Our study suggests that text does not have the hypothesized disambiguation value for emoji. In this section, we discuss the implications of this finding more broadly. An important question is why doesn’t text reduce emoji ambiguity? One reasonable hypothesis is that sarcasm plays a role. Our survey contained an open-ended text box to gather feedback from participants, and several participants highlighted the role of sarcasm in their assessments: “some of the emojis seemed sarcastic” “Wasn't sure how to analyze the sarcastic texts” Another insight as to why emoji were still ambiguous in context that was pointed out by a participant was that the texts containing the emoji were too short: “A couple of the texts could use a little extra context to tell what the emoji is supposed to reflect. For instance, the "I didn't expect to see her unexpectedly" text could be either positive or negative based on context.” With Twitter’s 140 character length restriction, using tweets as our source of texts limited the amount of context accompanying emoji in our study, whereas many platforms for emoji usage are not limiting in that respect. Similarly, while using Twitter as we did (e.g., with the filtering steps outlined above) allowed us to maximize general interpretability and successfully examine general consistency of interpretation (as reflected in broadcast communication like Twitter), this approach limited the amount of interpersonal context (or common ground (Clark 1996)) in the simulated communication. Future work should seek to explore emoji ambiguity in longer-form texts and in longitudinal communication in more established relationships. Interestingly, while our study controls for the presence or absence of text to study emoji ambiguity, the reverse relationship is also worthy of examination. In other words, future work should seek to investigate whether emoji affect the ambiguity of the text they accompany. Participants reflecting in the open-text box suggested that this could be the case. For example, one participant wrote: “[emoji] do have their value in that they give you a sense of security that you've gotten across the right tone in an email. Whenever I feel I need to be most clear rather than risk a misunderstanding, I insert an emoji” This sentiment was reflected in some qualitative responses in Cramer et al.’s (2016) recent work on emoji as well. Lastly, it is interesting to reflect on textual context’s effectiveness in reducing the ambiguity of Apple’s (former) rendering of the “grinning face with smiling eyes” character (U+1F601). Miller et al. identified a roughly bimodal distribution for sentiment interpretations for this rendering. Our results suggest that in these types of extreme ambiguity cases in which there are two clear senses that must be disambiguated, text may possibly help to distinguish between the two very different meanings. Examining this conjecture in detail would be a useful direction of future work.