The rational roles of experiences of utterance meanings

ABSTRACT

The perennial question of the nature of natural-language understanding has received renewed attention in recent years. Two kinds of natural-language understanding, in particular, have captivated the interest of philosophers: linguistic understanding and utterance understanding. While the literature is rife with discussions of linguistic understanding and utterance understanding, the question of how the two types of understanding explanatorily depend on each other has received relatively scant attention. Exceptions include the linguistic ability/know-how views of linguistic understanding proposed by Dean Pettit and Brendan Balcerak Jackson. On these views, to tacitly linguistically understand a sentence just is to possess the linguistic ability/knowledge-how needed to derive/infer what is said by different utterances of the sentence. Despite their focus on linguistic understanding, both views can straightforwardly explain utterance understanding as the output of a derivation/inference from a representation of the sentence uttered. Here, I take issue with these approaches to utterance understanding and then develop an alternative. More specifically, I distinguish two kinds of utterance understanding, experiential and doxastic, and then argue that experiences of what is said by utterances play distinct rational roles in the two kinds of utterance understanding. I conclude by addressing a recent challenge to my proposal.

KEYWORDS:

• Experience of utterance meaning
• dogmatism
• knowledge-how
• linguistic meaning
• speech comprehension
• utterance understanding

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 See e.g., Siegel (2006; 2010), Bayne (2009), Strawson (2009), O'Callaghan (2011; 2015), Azzouni (2013), Dodd (2014), Reiland (2015), Nes (2016), Brogaard (2018a; 2020), Longworth (2008; 2009; 2018), Gasparri and Murez (2019), Drożdżowicz (2021a; 2021b; 2023), Grodniewicz (2020; 2022). In many cases, what lies at the center of dispute is not the nature of language understanding as such but rather (i) the nature of experiences as of what is said (see e.g. Fricker 2003; O'Callaghan 2011; Gasparri and Murez 2019; Brogaard 2018a; 2020; Drożdżowicz 2021a; 2021b), or (ii) what it takes for our seemings or beliefs about what is said to be justified (see e.g., Brogaard 2018a; 2020; Balcerak Jackson 2019; Grodniewicz 2020; 2022).

2 The occurrent/dispositional distinction, which traditionally has been more deeply entrenched in philosophy, is a viable alternative to the explicit/implicit distinction. Mental state (or attitudes), such as beliefs, desires, and states of understanding, are said to be ‘occurrent’ when they are at the forefront of your mind. All of your occurrent mental states are plausibly constituents of your stream of consciousness. Dispositional (or standing) mental states, by contrast, endure in an unconscious form but are generally prone to become occurrent under certain eliciting circumstances. Whether the occurrent/dispositional distinction maps neatly onto the explicit/implicit distinction is a matter of dispute. While it is quite plausible that implicit mental states are always (or necessarily) dispositional, it remains an open question whether there are any mental states that are dispositional yet explicit. Here, I shall assume that the occurrent/dispositional distinction is commensurate with the explicit/implicit distinction.

A final distinction that should be mentioned is the personal-level/subpersonal-level distinction. Its application to mental states and activities derives from Daniel Dennett's (1969, 93) distinction between different explanatory levels on which to explain human behavior: ‘the explanatory level of people and their sensations and activities’ and ‘the subpersonal level’ of brains and cognitive subsystems and their operations and processes. Although personal-level states and activities can be a-rational (e.g., pain and perceptual states), a hallmark of personal-level states and activities is that they are assessable for rationality (e.g., ‘needs’, ‘desires’, ‘intentions’, and ‘beliefs’) (Dennett 1969, 164). By contrast, subpersonal-level states and activities are not subject to normative assessment (except perhaps in the minimal sense that they can be said to function properly or improperly). Due to the elusiveness of the personal-level/subpersonal-level distinction, I shall remain neutral on the question of whether it is commensurate with the explicit/implicit distinction.

3 See also Steven Gross (2010) on the gap between implicit and explicit linguistic knowledge.

4 It should be noted that while the term ‘utterance understanding’ (or ‘comprehension’) has wide currency in contemporary philosophy of language, different authors use the term differently. Most use it as a success term, which is to say that having an inaccurate attitudinal state about what the speaker said in uttering a sentence S and/or lacking sufficient justification for that attitudinal state precludes one from counting as understanding (or comprehending) what the speaker said (Pettit 2002; Fricker 2003; Brogaard, 2018a, 2020; Balcerak Jackson 2019). However, ‘comprehension’ (and ‘comprehension-based’) is also sometimes used more loosely to refer to/describe a hearer's belief about what a speaker said in uttering a sentence S, provided the hearer's belief is formed on the basis of direct exposure to the utterance (as opposed to, say, learning what the speaker said on the basis of testimony) (see e.g., Grodniewicz 2022). On the latter account, forming a belief about what a speaker said in uttering a sentence S on the basis of direct exposure to the speaker's utterance may suffice for the belief to be comprehension-based, even if the belief is inaccurate or unjustified. This is not how I intend to use the term ‘utterance understanding’ (or ‘comprehension’). Rather, in this paper, I shall use ‘utterance understanding’ (and ‘comprehension’) exclusively in the former sense (i.e., as a success term).

5 What is said by an utterance is not always truth-valuable (or alethic) and may not have a propositional structure. Utterances used to perform speech acts, for example, need not be truth-evaluable and may not have a propositional structure (e.g., an utterance of ‘Who stole the cookies?’). The same goes for the verbal articulation of phrases on a list for the purposes of committing them to memory, conveying them to an addressee, or practicing their pronunciation. But when what is said by an utterance is truth-valuable and propositionally structured, we can also refer to what is said as the truth-condition, or the proposition expressed.

6 John Austin (1962) distinguishes between the locutionary, illocutionary, and perlocutionary acts of speech. A locutionary act is an act of speech used to perform a speech act (e.g., ‘I demand that you close the door.’). The illocutionary act is the act performed by uttering the words (e.g., the act of demanding that you close the door). Finally, the perlocutionary effect is the effect the illocutionary performer intends to bring about by performing the illocutionary act (e.g., the door being closed). Here, I shall use ‘speech act’ to mean illocutionary act. The ‘content of a speech act’ refers to what the speaker asserts is the case, promises to do, demands that the addressee do, apologizes for having done, declares to be the case, etc., whereas the ‘force of the speech act’ refers to nature of the speech act performed (e.g., the speech act being an assertion, a promise, a demand, an apology, a declaration, etc.).

7 I am grateful to Brendan Balcerak Jackson (pers. comm.) for helpful discussion here.

8 Carter and Pritchard (2013) argue that if knowledge-how is a kind of knowledge-that, as intellectualists argue, then knowledge-how and knowledge-that must have the same epistemic properties. But Gettier-style cases illustrate that knowledge-how is compatible with a kind of epistemic luck that is not compatible with knowledge-that.

9 I am grateful to an anonymous reviewer for pressing this point.

10 In fact, in some cases, the subject must have an explicit action plan for how to get to φ and exercise that plan intentionally or purposely. To see this, consider Paul Snowdon's (2003) counterexample to the view that knowing how to φ just is being able to φ. As he puts it: ‘A man is in a room, which, because he has not explored it in the least, he does, as yet, not know how to get out of. In fact, there is an obvious exit which he can easily open. He is perfectly able to get out, he can get out, but does not know how to (as yet).’ (Snowdon 2003, 11).

11 Given that the explanandum of PKM says 'A subject knows what sentence S means in language L', it may be argued that Balcerak Jackson (Forthcoming) does not provide an account of propositional linguistic knowledge in terms of knowledge-how but rather an account of linguistic knowledge-wh in terms of knowledge-how. However, whereas it is a matter of controversy whether knowledge-how is a kind of propositional knowledge, it is nearly universally agreed that a sentence headed by 'know-wh' is equivalent to a sentence headed by 'know that', for instance, 'Jena knows what S means' is equivalent to 'For some m, Jena knows that S means m', 'Jeff knows where the meating is taking place' is equivalent to 'For some L, Jeff knows that the meeting is taking place at L', 'Alex knows when dinner is served' is equivalent to 'For some t, Alex knows that dinner is served at t', and so forth (see e.g., Lewis, 1982; Higginbotham, 1996; Braun, 2006; Brogaard, 2009). Not much hinges on how we settle this issue for present purposes, however. If it should turn out that Balcerak Jackson disagrees that the explanandum of PKM (viz. 'A subject knows what sentence S means in language L') is equivalent to a sentence headed by 'know that', this is of no consequence for what I say below. Thanks to Kim Pedersen Phillip and Anna Drożdżowicz here.

12 I am grateful to an anonymous reviewer for pressing this point.

13 The phenomenon of associative agnosia, a brain disorder that impairs object recognition, provides a key bit of evidence that utterance understanding is cognitively comparable to object recognition (Riddoch and Humphreys 2003; cf. Bayne 2009). Associative agnosia has been reported not only for visual object recognition but also for the recognition of the meanings of spoken words (Buchtel and Stewart 1989). Auditory associative agnosia (word meaning deafness) differs from verbal auditory apperceptive agnosia (pure word deafness), which impairs the ability to recognize word sounds rather than word meanings (Buchtel and Stewart 1989).

14 Fricker (2003) calls these kinds of experiences ‘quasi-perceptual’, thus leaving it open whether they are genuinely perceptual or perception-like.

15 That understanding what is said by an utterance typically involves undergoing an experience as of the utterance having a certain meaning is particularly salient when we are learning a foreign language and for the first time effortlessly understand what is said by an utterance in that language (Drożdżowicz 2021b).

16 Note that the alternative accounts on the table (i.e., Pettit's account and the strengthened version of Balcerak-Jackson's view) seem to handle this and the first case perfectly well. So, here, my proposal is in agreement with their accounts.

17 The term ‘apt’ is borrowed from Sosa (2021), although Sosa doesn't apply it to mental states representing linguistic meanings or utterance meanings, nor to experiences.

18 The Hugo case echoes Balcerak Jackson's (2019)'s Rabbit Habit and New Goat cases, which I turn to below.

19 More precisely, although the target of Pettit's third case is the view that understanding the linguistic meaning of a linguistic item requires having a belief about its linguistic meaning, I will argue that contrary to what Pettit seems to think, this case also presents a counterexample to the view that utterance understanding, in its generic sense, requires having a doxastic attitude about what is said.

20 For thinkers sympathetic to dogmatism, see e.g. Chisholm (1966), Pollock and Cruz (1999), Pryor (2000; 2004; 2012; 2013), Huemer (2001; 2005; 2005), Silens (2008; 2013), Tucker (2010a), Chudnoff (2012; 2013; 2014a; 2014b; 2018); Brogaard (2013; 2018b; in press), Ghijsen (2014; 2015), Moretti (2015), Fuqua (2017), Pace (2017), Brogaard and Gatzia (2024).

21 Since I initially drafted this reply to Balcerak Jackson (2019), J. P. Grodniewicz (2022) has independently developed related responses to Balcerak Jackson's (2019) counterexamples to dogmatism. A crucial difference between my reply and that suggested by Grodniewicz (2022) lies in our different approaches to Balcerak Jackson's (2019) second case (see note 21). Unconvinced by available dogmatist replies to this case, Grodniewicz (2022) offers a reliabilist alternative to dogmatism, according to which our beliefs about what is said must be reliably formed on the basis of a reliable exercise of our linguistic competence in order for them to enjoy prima facie justification. As I cannot now give this proposal the scrutiny it deserves, I refer the reader to Grodniewicz (2022) for the details of his account.

22 A reviewer for this journal convinced me that the second case is interestingly different from Rabbit Habit. So, let me briefly deal with it here. Interpretive Clairvoyance: Eddy is meeting Liz for the first time and asks her about her occupation. But when she answers, it sounds to him as if she is speaking Volapyk or some other foreign language entirely unknown to him. Even so, he has a quasi-perception as of her saying she is a vet (adopted from Balcerak Jackson 2019, 396). The problem for the dogmatist, Balcerak Jackson (2019) argues, is that Eddy has a quasi-perception as of what Liz is saying despite not being justified in believing what she said. By way of reply, the case seems to violate the grounding constraint on utterance understanding canvassed in the last section. But I don't think dogmatism need to appeal to etiological issues to adequately reply to Interpretive Clairvoyance: As Balcerak Jackson lays out the example, Eddy's auditory experience as of Liz making ‘unfamiliar vocal noises’ undoubtedly counts as an undercutting defeater of the prima facie justificatory status of his experience as of her saying she is a vet. So, dogmatism will yield the intuitively correct verdict that Eddy is not justified in believing that Liz said that she is a vet. Balcerak Jackson could have assimilated his counterexample to the standard clairvoyance objection which has been leveled against reliabilism (BonJour, 1980) by adding that Eddy has a perfectly reliable interpretive clairvoyance faculty, and that he is aware of this fact. In the envisaged circumstances, Eddy's auditory experience as of Liz making ‘unfamiliar vocal noises’ would not count as a defeater of the prima facie justificatory status of his experience as of her saying she is a vet but he would not count as understanding her utterance on the basis of his auditory experience of the speech sounds she makes because he would be relying on his perfectly reliable interpretive clairvoyance faculty rather than his ordinary linguistic competence and hence would violate the grounding constraint.

23 Grodniewicz (2022) disagrees with me here: ‘I think that Balcerak Jackson is simply mistaken in assuming that a warning about misperception of voice pitch or accent would not defeat the justification of comprehension-based beliefs. In fact, it would have the exact same effect on [Susie]’s justification as the warning about possible confusion of phonemes; it would be an undercutting defeater’ (2022, 114). However, one of the wonders of natural language is that fluent speakers of a language often have no trouble understanding each other despite enormous variations in voice pitch (i.e., the lowness or highness of a person's voice) and accent (e.g., a Boston vs a Texas accent). The pitch/accent constancy found in natural language is akin to the shape/size/color constancy found in visual object (or scene) perception. To revisit the Rabbit Habit case: if Susie had been told by her doctor that she is likely to misperceive people's voice pitch and accent, this by itself would not undermine the justificatory force of her experience as of Elvis saying that he has too many bad rabbits. To be sure: Susie's background information about Elvis might give Susie a reason to doubt the accuracy of her experience as of what Elvis said, but that has nothing to do with accent or voice pitch. Grodniewicz (2022) is right, of course, that intonation (i.e., the variation of pitch across a phrase/sentence) can convey a difference in force (e.g., interrogative versus assertoric force) and expressive meaning (e.g., sounding serious), which by Balcerak Jackson's light (Forthcoming) might count as a change in what is said (e.g., ‘I work here?’ vs ‘I work here!’). For instance, if Susie were told by her doctor that she is likely to misperceive intonation, then she would have reason to question the force of what Elvis said. In that case, she might have wondered if he was asking her whether he had too many bad rabbits rather than asserting that he did have too many, which was how it sounded to her. So, Susie's justified belief that she is likely to misperceive intonation would undermine the justificatory status of her experience as of the force of Elvis's utterance. This is a nice point. But we can grant that point and still maintain that if Susie had come to justifiably believe that she was misperceiving Elvis as uttering (23) with a low-pitched Boston accent rather than, say, a high-pitched New York accent, she would not thereby have had good reason to doubt the veracity of her experience as of what Elvis said in uttering (23).

24 I am grateful to the organizers and participants of the Linguistic Understanding: Perception and Inference conference at the University of Oslo in 2021 and an audience at NYU for the opportunity to present some of these ideas and for excellent discussion and feedback. I am also grateful to Anna Drożdżowicz, Kim Pedersen Phillip, and three anonymous reviewers for this journal for helpful comments on a previous version of this paper and to Casey O'Callaghan, Elijah Chudnoff, and Song Oh for insightful discussion of these and related issues.