose relatedness to
certain identified trigger words could not necessarily
be due to race. While a careful composition of
the seed word lists helps to minimize this issue, a
systematic approach to polysemy would certainly
be desirable. The seed word list may also be
expanded, either manually or exploiting existing
lexicons such as HurtLex
In future work, we plan on improving our model
to account for lexical ambiguity, and the
complexity derived from the interference between
pragmatic phenomena and aggression, e.g., in
humorous and ironic messages, following the intuition
in recent literature
This work of Valerio Basile is partially funded by the project “Be Positive!” (under the 2019 “Google.org Impact Challenge on Safety” call). 2See for instance this compendium of LGBTQIA+ terminology: https://www.umass.edu/stonewall/si tes/default/files/documents/allyship ter m handout.pdf