‘Eish’ in South African English: an analysis of the word in informal written public discourse, with a speculative enquiry into its etymology

The lexical item ‘eish,’ with a speculative enquiry into its etymology | Corpus linguistics and corpora | High-frequency SAE in the corpus of online comments | ‘Eish’ in informal written public discourse | The EISH-YOU collocation | Considering the EISH-YOU collocation in online comments as informal written public discourse | Conclusion

In what follows, I consider the lexical item “eish” as it is used in South African English (SAE), and speculate on its possible etymology, before using corpus linguistics to analyse the interjection’s collocation with “you” as it occurs in informal written public discourse.

The lexical item ‘eish,’ with a speculative enquiry into its etymology

“Eish” is a common South African expression1. It is an emphatic exclamation of surprise, disappointment, frustration, annoyance, disbelief, resignation, approbation, or less frequently, agreement or approval. The etymology of “eish” is not yet fully understood (see below), but the word has been used in English in South Africa since at least the 1990s (and, before that, in other South African languages)2:

1997 You might hear that so-and-so has gone (in an accident) and you think: “Eish! Our brothers are passing away.” (Informant in Campbell, 1997: 276)

2001 “Eish!” exclaimed Sobahle with vigour, “coal stoves make too much smoke in the house. And that makes a lot of work!” (Meintjes, 2001: 354-355)

“Eish” has already entered a number of dictionaries. Most significantly, it has an entry in the second edition of the Oxford South African Concise Dictionary (2010), and will have one in the (future) online-only second edition of A Dictionary of South African English on Historical Principles (the online version of the 1996 first edition can be found here). The online international English dictionaries where it is listed include Oxford Dictionaries (English), Collins English Dictionary, dictionary.com, Wiktionary, the Online Slang Dictionary (OSD), and Urban Dictionary.

While OSD doesn’t give any information on the derivation of the word, in some of the (far-fetched and, frankly, wack) entries in Urban Dictionary, it is claimed it is from Xhosa. According to Wiktionary, it comes from Afrikaans. These are all wiki-type dictionaries, and each derivation claimed appears to be a shot in the dark. Collins and dictionary.com claim that the word comes from Zulu, which may be closer to the truth, but not the whole story. Oxford Dictionaries, on the other hand, says that is from Tsotsitaal. This, I think, is probably the best answer in the online dictionaries – the spelling of the word suggests that the orthography may have undergone a shift from a vernacular language (e.g. Zulu or Xhosa) to a patois (e.g. Tsotsitaal or Isicamtho3): “ei” does not occur as a unit in the Nguni languages, though it can in Tsotsitaal and Isicamtho (e.g. “heita”). Along with Afrikaans, Zulu has been a prime influencer of Tsotsitaal4. None of these entries, however, provide any evidence for their contentions, nor any information on how “eish” may have been derived from any of these languages, and what it might have meant originally.

The origin of “eish” is an etymological puzzle calling out if not for resolution, then at least proper research. Speculating on its roots, I trace out the following:

1. It’s possible that “eish” is derived from the Xhosa interjection “ishi” (or “ish”)5. “Ishi” is historically an interjection of surprise and prohibition, with the meaning “Go away! You tire me out!”

2. However, given the fact that Zulu is more prevalent than Xhosa, particularly in Gauteng, where Tsotsitaal and Isicamtho also originate (a matter I’ll return to in the next point), it’s more likely that “eish” is derived from one or more of a number of related Zulu interjections6:

a. “ashi,” an interjection of “very strong disapproval or negation”
b. “ishi,” an interjection of approval (“Just so”)
c. “yeshi,” an interjection of “surprise, admiration” (“Oh my! Just fancy!”)
d. “yishi,” an interjection of approval (“Just so”)

3. It is noticeable that these similar interjections all contain some seed of the senses in which “eish” is used in South Africa today. Taking the speculative line a step further, it might not be going too far to suggest that this multiplicity of deep Zulu interjections was somehow collated, simplified and “flattened out” in an uptake of one or more of them into Tsotsitaal or Isicamtho, so that their particular individual meanings were crowded together under the rubric of the single lexeme “eish,” and as a result, became intertwined in this one word.

4. It may also be the case that an interjection like “ishi,” which – if this line of reasoning is on track – must (at least phonetically) be one the frontrunners to be considered a precursor to “eish”; and which, according to the sources used here, has almost opposing meanings in Xhosa and Zulu, experienced a similar convergence of the senses pertaining in those two languages, so that this opposition was flattened and simplified in one new word. These factors would explain why “eish” is meaningful in so many different contexts, why it can index such a variety of emotions (frustration, disappointment, annoyance, etc.), and why, in separate uses, it can yoke together opposing meanings (e.g. disbelief and agreement, approval and approbation).

5. In summary, what I am suggesting is that one or more of these interjections dropped its final “-i” in the loanword-saturated informal speech of Zulu- and/or Xhosa-derived Tsotsitaal or Isicamtho. Most, if not all, loanwords in these languages are in noun class 5, where they get an “i-” prefix in their singular form. Dropping the final “-i” of the interjection would particularly be the case if the prefix of the following word was “i-”7. Thus, it is quite possible to have constructions like “Ash’, ipolisa!” or “Ish’, ipolisa!” – “Hey, (a) police(-man/-woman)!”8.

Corpus linguistics and corpora

Corpus linguistics is a sub-discipline within linguistics that involves analysing of large amounts of textual (and, increasingly, these days, sound and video) data. It works on a linguistic corpus (plural: corpora) – a collection of text files – using a type of analytic software usually referred to as a concordancer. Two popular concordancers are WordSmith and Laurence Anthony’s AntConc. WordSmith is fairly expensive, but AntConc, which has all you need in a concordancer, is free – you can download it here. My analysis covers the frequency of the lexical item “eish;” its most significant collocate (collocates are other words one can expect to find in proximity to the keyword, i.e. to the search term “eish”); and considers information retrievable from keyword-in-context (KWIC) concordances of “eish”, that is, of the use of the word in context of the sentence. Collocations and concordances in corpora are among the things corpus lexicographers look at to distinguish senses of a lexical item and to generate a dictionary description of the word.

The corpus I am working from contains just over 2.5-million words (tokens), just over 50 000 of them unique (types)9. The texts in the corpus were retrieved semi-automatically from online comments to news articles in 2014, most of them from IOL, as part of a short-lived corpus project at the Dictionary Unit for South African English. The corpus is not particularly large – for an example of a large corpus, see GloWbE (1.9-billion words) – but the material in it is all South African, all public, and all informal writing. I refer to the online comments as “public” and “informal” written discourse because they are, respectively, freely accessible, and not edited for publication. They are largely written in an informal style, and thus contain a fair amount of informal, colloquial and slang material that is not as likely to be found in written form elsewhere. (Informal written public discourse has a long history, going back to ancient graffiti, but it was not until the coming of the worldwide web in the 1990s, and more particularly, the advent of Web 2.0 in the latter part of last decade, that this type of language was freely retrievable and easily analysable in large quantities.)

High-frequency SAE in the corpus of online comments

SAE is low frequency relative to the rest of the words we use every day. Though it’s possible to find language that contains a high frequency of SAE, such as an Athol Fugard or a Paul Slabolepszy play, in general little of the language we use in English in South Africa is actual SAE. This fact is reflected in the corpus of online comments. Leaving aside acronyms and abbreviations (e.g. ANC, R, DA, EFF, ESKOM, BEE, etc.), which are among the most frequent SAE types in the corpus; phrases; and words that otherwise need to be disambiguated (e.g. BLACK, WHITE, WHITES, CAPE, BLACKS, SHAME, etc.), the most frequent SAE word of interest is APARTHEID (1 779 instances, or tokens), followed at some distance by VIVA (318), ZULU (317), EISH (314), RAND (275), CADRES (270), MADIBA (233), JA (216), CADRE (215), and COMRADE (188)10. What is noticeable about EISH here is that, for an informal word, it is fairly highly frequent, and occurs among other SAE words that are mostly more formal, or at least, not colloquial. “Eish” is both applicable in a wide variety of contexts, and – in this corpus – is used more frequently than many other informal SAE lexical items11.

‘Eish’ in informal written public discourse

Table 1 below shows the most frequent statistically significant collocates of EISH in the online comments corpus, down to a frequency of 5 left or right12. The collocates of a word (the keyword) are those words that tend to occur in the vicinity of the keyword in a corpus, here up to five words and the left (5L) and right of the keyword (5R).

Table 1

EISH collocates with 1 147 collocation types, making up 3 056 collocation tokens. The collocates in Table 1 represent the 41 most frequent collocation types (i.e. 3.6% of the collocation types), and their 391 collocation tokens (i.e. 12.8% of the collocation tokens). The most frequent of these collocates, by far, is YOU (85). Notice that by adding WENA (i.e. singular “you” in Xhosa and Zulu, 7) and U (6) to YOU, its prominence as a collocate of EISH is magnified, to the extent that it is more than six times more prevalent than the next most frequent (LOL). Put it this way: there are 51 449 other words in the corpus that EISH could cuddle up to, yet in the 5L-5R range, it only does so with 1 147 – and every 12th one of these is a second-person pronoun. Some other personal pronouns are also highly frequent: IT (15)13, and the first-person ME (11) and US (7); as are auxiliaries of the singular first-person pronoun: M (i.e. I’M, 7) and AM (7). In other words, “eish” attracts certain personal pronouns in statistically significant way. Below, I restrict the analysis to YOU.

The EISH-YOU collocation

Figure 1 below shows some examples of KWIC concordances in AntConc in which the EISH-YOU collocations are statistically significant.

Figure 1

The two most noticeable features about the 85 instances of EISH in these concordances are that it appears to have specific types of positioning in a sentence, and that the YOU element is mostly a direct address to another person.

On the first point: EISH occurs at the end of a sentence only 9.6% of the time. It occurs as a stand-alone interjection, with periods on either side of it (or a period before and an exclamation mark after), in 20.5% of cases. In 48.2% of the instances, EISH occurs at the beginning of a sentence, in a manner that suggests that functions pragmatically to set the tone for what follows. Furthermore, it is followed by an exclamation mark (immediately, or a few words down) only 15.7% of the time, suggesting that, on the whole, it does not rely on extreme punctuation to carry its load – it is not necessarily an interjection that is shouted out, but can be as effective somewhat muted.

On the second point: in the vast majority of cases, the YOU collocate is a direct address to another person: mostly, “eish” is being said to someone else (rather than being a generic stand-in for “one” or “anyone,” as in the last example below). This “other person” is, in general, another commenter, and this feature of “eish” may be an artefact of the anonymity and distance that is set up in the world of online comments by means of the masks of usernames and avatars. That is, it may be an artefact of the genre of informal written public discourse (think about the crassness of some graffiti, which it surely owes to the fact that the writer is absent when the passerby reads the text)14.

Here are some examples of EISH-YOU collocations from the corpus:

Eish… You seem to be late for your comprehension lessons.

Eish Liz, what you are saying is so true, and the whole thing is just too sad.

Eish baba, you musn’t steal the gravy.

Eish you are a closed book, I don’t if I should cheer you or BOOOO you.

Eish … are you trying show what a simpleton you are?

Eish but you are a fogon stoopid

Eish you are one huge moron

Eish, you are confusing his poor brain.

Eish, did you major in ‘Stupidity’ at UNITRA?

Then they say to you who burnt the library, apartheid made you burn the library. Then you say eish, APARTHEID!

Eish, no, you know, there is no crime in SA – we are free here

Considering the EISH-YOU collocation in online comments as informal written public discourse in South Africa

The South African online commenting community (if such a thing can be said to exist), is known to be fractious, antagonistic, and hostile. Even when the text producer is identifiable, such as on social media, some people, especially white South Africans, often don’t give a second thought to posting inflammatory and sometimes straight-up racist material. The incidents are well-known, make it into mainstream news, and don’t bear repeating here. White privilege may lead well-known liberal politicians to make narrow-minded and ahistorical comments, such as Western Cape Premier Helen Zille’s recent defence of colonialism on Twitter. (Like so many others, Zille has since “apologised,” claiming her posts were misconstrued.)

What has been as problematic are the conversation threads Zille’s remarks have spawned in social media forums like Facebook over the last few days, where numerous commenters have defended Zille’s views. What we are seeing is an inability of people making such remarks to attempt to understand the social, political, and historical environment of South Africa, and indeed, Africa as a whole. Many times, this is merely racism disguised as contempt.

The situation with news sites, and the reader comments made there, is even more dire. The possibility of anonymity, and hence the feeling of a lack of restraint, and the distance between the producers of these informal texts and their readers, is an important contributing factor to this direness. A few years back, many local news sites either disabled comments to their articles, or implemented active moderation practices due to the denigrating and disparaging nature of the texts posted there. At that time, the comments sections of News24 articles were notorious for if not in-your-face racist statements (though these were also sometimes found there), then comments that evidenced astounding levels of failure to empathise, or to attempt to progress beyond the commenters’ restricted view of the world. (People would sometimes set up dummy Facebook accounts and log in via these in order to be unaccountable for the material they posted.)

The Daily Maverick, one of the few sites that required commenters to use their RL identities – and attempted to sanction those who didn’t – has “suspended comments on the site” until “the interwebs figures [sic] out a better way to deal with the naughty kids in the class.” This message is present as a footer at the end of each of their article pages. It seems that not even their commenting requirements, and the threat of banning, could influence commenters to adopt appropriate etiquette and a mannered and sociable tone.

The Mail & Guardian Online has had moderation measures in place for years. Recently, they seem to have upped their game – yet, it is still possible to come across comments on the site that are arguably a “subtle,” indirect form of racism. At the same time, however, it doesn’t seem possible to post a comment that calls out such problems with other comments – “your comment is awaiting moderation” results in non-posting of the call-out comment.

(To my mind, these are not really attempts by online publishers to shut down racist discourses on “the interwebs” – these have proliferated. Rather, they are measures taken by the publishers to disassociate their brands with such discourses. These are not benevolent, progressive gestures; they are made from a position of self-interest.)

In the end, though, the problems of online comments as informal written public discourses in South Africa are not essentially ones of the possibilities for anonymity and distance provided by online technologies – rather, they are re-articulations of social problems, where people with certain levels of social comfort and privilege feel little need to reconstruct their attitudes and views towards people who, to their minds, are obviously not “like” them.

So where does “eish” come into all this? We’ve established that the most common collocation of EISH is the second-person pronoun (YOU, WENA, U). Within this group, EISH is often a sentence-initial tone-setter, and YOU is mostly used as a direct address to someone else (rather than a generic stand-in for “one”). The most salient difference between addressor and addressee in these comments is that the former is usually white, and the latter often black. Here, “eish” may function as an indicator of a type of abuse that is common in South African online communities. This ranges from a lack of respect for others, through a determined failure to understand other points of view, to talking down to people whose identities you don’t share, to more direct forms of abuse, as is evident in some of the examples of EISH-YOU collocations above. “Eish” itself is not directly abusive, and as a rebuke it is mild, yet it can be weaponised within a larger structure of disempowerment and inequality.15


The South African interjection “eish” is probably derived, via Tsotsitaal or Isicamtho, from one or more traditional Xhosa or Zulu interjections, most likely “ashi” or “ishi.” “Eish” is common to most if not all languages and demographic groups in South Africa and, in SAE, is probably the most frequently occurring informal term. The most frequent collocate of “eish” in the online comment data analysed here is “you,” where the latter usually occurs as a direct form of address, usually referring to another commenter. In the fractious South African online commenting community, the relationship between “eish” and “you” is potentially problematic as it can signal attitudes of superiority in the context of the inequalities in South African society.


1. ^ Along with “braai,” “ja,” “shame,” “sjoe,” “voetsek,” “yho,” and a few others (and variations of these), it appears to be in use in English and other South African languages across the full demographic spectrum.
2. ^ It is also used in Botswana, and may be used in other southern African countries.
3. ^ Bembe and Beukes (2007: 469) claim “eish” comes from “Is’camtho,” a more contemporary relative of Tsotsitaal. It may have existed in Tsotsitaal first and found its way into Isicamtho.
4. ^ There is less Afrikaans influence in Isicamtho.
5. ^ “Heje” (a form of “he” or “heke”) might also be a contender, though this is less likely. These are historically interjections of approbation or praise, where the latter has the meaning “Well! Right! Good! Well done!” For dictionary entries on these words, and the information on them presented here, see Kropf 1915: 147 and 167.
6. ^ For these entries, and the information on them presented here, see Doke and Vilakazi 1972: 13, 355, 878, and 881, respectively. There are also the following less likely contenders: “hoshi,” an interjection of surprise (305); “oshi,” an interjection of “disapproval, dislike, contempt” (“Ha”) (637); and “woshi,” an interjection of “surprise, admiration, excitement” (“Oh my! Just look!”), or of “contempt, depreciation” (“What nonsense! What an idea!”) (855).
7. ^ When the next word is prefixed by a different vowel, the two vowels combine to form a third which displaces the other two. So: “isiNgesi” (“English”), but “nga-” + “isiNgesi” = “ngesiNgesi” (“in English”).
8. ^ The informal interjections “ey” and “ai” may also have had a phonetic influence here.
9. ^ In other words, there are on average about 500 or so tokens of each type in the corpus. The reality though is that, as in everyday language, there are a relatively small number of words with extremely high frequencies (e.g. THE, TO, AND, etc.), a middling number with moderate frequencies (e.g. REAL, END, GUY, etc.), and an extremely large number with very low frequencies (e.g. MINGLE, RENOWN, BURGLE, etc.).
10. ^ “Cadre” is of course also a word in general English, where it refers to a group of people. What makes its use in South Africa particular is that it is invariably refers to an individual. (It is worth noting that, relative to their national demographic profile, white male South Africans are generally overrepresented in online comments. Their use of “cadre” is often not with the same tone as use by, say, government officials to refer to ANC cadres, and is often disparaging. The same probably goes for “viva” and “comrade.”) Of course, if one were to consider CADRE as a lemma (for a noun, the singular form, with all other forms of the lexical item grouped together with this), then there would be 485 instances of the lemma CADRE (CADRE + CADRES). Similarly, grouping the lexical items COMRADES (159 instances) and COMRADE, there would be 347 instances of the lemma COMRADE. As lemmas, these words – and the discourses in which they occur and co-fashion – are more prevalent than they appear at first sight.
11. ^ From here on, I use small caps for EISH when I am referring to the lexical item in the corpus.
12. ^ In Table 1, T = the total number of times a lexical item collocates with EISH, L = the number of times the collocate appears to the left of the keyword, and R = the number of times it occurs to the right. COLL. = the collocates, the lexical items, themselves. MI = mutual information score, a measure of statistical significance. The higher the MI score, the more likely it is that the nearby lexical item is a collocate. I have included in the table only items with MI ≥ 4 to ensure that I am dealing with collocates and not random co-occurrences. The scores are colour-blocked as follows: yellow – MI ≥ 4, but < 5; orange – MI ≥ 5, but < 6; red – MI ≥ 6, but < 9; purple – MI ≥ 9.
13. ^ In the corpus, the 15 instances of “it” as a statistically significant collocate of EISH are all written as ITS (all but one should in fact have an apostrophe). IT is more frequent (40), but it is not statistically significant (MI = 3.468). It may be the case that ITS is statistically significant because it is an error.
14. ^ I am thinking here less about graffiti as it occurs as a component of contemporary hip-hop, and more about a textual message (whether ancient or modern) tagged on a wall, or behind a public toilet door.
15. ^ In my experience, however, this is not how “eish” usually occurs in real life. If “you” is involved, it tends to be a generic “you” rather than the addressee (as in “Eish, you know, …”).


Bembe, M. P. & Anne-Marie Beukes (2007) “The Use of Slang by Black Youth in Gauteng” in Southern African Linguistics and Applied Language Studies 25(4)

Campbell, C. (1997) “Migrancy, Masculine Identities and Aids: The Psychosocial Context of HIV Transmission on the South African Gold Mines” in Social Science & Medicine 45(2)

Dictionary Unit for South African English (1996) A Dictionary of South African English on Historical Principles. Cape Town: Oxford University Press

Dictionary Unit for South African English (2010) Oxford South African Concise Dictionary. 2nd edition. Cape Town: Oxford University Press

Doke, C. M. & B. W. Vilakazi (1972) Zulu-English Dictionary. 2nd ed. Johannesburg: Witwatersrand University Press

Kropf, A. (1915) A Kafir-English Dictionary. 2nd ed. edited by Robert Godfrey. South Africa: Lovedale Mission Press

Meintjes, H. (2001) “‘Washing Machines Make Lazy Women’: Domestic Appliances and the Negotiation of Women’s Propriety in Soweto” in Journal of Material Culture 6

© GQOM 2017


3 thoughts on “‘Eish’ in South African English: an analysis of the word in informal written public discourse, with a speculative enquiry into its etymology

    1. Hi Graham, no, it’s unfortunately not available online, as far as I am aware. It belongs to the Dictionary Unit for South African English at Rhodes. You should be able to contact them at dsae@ru.ac.za to request access.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s