Corpora

The corpora used to investigate the frequencies of use of aren't I and other alternatives include Corpus of Contemporary American English (COCA), British National Corpus (BNC) and the Scottish Corpus of Texts & Speech (SCOTS). COCA is probably the most widely-used corpus of English. It contains more than one billion words from eight genres: blogs, web pages, TV and movies subtitles, spoken, fiction, popular magazines, newspapers, and academic texts. BNC was originally created by Oxford University Press in the 1980s to early 1990s. It contains 100 million words of text from a wide range of genres (e.g., spoken, fiction, magazines, newspapers, and academic). SCOTS, which aims to cover the wide range of Scottish English texts today (1945 to the present day), currently contains over 4.5 million words of written and spoken texts. The reasons for choosing COCA and BNC involve that they are large-scale and representative corpora of American English and British English, and they are publicly available. In addition, SCOTS is included on the account that many scholars claim that amn't I is mainly used in Scottish and Irish English (Quirk et al. Reference Quirk1985; Crystal Reference Crystal2021).

Search strategies

In COCA and BNC, the automated search for the question tags aren't I and its alternatives (amn't I, ain't I, am I not and an't I) was implemented respectively. In SCOTS, the automated search for only amn't I was implemented. Considering that the retrieval of question tags is not easy (Tottie and Hoffman Reference Tottie and Hoffman2006), a manual selection was conducted after each automated search in order to avoid the occurrence of which that do not function as tags, and which were counted as irrelevant instances. The exclusion criteria were adopted as follows:

• • The exclusion of instances that do not function as tag questions. (e.g., Why aren't I ?)

• • The exclusion of instances where the question tag does not function as a contraction of am not. (e.g., I got to write something, ain't I ?)

Frequency of use: aren't I

The search results for aren't I in COCA and BNC are demonstrated in Table 1 and Table 2 respectively.

Table 1. Frequencies of aren't I in each section of COCA

Table 2. Frequencies of aren't I in each section of BNC

WORDS(M)*: the total word counts for the whole corpus and each section in the unit of million.

PER MIL*: frequency of occurrence every million words.

As displayed from the data in the above two tables, aren't I was found to occur 0.75 times every million words in COCA and 1.08 times in BNC in the same proportion. A chi-square test was conducted here (chi-square = 12.3836, p = 0.000 < 0.01). The result shows that there is a significant difference between 0.75 and 1.08 per million, which means that the frequency of use of aren't I is higher in British English than in American English. In COCA, aren't I occurred most frequently in the TV/movies section (4.07 times every million words), with the fiction section following after (1.30 times every million words). In comparison, the section with the highest frequency of use of aren't I in BNC is the spoken (5.7 times every million words), with the fiction section at the second place (2.83 times every million words). Based on the above data, we can reasonably conclude that aren't I has already become a commonly accepted contraction form in both British English and American English.

Frequency of use: alternatives to aren't I

The search results for amn't I in COCA, BNC and SCOTS are demonstrated in Table 3.

Table 3. Frequencies of amn't I in each corpus

According to Quirk et al. (Reference Quirk1985, 129) and Crystal (Reference Crystal2021, 37), amn't I is mainly used as the colloquial norm in Scottish and Irish English. That may be the reason why amn't I is rarely used in both American English and British English (3 occurrences in COCA and 0 occurrence in BNC). Considering the fact that SCOTS is a much smaller corpus compared with COCA and BNC, the search result for amn't I in SCOTS (6 occurrences) suggests that amn't I is much more frequently used in Scottish English than in American English and British English.

The search results for ain't I and am I not in COCA and BNC are exhibited in Table 4.

Table 4. Frequencies of ain't I and am I not in COCA and BNC

According to the data above, ain't I was found to occur 0.08 times every million words in COCA and 0.06 times in BNC in the same proportion. An interesting comment is found in the Dictionary of Contemporary American Usage: “The difference between the English aren't and the American ain't is simply the difference we have in the two pronunciations of ‘tomato'”. (Evans and Evans Reference Evans and Evans1957, 23) It is meant to imply that ain't is used more widely in America, as the American equivalent of the British aren't. This result that the frequency of use is higher in COCA seems to coincide with many linguists’ perspective that ain't I is more of a typical American English expression. In addition, compared with aren't I, the frequencies of occurrence of ain't I in both COCA and BNC are much lower (COCA: 0.08 times every million words for ain't I vs. 0.75 times every million words for aren't I; BNC: 0.06 times every million words for ain't I vs. 1.08 times every million words for aren't I). This contrast points to the fact that the contraction form aren't I is used much more frequently than ain't I in both British English and American English.

As shown from the data above, am I not was found to occur 0.04 times every million words in COCA and 0.07 times in BNC in the same proportion, which are much lower than the frequencies of occurrence of aren't I in both COCA and BNC (COCA: 0.04 times every million words for am I not vs. 0.75 times every million words for aren't I; BNC: 0.07 times every million words for am I not vs. 1.08 times every million words for aren't I).

Incidentally, the search results for an't I in COCA and BNC (both zero occurrence) reveal that this contraction form is out of date.

Why aren't I

Crystal (Reference Crystal2021, 37) has a very interesting explanation for the origin of aren't I, which came from people mistaking an't (a substitute people created for amn't) for aren't since [r] after the vowel is silent in the newly emerging Received Pronunciation around 1800. This explanation seems to be intriguing, however, aren't I is first attested in the Literary World (in Yang et al. Reference Yang2017, 207):

Given that writing is more conservative than spoken usage, it is probable that aren't I appeared in speech much earlier. In the earliest attestation from 1872, aren't I was already perceived as a correct and standard form. Due to the fact that it has been standard for more than 150 years, some scholars approve of the usage of aren't I. Thomson and Martinet (Reference Thomson and Martinet1986, 109) consider aren't I as an irregular contraction of am I not. Fowler and Butterfield (Reference Fowler and Butterfield2015, 96) maintain that whatever its origin, aren't I is used in regular and natural tag questions in standard Britain English. Crystal (Reference Crystal2021, 37) holds that aren't I is the standard form in British English, and it is widely used in American English too. However, there are those who put rules above reasonability and consider aren't I illogical for the reason that “I” and “are” are not consistent grammatically. (Phillipps Reference Phillipps1984; Partridge Reference Partridge1973)

Nevertheless, from our point of view, the use of aren't I is an ideal method to deal with the real difficulty faced in question tags: there is no appropriate contraction form for am not (amn't is considered unacceptable by most English-speaking people). We will find its wide use reasonable if we analyze it from another perspective. As is known, two interlocutors are needed at most to start and propel a conversation: the addresser and the addressee. In speech communication, the addresser and the addressee usually ask and answer questions in turn. There is also the case of asking a question and answering it by oneself. An example is shown in (1):

1. (1) Person A: Am I a student?

   Person B₁:Yes, you are. (No, you are not.)

   Person B₂:Yes, I am. (No, I am not.)

In this example, Question A has two possible answers: B₁ and B₂. B₁ is the addressee's answer to the question (dialogue), while B₂ is the addresser's answer to the question he or she poses (monologue). In the latter case, the role “I” is both the addresser and the addressee. An instance is shown in (2):

1. (2) Person A: I am a student, aren't I?

   Person B₁:Yes, you are. (No, you are not.)

   Person B₂:Yes, I am. (No, I am not.)

In this instance, similar to Example One, Question A also has two possible answers: B₁ and B₂. B₁ is the addressee's answer to the addresser's question (dialogue), whereas B₂ is the addresser's answer to the question he or she poses (monologue). Therefore, “I” in this question tag may play double roles of both the addresser and the addressee. Given the fact that people refuse to use amn't, the use of aren't I here appears to be a smart coincidence: “I” in the question tag still represents the addresser, while “are” in the same question tag marks its potential role as an addressee, for “are” can be collocated with the singular addressee “you”.

Why not amn't I

The fact that most English-speaking people decline to use amn't I may result from its pronunciation problem. In terms of articulation rules in modern English, the consonant cluster [mn] is awkward. Although [m] and [n] are both nasal consonants, their places of articulation are distinct. The bilabial [m] is produced using both lips, while the alveolar [n] is produced with the front part of the tongue on the alveolar ridge. Therefore, the related articulators cannot transform from [m] to [n] instantly. Moreover, the power in chest cavity used to produce these two consonants in succession is not sufficient. Hudson (Reference Hudson2000, 302) remarks on the amn't form like this: “we don't know how to pronounce (or write) the word, and we can't use it.” He is not the only linguist to make such kind of comment on amn't. Rothstein and Rothstein (Reference Rothstein and Rothstein2009, 50–51) also hold that we cannot say amn't I for the reason that m and n adjoin, and we are forced to pronounce each word in full.

However, despite the fact that the consonant cluster [mn] is difficult to pronounce, there are some English words with mn combination in spelling, the common instances include mnemonic, column, condemn, damn, hymn and solemn. When pronouncing these words, either [m] or [n] is not voiced. For example, mnemonic is pronounced as [nɪˈmɒnɪk], and column is pronounced as [ˈkɒləm]. In that case, why can't this elision rule apply to the pronunciation of amn't as it does to the above-mentioned words? The reason lies in that when am and not is contracted to amn't, this negative contraction should be stressed (Swan Reference Swan2019, 680), and thus the elision rule cannot apply. In addition, the consonant cluster [mn] in amn't is followed by a consonant rather than a vowel in spelling, then [m] and [n] cannot form syllables with their neighboring vowels respectively like in the case of the word omniscient whose pronunciation is [ɒmˈnɪsɪənt].

Other alternatives

Apart from amn't I, there are some other alternatives that are available for aren't I: am I not, an't I and ain't I. The “bookish” am I not is often used in intensely formal situations, such as in literary language. For example, Mrs. Hunnard gave delighted little cries. “If only I had a pretty voice -- I'm quite without tune, am I not?” (From BNC, fiction: The diamond waterfall) However, this form is rarely used in colloquial speech, or it will sound completely unnatural and stuffy.

As for the obsolete an't I, there are two different explanations for its origin. According to Jespersen (Reference Jespersen1917, 118), in earlier times, am not became an't in a similar way that cannot and shall not became can't and shan't, with a change of the vowel from [æ] to [a:]. Crystal (Reference Crystal2021, 37) provided another explanation for the origin of an't: The form amn't, which owns an awkward consonant cluster, was naturally simplified as an't for the reason of pronunciation convenience that [n] and [t] are both articulated on the alveolar ridge behind the top teeth. Nevertheless, whatever the origin of an't, this form is obviously outdated nowadays. Despite the fact that an't I was very popular a little earlier in modern English, as in Smollett and Dickens (Curme Reference Curme1931, 137), we can find little support for this contraction form in actual modern usage.

With respect to the form ain't I, which is generally denounced as illiterate or vulgar, people seem to hold a more tolerant attitude. Regarding the origin of this form, my speculation is that it may have something to do with the Great Vowel Shift, through which the pronunciations of all Middle English long vowels were changed. Among them, the pronunciation of the long vowel [aː] experienced several changes ([æː], [ɛː] and [eː]), and finally shifted to [ei]. (Stockwell Reference Stockwell, Minkova and Stockwell2002) Given this fact, it is probable that the form ain't I came from the form an't I whose pronunciation was changed in some dialects during the Great Vowel Shift, and then when recording this form in writing after the Great Vowel Shift, it became ain't I. As Random House Webster's College Dictionary (2016, 36) describes, ain't is nonstandard except in some dialects, and it can not only replace am not, are not and is not, but also substitute have not, has not, do not, does not and did not. As regards the usage of ain't, this Dictionary (2016, 36) further claims that ain't is more common in uneducated speech than in educated, but it occurs occasionally in the informal speech of the educated, especially in the interrogative ain't I used as a substitute for the formal am I not or for the ungrammatical aren't I or for the awkward amn't I. However, from my standpoint, the fact that ain't can replace am not, are not and is not and even have not, has not, do not, does not and did not in speech will lead to greater chaos and complexity in actual usage. Superficially, using ain't in speech will reduce addressers’ burden because it can express a variety of meanings in communication, and addressers can thus simplify their utterance. However, the consequence of addressers’ “least effort” may be that addressees have to spend more time determining the appropriate meaning of ain't from a range of possibilities.