Haba! Bilingual interjections in Nigerian English: A corpus-based study

Interjections and emojis in Nigerian online communication

“…The analysis relied on an 843‐million‐token corpus of written web forum interactions between young Nigerians, and automatic detection and measurement of Nigerian Pidgin and positive and negative emotions in the forum writing. Our multinomial regression model suggests that the probability of Nigerian interjections is higher in messages containing NigP (expectedly), orthographically lengthened words (suggesting informality), and positive rather than negative emotions (somewhat surprisingly, considering Unuabonah & Daniel, 2020), while the shocked emoji occurs in strongly negative contexts, in messages by established forum members, and alongside other emojis – not only angry and displeased but also amused or flirtatious ones. The sad emoji, however, favors English‐language interjections, as do high arousal (estimated through the use of capitalization and punctuation) and a high density of emotionally laden words.…”

“…Positive messages are also associated with Nigerian interjections when arousal is low to medium (probability 0.5–0.7 for arousal < 0.5). After Unuabonah and Daniel (2020) described kai , chai , and haba as strongly negative, we expected Nigerian interjections to favor negative contexts. This turned out not to be the case, but instead, each of the six Nigerian items occurs in messages expressing positive surprised emotions as well, as in examples (9–11): (9)Kai, never seen a Brother that know more about BBQ like you, damn… I am so much loving this… [laughing emoji] (10)haba!…”

“…The positive and negative emotions across a message are combined into an overall sentiment score; the process is described in Section 3.4. In addition to surprise, Unuabonah and Daniel find the ethnic interjections haba , kai , chai , and chei to ‘express strong negative emotions’ (2020, pp. 70, 75); the same applies to NigP na wa (Unuabonah, 2020, p. 158).…”

“…69-71). Studies on interjections have tended to be qualitative and focus on functions; this holds true even for research based on large CMC data sets which would permit quantification, such as Unuabonah and Daniel (2020), or Lockyer's (2014) Twitter study. Our study departs from this tradition, striving to detect preference patterns based on a data set much larger than what could be investigated qualitatively.…”

“…This study contributes to the latter emerging tradition. Previous work on discourse phenomena in NigE has compared the frequencies and distribution of discourse markers across contexts to British English (Gut & Unuabonah, 2019; Unuabonah, 2019; Unuabonah & Gut, 2018), described the extended functions of English elements (Adegbija & Bello, 2001; Fuchs et al., 2013; Ogoanah, 2011), and explored the usage of indigenous discourse‐pragmatic elements (Honkanen, forthcoming; Unuabonah & Daniel, 2020; Unuabonah & Oladipupo, 2018). Most studies are based on the Nigerian component of the International Corpus of English (Greenbaum, 1996; Gut, 2012), which allows comparisons across various spoken and written genres, and across varieties.…”