Expressing my feelings, not yours

One aspect of self-talk of the sort in (1) that is not really mentioned in the literature, as far as I have seen, is that not all people engage in it.

It is recognized in the literature that there is variation regarding “inner speech” or “inner monologue”: some people experience a lot of it, other people less, or even none at all (Hurlburt et al. 2013, 2021, Nedergaard and Lupyan 2024). But I have not seen it discussed, or even mentioned by a word, in relation to self-talk as in (1), for instance in the fairly rich literature on self-talk within psychology, particularly sports psychology, where much research on self-talk is geared toward finding out to what extent people’s sports achievements are affected by positive or negative self-talk (Van Raalte et al. 2016).

But there is variation. It is not just that some people choose not to practice this kind of self-talk like some people choose not to swear or not to use a local accent, but that this kind of language use is alien to them. They find it odd that any otherwise mentally stable person would engage in it. I know this because my wife is one of them. My son, too, doesn’t recognize this use of language. But my daughter does, and I do (not surprisingly). So it isn’t a matter of social or cultural variation, not a matter of nurture, but of personality, presumably with some genetic factor involved. When I gave talks, some years ago, about self-talk, I used to ask the audience if they talked like that themselves. There would typically be some, a definite minority, who said they would never talk like that. My estimation, based on these highly unsystematic observations, is that maybe 20 % of people (British, European, highly educated) don’t practice self-talk as in (1), at all. This suggests that self-talk of the sort described and analyzed in the target paper by Martina Wiltschko (MW; I will refer to both the paper and the author as MW) is not just an extension of ordinary language use, like, say, code switching or excessive swearing, but involves some cognitive mechanism which isn’t universally available, which would be very interesting if it turns out to correlate with other cognitive–psychological properties. It may well be that this is but one facet of the variation regarding inner speech more generally, as investigated in particular by Russell Hurlburt. But this is not obvious. Hurlburt’s research appears to show that there is rather less variation regarding “inner monologue” than claimed in some of the literature (Hurlburt 2011, Hurlburt et al. 2021, Hurlburt and Krumm 2020, but see Nedergaard and Lupyan 2024). In my experience, admittedly limited, the variation regarding (1) is absolute: You practice it more or less frequently or you never do.

MW presents an explanation of the contrast between (2a, b) as self-talk, which is very interesting but appears incomplete in a certain respect.

(2a) is fine as self-talk, (2b) is impossible (here and in the following I will be referring to self-talk of the “speaker-as-listener” type, not the one performed in front of a mirror). In this type of self-talk, you can refer to yourself as either “I” or “you.” In both (2a,b), the two pronouns both refer to the speaker, and they are meant to signify that the speaker is sick of themselves, yet only (2a) can have that meaning. (2b) is oddly uninterpretable as self-talk. It’s interesting that you can have such a robust sense of grammaticality/interpretability in the case of an utterance that is completely private. In the case of (2b), I literally can’t understand what I’m telling myself!

In Holmberg (2010a, 2010b)^[1] I proposed a condition to the effect that “you” in self-talk can’t have the role of experiencer. This condition rules out (2b). MW (following Ritter and Wiltschko 2021) proposes an explanation of this condition, which I’m convinced is on the right track. Assignment of the role of experiencer to the small clause (SC) subject of “sick” means accessing the mind of the referent of the SC subject, in a certain sense, ascribing the property of feeling sick to that mind. This is fine in (2a), where the experiencer, the mind feeling sick, is referred to by the first person pronoun. But (2b), where the experiencer argument is a second person pronoun, doesn’t allow that interpretation. As made clear in MW, the interpretation is ruled out by the grammar, which treats the addressee, encoded as a grammatical category, as another person even when the grammatical representation actually refers to the self, leaving the sentence without interpretation, as self-talk.

But then, how can (2b) be used in ordinary conversation with an interlocutor, as in (3):

Here, the speaker makes a claim about the addressee’s mind although they obviously don’t have access to it. In ordinary talk, you can make a claim about the addressee’s state of mind based on your own assumptions, which may be based on empirical evidence or just speculation. But then, why can’t you do that in self-talk? You don’t have to actually access an addressee’s mind to make a claim about it (which may or may not be true), as in (3). So why do you, in the case of (2b)?

A property of self-talk of the kind exemplified in (2) is that it is not descriptive-informative. It does not put forward a proposition, as (3) does. What it does is express the feelings of the speaker. In terms of speech acts, it is a form of expressive speech act (Ronan 2015; Searle 1969, 1976). These are speech acts that do not put forward propositions, but express the state of mind, the feelings, and attitudes of the speaker. Canonical expressive speech acts include apologies, thanks, and congratulations. I submit that self-talk of the sort in (1) is a variety of expressive speech acts (this is hereby my new contribution to speech-act theory).

What we can’t do is express somebody else’s feelings by an expressive speech act. We can describe them, but we can’t express them. What is interesting about self-talk is that we can’t express even our own feelings if we use the pronoun “you” to refer to ourselves. The grammar doesn’t allow this, as MW argues. This is what rules out (2b) as self-talk, while allowing (2a), where the experiencer argument, the one whose feelings are not just described but expressed, is “I.” The expressive, as opposed to descriptive-informative-propositional, aspect of self-talk as in (1) is not elucidated in MW paper.

Not all self-talk is expressive. Imperatives are common in self-talk (as discussed in MW).

Questions as in (5) occur as well. As self-talk, these are not information questions, but exclamations in the guise of a question. Information questions, arguably, don’t occur as self-talk (of the relevant type).

None of these put forward a proposition. In none of them are you making a claim about the addressee’s mind. But also, in none of them, are you expressing the adressee’s mind, So the use of “you” instead of “I” is not a problem.

So the type of speech act is crucial for self-talk as in (1). Does this mean that the type of speech act, specifically whether it is expressive or not, should be grammatically encoded as a feature of the left-periphery? MW rejects the idea that a dedicated Speech Act Phrase is called for by facts relating to self-talk: “[…] speech acts are constructed and hence cannot define grammatical categories.” With regard to the present case I tend to agree with this: Once the grammar treats the grammatically defined addressee as another person, whose mind is therefore inaccessible to the speaker, as formalized in MW, real world pragmatics does not allow the speaker to express (as opposed to describe) the addressee’s feelings. Grammatical encoding of speech act type is, arguably, superfluous.

Finally, I would like to add some remarks about the use of “we” as pronoun in self-talk. Some people can use “we” when addressing themselves in self-talk, as in We can do it! as encouragement to yourself, instead of the more common You can do it! or I can do it!. MW mentions this in her footnote 6, adding that she will not discuss this case in the paper. I’m not aware of any research on “we” in self-talk reported in the literature, but I wouldn’t be surprised if there is some recent work on this, by now. One paper referenced by MW is Son et al. (2011), a representative of the rich literature on self-talk in sports-related psychology, which reports an experiment comparing “we” versus “I” in self-talk. However, they discuss the use of “we” when referring to the group they have been assigned to as part of the experiment: “we” = I and this group of people. So this is not using “we” to refer to yourself, the speaker speaking to themselves. As well known (at least in the typological literature; Filimonova 2005), “we” can be inclusive, referring to the speaker and the addressee, or exclusive, referring to the speaker and some other person(s) but excluding the addressee. In Son et al.’s experiments, “we” is exclusive. In self-talk, “we” is inclusive. It includes “I,” the speaker, and “you” the addressee, which, in self-talk, is also the speaker. This is not self-evident, but I think it is the right interpretation of “we” in self-talk.

Some years ago I talked about self-talk as part of a third year undergraduate syntax course at Newcastle University, assessed by a research paper. One student, Stephanie Hall, wrote a paper on “we” as self-talk argument.^[2] Based on data from 205 people contacted on social media, she found that about 31 % of them acknowledged using “we” as a pronoun when referring to themselves in self-talk. She didn’t take into account, at that point, that some people don’t do self-talk at all. Assuming, for the sake of argument, that my assessment is right that 20 % of people (in the United Kingdom) don’t ever do self-talk, her findings indicated that about 37 % of people who do use self-talk can use “we” as argument.

Why some people do and others don’t is probably just fortuitous. I think I have started using “we” occasionally, myself, now that I have become accustomed to the idea.

In a follow-up experiment, Hall asked 120 people who reportedly accepted inclusive “we” in self-talk to judge the acceptability of a set of expressions along a 1–5 Likert scale. A hypothesis that she tested was that “we” cannot be used as an argument in self-talk when expressing the feelings of the speaker (as predicted by Holmberg 2010a, 2010b). This was confirmed: (7) was almost universally rejected as self-talk, in contrast with (6).^[3]

As mentioned, according to MW, the reason why “you” as experiencer argument in self-talk is ruled out is that the grammar doesn’t allow access to the mind of the addressee “you,” even in the case where the holder of “the addressee-ground” is actually the speaker. Therefore, the feelings of “you” cannot be expressed in self-talk. Inclusive “we” includes the addressee, so having the addressee express their feelings along with the speaker expressing theirs by using “we” should be ruled out by the same reasoning, which apparently it is.

Expressions such as in (8), canonical examples of expressive speech acts, can have “we” as subject, which may seem to contradict the claim that “we” cannot be the subject of an expressive speech act.

This is not inclusive “we,” though. It refers to the speaker and a person or group not including the addressee, who is instead the recipient of the congratulation or gratitude. The claim here is that inclusive “we” cannot express the feelings of the speaker, because the grammar does not allow access to the mind of the addressee.

(7) is fine when putting forward a proposition in ordinary dialogue, as in (9).

This is expected, as the sentence is not expressing the state of mind of either the speaker or the addressee, but communicating a conjecture about their state of mind. The prediction is, though, that (10) should be unacceptable on a par with (7) in a dialogue, that is if it is an expressive speech act, expressing the feelings of the speaker and the addressee.

Arguably (10) isn’t expressing the feelings of the addressee, though, but those of the speaker, along with a conjecture that the addressee shares those feelings.

With this I leave the topic of “we” in self-talk for future research. I count on Martina Wiltschko to contribute to it.