Introduction
To prescriptive grammarians, the nonstandard use of pronoun cases can evoke a reaction similar to that of nails on a chalkboard; however, the practice is not uncommon, and not only among language learners. But why is this, and why do some forms appear to be more susceptible to creative case use than others? Most proficient English speakers would never dream of saying Give those papers to I, but many would happily say Give those papers to John and I, suggesting that something about the coordinate construction makes some speakers believe pronoun case is flexible. Furthermore, the first‑person singular seems to be most prone to nonstandard use. This article looks at corpus data to see whether there are any clues regarding why this is so and posits the following explanations: First, that X and I and X and myself are constructions (in the construction grammar sense) which are learned as chunks and are then popped into sentence structures regardless of the grammatical function of the coordinate construction. Second, that the near‑obsolescence of case in English leaves English speakers vulnerable to overgeneralizations triggered by prescriptivist corrections, resulting in a general uneasiness regarding the use of the pronoun me in coordinate constructions, even when the coordinate construction is the object of the sentence.
Research questions
The current study seeks to answer the following research questions:
- •
Why and when are first‑person singular nominative or reflexive pronouns substituted for object pronouns in coordinate constructions?
• What evidence is there for the existence of X and I and X and myself constructions?
Literature review
The variability of pronoun case use in English has been a significant cause of heated debate among linguists and occasional hand‑wringing by grammarians dating back generations (e.g. Jespersen Reference Jespersen1933). There are differing theories regarding the cause of this phenomenon (e.g. Quinn Reference Quinn2005); however, there has not been much corpus‑based research into the matter (Maier Reference Maier2013). Below is a brief summary of some of the research that has been done on the use of nominative or reflexive pronouns in object position.
Subject pronouns in object position
The term hypercorrect is commonly used to denote a subject pronoun used in lieu of an object pronoun. Whether this term is appropriate is certainly debatable, given that it is not clear how using an incorrect (from a prescriptivist standpoint) case could be considered overly correct. It seems to suggest that the motivation for the practice is aspirational. While it is difficult to ascertain the speaker’s or writer’s motivation through corpus analysis, we will come back to this question in the discussion. In the meantime, in order to avoid confusion and keep with commonly used terminology, this paper will use the term hypercorrect.
Kirner–Ludwig et al. (Reference Kirner-Ludwig, Oberhofer and Heiss2023) examined scripted television shows to conduct a corpus‑pragmatic study of hypercorrect uses of subject pronouns in object positions. They argued that the practice is growing in use, but little quantitative evidence exists; in fact, the authors only identified one previous study that used corpora to examine this phenomenon. The researchers identified ‘correct’ subject and ‘correct’ object forms of pronomial binomials, as well as ‘incorrect’ object forms in subject position and ‘hypercorrect’ subject forms in object position (p. 383). They found 77 cases of hypercorrect pronouns in coordinate constructions, and in all cases the hypercorrect pronoun was found after the coordinator. In other words, people are infinitely more likely to say give it to John and I, than give it to I and John. In two cases both elements were hypercorrect. Most of the cases found involved the first‑person singular pronoun, and the coordinator and. Prior to the study, the researchers predicted that the hypercorrect form would be used more in UK than in US English given that ‘UK English is usually associated with a higher level of prestige’ (p. 384). The authors did not clarify why they believed UK English to be associated with higher prestige, nor why using nonstandard grammar would be considered prestigious. However, their prediction turned out to be correct as the X and I hypercorrect form was more frequent in British than in US shows at a rate of 14 per cent to six per cent.
Angermeyer and Singler (Reference Angermeyer and Singler2003) argued that using hypercorrect pronoun binomials is very common in modern American English, being implemented in any range of registers from daily speech to edited texts. It is a practice which dates back at least 400 years.
Parker et al. (Reference Parker, Riley and Meyer1990) stated that case assignment rules are relaxed in all coordinate structures. They argued that the normal rules are relaxed because intervening noun phrases act as a ‘barrier’ (p. 56).
Reflexive pronouns in object position
According to binding theory, ‘reflexive pronouns bind with their antecedent in their local domain’ (Paterson Reference Paterson, Brezina, Love and Aijmer2018, n.p.). Using a reflexive in the object position violates this if the reflexive cannot bind with the subject. Therefore, I threw the ball to myself would be considered grammatical, since myself binds with the first person singular I; whereas John passed the ball to myself* would be considered ungrammatical, since myself cannot bind with John. It is thus ‘referentially incomplete’ (Sportiche Reference Sportiche2013, 189). Reflexives with nothing to bind to are known as untriggered reflexives (Paterson Reference Paterson, Brezina, Love and Aijmer2018). However, it is not clear that all speakers have the same compunctions about using untriggered reflexives as do grammarians (Lederer Reference Lederer2013).
Identifying ‘ungrammatical’ untriggered reflexives in a corpus is less straight forward than identifying hypercorrect forms, since the binding pronoun can be implied. This is a picture of myself (example from Paterson Reference Paterson, Brezina, Love and Aijmer2018, n.p.) could be considered grammatical if the photo in question were a selfie. In this case the which I took could be implied but omitted from the middle of This is a picture which I took of myself. If the picture were taken by another, then this is a picture of me would be considered more grammatical. This means that the researcher must make great efforts to unravel the context when making grammaticality judgments, and sometimes the context provided is not sufficient. However, some researchers (e.g. Parker et al. Reference Parker, Riley and Meyer1990, 51) deem the untriggered reflexive ‘acceptable’ in such situations even without including the caveat that it should be a self‑taken photo. They do not specify what their judgment of acceptability is based on, but it is likely intuition.
Parker et al. (Reference Parker, Riley and Meyer1990, 50) further argued that untriggered reflexives are ‘acceptable’ as part of coordinate noun phrases, or prepositional phrases modifying noun phrases, such as It is for people like yourself. They also posited that untriggered reflexives are more likely to be substituted for first‑ and second‑person pronouns than for third‑person pronouns.
Paterson (Reference Paterson, Brezina, Love and Aijmer2018) searched spoken English in the BNC2014 and classified reflexives as triggered, implied triggered, self‑intensifiers, untriggered, and unclear. She found that reflexives were only untriggered two per cent of the time. Also, the first‑person singular was by far the most common culprit, where five per cent of reflexives were deemed untriggered, followed by the second‑person singular at 3.1 per cent. All the other untriggered reflexives accounted for only one per cent of the total reflexive use. The normalized rates for first‑person singular untriggered reflexives was 6.23 per million words and second‑person singular was 3.34 per million, while others were less than one per million. Thirteen out of 30 tokens of untriggered myself were either preceded or followed by a coordinating conjunction. In other words, they were part of coordinate constructions. The examples given were all in object position. The use of untriggered reflexives was correlated with social class, where, perhaps counter‑intuitively, the upper classes, and also the very lower class, were more likely to use them. Paterson further concluded that 90 per cent of speakers produced no untriggered reflexives, and suggested it is a stable phenomenon.
In their examination of scripted television shows, Kirner–Ludwig et al. (Reference Kirner-Ludwig, Oberhofer and Heiss2023) found that the untriggered reflexive myself was used in object position 2.7 per cent of the time in situations when it could have been used.
Procedure
The current study examined the nonstandard use of myself and I in coordinate constructions using the Corpus of Contemporary American English (COCA)(Davies Reference Davies2008–). After the search was conducted, the results were randomized, and the first 50 items from each search were examined in detail and were used for a quantitative analysis. The remaining results were also examined for interesting qualitative examples. The various terms searched are as follows:
In order to find examples of I used as an object in coordinate constructions, the search terms ‘and I .’ and ‘I and PRON .’ were used. The period was included in the search terms because, since English is a Subject Verb Object (SVO) language, it was hypothesized that searching at the end of sentences would yield a higher hit rate for subjects used as objects. (It was not possible to confirm this since a search of ‘and I’ alone yielded 894,260 results which is too much for a context search in COCA. However, a search of ‘you and I’ with no period yielded only one hypercorrect use out of 50, whereas ‘you and I .’ yielded 23.) In order to compare the findings to cases where I was used as an object that was not necessarily part of a coordinate construction, the search term ‘I .’ was also included.
In order to determine whether the first‑person singular pronoun was indeed the most likely culprit for hypercorrection in coordinate constructions, searches were conducted for ‘and he .’; ‘and she .’; and ‘and we .’ No searches were conducted for ‘and you .’ since the second‑person pronoun is ambiguous, being used for both object or subject case.
In order to find examples of untriggered reflexives being used in lieu of object pronouns, the search term ‘and myself .’ was used. The rationale for the inclusion of the period was the same as that above. A separate search for ‘myself .’ was also conducted to determine whether the presence of a coordinate construction made the use of an untriggered reflexive more likely. To potentially find evidence for an X and myself construction which could be swapped in and out of sentences regardless of the function, searches were also conducted at the beginnings of sentences for ‘. * and myself’ and ‘. Myself’, examining the likelihood that untriggered reflexives within or without a coordinate construction could be used in lieu of a subject pronoun.
Results
General information regarding the search terms and results is presented in Table 1.
Table 1. Search results
The Total column represents the number of concordance lines returned for each search, while the Sample Size indicates how many of those were examined in detail. Relevant Tokens refers to the number of hypercorrect or untriggered reflexives found within the sample. Raw frequency and relevant token frequency are presented in words per million (wpm). Where tokens were drawn from a random sample of 50, relevant token frequencies are estimates based on the proportion of relevant tokens within the sample.
Hypercorrect subject pronouns
The search for ‘and I .’ yielded 14 unambiguously hypercorrect usages among the first 50 randomized items drawn from 810 results. All numbered examples throughout this paper are drawn from COCA (Davies Reference Davies2008–). An example of an unambiguously hypercorrect form is shown in (1):
(1) [ … ] you’re boring your mother and I.
Of the remaining search items, 22 were subject disambiguations, such as the one seen in (2):
(2) [ … ] we ended up staying, my wife and I.
Here my wife and I are clarifying the subject we, and therefore are quite rightly presented in the subject case. The rest were sentence fragments, subjects followed by ellipses, or otherwise unclear. One of the most interesting examples for the current purpose was the sentence (3) from a scripted television show which was drawn from among the hypercorrect examples:
(3) The aliens are coming for us, you and I.
Although the author uses the object for the initial us, the disambiguation involves the subject I.
A search for ‘I and PRON .’ yielded only 18 results in total. Most of these were meta statements about grammar, such as in (4):
(4) You know this because the author uses pronouns like I and me.
None of these 18 results featured hypercorrect forms.
A search for ‘I .’ yielded 6029 results. Out of the first 50 randomly selected results, only three were hypercorrect forms, and all three of these were the second part of coordinate constructions, using the coordinator and. An interesting result was phrase (5), taken from a movie:
(5) Let’s get out of here, you and I.
Since the you and I is a disambiguation of the object ‘s (us), it is noteworthy that the scriptwriter changed to the subject case for the and I.
As compared to the 810 results yielded by the search for ‘and I .’, searches for other subject pronouns at the ends of sentences resulted in far fewer hits: ‘and he .’ yielded 28 results; ‘and she .’ yielded 21; ‘and we .’ yielded 25; and ‘and they .’ yielded 13. Of course, as with the ‘and I .’ searches, not all of these results are hypercorrect forms. For example, looking at the 28 ‘and he .’ results, the most common form, we find that actually only five are examples of hypercorrect usage. The 25 results for ‘and we .’ yielded no hypercorrect results.
Untriggered reflexives
A search for ‘and myself .’ yielded 651 results, and the first 50 from a random selection included 24 unambiguously untriggered examples, such as examples (6), (7) and (8):
(6) Well, it’s just the cameraman and myself
(7) He hands a list of some sort to both Drummond and myself
(8) Here’s a picture that someone took of Mr. Baccigalupi and myself.
It is much more difficult to judge correctness as the binding pronoun can be quite far from the reflexive, so it is often necessary to look at a much broader context before making a grammaticality judgment.
A search for ‘myself .’ without the and yielded 40,265 results, and the first 50 from a random selection included only three untriggered examples, one of which was part of a coordinate construction following an and.
A search of ‘. * and myself’, looking for untriggered coordinate constructions containing myself used in lieu of subject forms yielded only 36 results. However, of these, 17 were eliminated from the sample because they were not actually sentence initial, but rather came up in the search because of the period following an abbreviated title, as in example (9):
(9) [ … ] Mr. Thornton and myself [ … ].
Another two were eliminated because they were part of fragments where grammaticality was difficult to assess, and one was eliminated because it was part of a deliberate idiomatic expression, seen in example (10):
(10) Me and myself making cranberry molds.
This left only 16 examples of reflexives being used in lieu of subjects in coordinate constructions. Of these, perhaps the most significant for the current purposes was the one which used a subject pronoun as the part of the coordinate construction before the coordinator, seen in example (11):
(11) You and myself used to hope [ … ].
A search of ‘. Myself’ yielded 425 results. A randomized selection of the first 50 yielded the following items: 22 were triggered, where the reflexive was fronted, as in example (12):
(12) Myself, I got three [ … ].
Seven were fragments, where the grammaticality was difficult to assess. Two were fragments where the triggering pronoun was relatively clear, as in example (13):
(13) # Just. Can’t. Help. Myself.
In this case the triggering pronoun is clearly implied, this being a modification of the commonly used construction ‘I just can’t help myself.’ Finally, 19 were untriggered, as in example (14), or as part of a coordinate construction as the part before the coordinator, as in example (15):
(14) Myself felt no different [ … ]
(15) Myself and Jamie met whilst studying in university about 5 years ago.
In example (15), the use of another marked, potentially aspirational form whilst could provide some insights as to the speaker’s motivation.
Discussion
In order to establish that these forms are constructions, we should be able to demonstrate, among other things, that they are used in a way that is contrary to standard ‘rules’ of grammar (Goldberg Reference Goldberg1995, 4; Croft Reference Croft, Geeraerts and Cuykens2007, 467), that they have idiosyncratic properties (Goldberg Reference Goldberg1995), and that they are learned and stored as chunks (Croft Reference Croft, Geeraerts and Cuykens2007). The following discussion looks at the evidence from the corpus search in this light.
Hypercorrect subject pronouns
If we assume that the random sample of 50 concordances from the ‘and I .’ search, which contained 14 hypercorrect forms, is representative of the whole population of 810 results, then we could extrapolate that COCA contains roughly 227 examples of the hypercorrect and I at the ends of a sentences. If we compare this to the zero hypercorrect results from the ‘I and PRON .’ search, it seems to confirm the finding of Kirner–Ludwig et al. (Reference Kirner-Ludwig, Oberhofer and Heiss2023) that the hypercorrect form is more likely to follow the coordinator. The five total incidences of hypercorrect forms for the ‘and he .’ search, and the zero for the ‘and we .’ search, also strongly support the above cited researchers’ finding that the first‑person singular is overwhelmingly the main culprit for hypercorrection. All this would be strong evidence for the existence of an X and I construction.
However, as noted by Parker et al. (Reference Parker, Riley and Meyer1990, 56), an alternative explanation could be the distance between the verb and the pronoun, placing an additional burden on short‑term memory and mental computation. Since the hypercorrect form is more likely to occur after the coordinator, this means that at least one other object and a coordinator are added to the ‘barrier’ between the verb (or the preposition) and the pronoun. While saying give it to I would just sound wrong, the more clutter that is interjected between the to and the I, the less intense the feeling of wrongness would become. This would further explain why ‘I and PRON .’ is not attested, and, given that politeness dictates the first‑person pronoun should come last in a sequence, would explain why I is the most common culprit for hypercorrection.
Nonetheless, the case for an X and I construction is still robust, and is further strengthened by evidence provided by a closer look at the concordances. The following examples are all from COCA, although not from the 50 random items used for the quantitative analysis.
One interesting concordance is a list, seen in example (16):
(16) Heaven and Earth. Adam and Eve. You and I. Us and them. It’s what creates divisions [ … ].
Here, since it is a fragment, there is no unambiguously correct case form to use. However, one would expect the default consideration in a sequence to be consistency. It is interesting, then, that the author used the object form for us and them, but the subject form for the first‑person singular portion of the coordinate construction. This could indicate that the drive for an X and I construction outweighs any consideration for consistency within the list.
Another example (17) suggests that the speaker uses standard pronoun forms when they are not part of coordinate constructions:
(17) We have an alliance of Mike, rc, Peter, and I. Peter has been a bit more genuine with me.
In the second sentence, the speaker demonstrates an understanding of standard case use, but this understanding seems to be eclipsed by the urge to use the X and I construction in the first sentence.
Sometimes, there is a mixing of object with subject pronouns within the same coordinate construction. This happened in cases of subject complements, where the prescriptivist choice might be the subject form, but the overwhelmingly most common usage would be the object pronoun (Siemund et al. Reference Siemund, Maier, Schweinberger, Paulasto and Penttilä2009). Some speakers seem to deal with this ambiguity by mixing and matching cases, but it appears to be by far the most common for the subject case to be represented by the first‑person singular pronoun following the coordinator. For example, in (18):
(18) it was just him and I.
This could also occur in sentences where the subject form should have been used to disambiguate the subject, as in example (19):
(19) We are good friends, her and I.
Once again, it appears the and I construction can be popped in wherever the first‑person singular pronoun plays a part in a coordinate construction, even if the natural inclination of the speaker seems to be towards using the object case.
To provide conclusive evidence of the existence of an X and I construction, we should be able to demonstrate that it is learned as a chunk. This is not easy to do, but there is some circumstantial evidence which could support the hypothesis.
If we subscribe to the argument that English does not actually have cases, and that the pronouns are vestiges of an obsolete grammatical element (Hudson Reference Hudson1995), then English speakers would presumably be significantly less familiar with the function and application of cases than would be speakers of languages with a well‑defined universal case system, like German. In fact, it does appear that pronoun cases have more flexibility in English than in German. For subject complements, for example, many prescriptivists demand the subject form, and yet, as mentioned above, corpus studies find that 99 per cent of UK speakers use the non‑prescriptive object form (Siemund et al. Reference Siemund, Maier, Schweinberger, Paulasto and Penttilä2009). To some degree, both it is I and it is me could be considered correct, depending on the register. However, in German, es bin mich (object form) would be unambiguously incorrect. Conversely, in French, saying c’est je (subject form) would draw strange looks from native speakers. (In fact, it was possibly the mix of Norman French grammar with Anglo‑Saxon grammar which weakened the English case system, but unraveling this conjecture is beyond the scope of the present paper.)
If it is indeed true that English speakers have a tenuous grasp on case, then this would leave them vulnerable to errors of overcorrection. One hypothesis, then, would be that the X and I construction is learned as an overgeneralization of corrections that we are subject to in childhood. When small children are learning English, it is common for them to say things like ‘Me and Johnny went to the park’. Well meaning adults will often correct them by saying ‘Johnny and I went to the park,’ thereby correcting not only the case of the pronoun but also implying the politeness principle of putting oneself last. The child, who does not understand that it is the case being corrected, instead thinks it is the coordinate construction being corrected and begins to overgeneralize by using hypercorrect forms in lieu of object pronouns. The prevalence of this phenomenon with the first‑person singular would provide further support for this hypothesis; since children generally have a self‑centered world view, they spend more time talking about themselves, which results in more potential for errors in the first‑person, and, in turn, more corrections.
This phenomenon may extend beyond childhood. A meta example from COCA (20) perfectly outlines how this situation plays out:
(20)
Me and my friends have driven past this place so many times, but this is the first time I've actually come into this bar.
My friends and I.
Pardon me?
Well, you said, ‘Me and my friends.’ But the correct phrasing would actually be ‘my friends and I.’
(Married in a Year, as quoted in COCA (Davies Reference Davies2008)
In this case the corrector is absolutely right, but the correctee may not understand why they are being corrected, and may, like the child, assume it is the coordinate construction being corrected rather than the case. This lack of understanding may also cause adults to make erroneous corrections of forms that are actually correct, which would compound the confusion which some speakers feel regarding pronoun case and might lead to a general unease about using the pronoun me in coordinate constructions, even when it would be the grammatically correct form. Some speakers might tend to replace it with an I, whereas others might opt for an untriggered reflexive, which brings us to our next section.
Untriggered reflexives
The fact that ‘and myself .’ yielded a far higher per centage of untriggered reflexives than ‘myself .’ provides support for an X and myself construction. However, the relatively small number of X and myself forms being used in the subject position, as well as the relatively high incidence of myself being used as a subject on its own or as the initial part of a coordinate construction could cast some doubt on this.
If an X and myself construction does not explain the majority of untriggered reflexives, there must be an alternative explanation. The fact that the first‑person (followed by the second‑person) is the most likely to be used untriggered, whereas the third‑person rarely is (Paterson Reference Paterson, Brezina, Love and Aijmer2018), could point to a pragmatic explanation. The third‑person can produce grammatical forms regardless of whether the object is in the accusative or the reflexive. For example, he looked in the mirror and saw him and he looked in the mirror and saw himself are both grammatical, although they have different meanings. In the first case, we can imagine a second man standing behind the man who is looking in the mirror. However, in the first‑person (as well as the second‑person) this is not possible. I looked in the mirror and saw myself would be considered grammatical, whereas I looked in the mirror and saw me would likely not, since there can be no additional first‑person entity. Thus there is no danger of ambiguity – the chief nemesis of good communication. Only third‑person object and reflexive pronouns appear in complementary distribution, leading to the potential of misunderstanding, which could be why people are more careful when using the third‑person reflexive, leading to a lower incidence of untriggered forms. A Further example of this is provided by Lederer (Reference Lederer2013, 484): Peter said that John loves him and Peter said that John loves himself are both grammatical and unambiguous. Building on Lederer, we can further argue that while I said that John loves me and you said that John loves you are both fine, I said that John loves myself and you said that John loves yourself would both be considered nonstandard English.
Conclusion
There is strong evidence for the existence of an X and I construction. Among the most compelling findings are: ‘I and PRON .’ is not attested; ‘and I .’ is more attested than ‘and he .’, ‘and she .’ and ‘and we .’; and a high per centage (28%) of ‘and I .’ results were hypercorrect, whereas no hypercorrect instances that were not part of a coordinate construction were found in a search of ‘I .’.
The most compelling evidence for an X and myself construction is that a search of ‘and myself .’ yielded 48 per cent untriggered results, whereas ‘myself .’ yielded only six per cent; also that ‘PRON and myself .’ is more attested than ‘myself and PRON .’ Conversely, the most compelling evidence for an alternate explanation for the use of untriggered reflexives as replacements for object pronouns rests in the relatively high incidence of the untriggered myself being used as a subject regardless of whether it was part of a coordinate construction or not. As such, the evidence for an X and myself construction is not as compelling as that for an X and I construction.
However, despite the existence of various potential contributing factors, overgeneralization might still be a root cause behind the nonstandard use of both first‑person singular subjects and reflexives in object position. If we subscribe to the construction grammar idea that computation is expensive and storage is cheap, it would make sense that speakers who, for whatever reason, feel uneasy about using the object pronoun me in certain situations would develop strategies that help them avoid the word without having to invest much energy into thinking about case. This strategy could take the form of an X and I construction, or of simply using an untriggered reflexive.
CHARLIE TAYLOR is a graduate student in the Department of English Studies at the University of Graz. He holds a professional Master of Education degree from Queen’s University in Canada and is completing an MA in English linguistics. His research focuses on second language acquisition, particularly English as a foreign language and learner attitudes. He has extensive experience teaching languages at the secondary and post‑secondary level and publishes regularly in refereed journals as well as professional teaching magazines. Alongside his core research, he has developed a growing interest in empirical approaches to language use and grammatical variation, which is the focus of the current article. Email: 15cgt1@queensu.ca
Open access
