Foundational Terms
Speech Act: JL Austin’s idea that when we say things, we’re also doing things. Saying “I promise” isn’t just describing a state of affairs, it’s performing the act of promising.
JL Austin distinguished three layers:
- Locutionary act: producing meaningful sounds
- Illocutionary act: what you’re doing in saying it—promising, refusing, asserting, etc.
- Perlocutionary act: the effects your utterance produces in the audience—persuading them, scaring them, etc.
Illocutionary force: The particular kind of action a speech act performs. “It’s cold in here” can be all of an assertion, a request, or a complaint.
1. Introduction
The usual reasons people give for caring about language loss (extinction) are instrumental — they treat the language as a means to some other end. Languages are useful for science, or language loss is a symptom of cultural oppression, so we should fight the oppression.
But imagine if you could fix all of those things and a language still dies out because everyone simply gradually switches to the dominant language for convenience. No coercion or injustice. It seems there is still something to mourn. The loss of the language matters in itself, independently.
The three main claims, each foundational or motivating for the other:
- “I will argue that understanding what someone is doing with her words often requires attending not just to what she says, but to how she says it.” (Nowak, 2020, p. 833)
- The methodological thesis statement. Understanding speech requires attending to how something is said and not just what is said. Thus evidence for this claim is going to come from poetry, colloquial speech, etc. A methodological divergence from disciplinary boundaries that typically study constructed speech acts such as “the cat is on the mat.”
- “I will argue that if no one else can speak that language, or if she is cut off from the community of speakers, she is deprived of the ability to activate the particular network of associations, historical and otherwise, that gives that language a distinctive character, and its speakers each a distinctive voice.” (Nowak, 2020, p. 833)
- The bridging claim: if speech acts depend on language-specific features, and those features depend on a community of speakers who share the relevant network of associations, then being cut off from that community deprives you of the ability to perform those acts—speakers are “silenced.”
- “I will claim that speakers of threatened languages are harmed when their languages disappear, because (i) humans have a fundamental interest in being fully empowered members of a speech community, and (ii) the silencing that occurs when languages are lost is silencing produced systematically and by unjust mechanisms; that is, it is silencing that people suffer in virtue of being part of disempowered communities.” (Nowak, 2020, p. 834)
- The normative claim: the fact of silencing constitutes a morally significant harm.
2. Linguistic metadata and the identity of speech acts
2.1 The history of a word shapes what can be done with it
Philosophers have spent a lot of energy trying to explain what makes a slur different from its neutral counterpart. “Boche” and “German” pick out the same people—so what’s the difference? Traditional approaches say there’s a difference in meaning.
Nowak draws on more recent thinking that says the difference is about the context and meaning around the word rather than in it.
Lepore and Stone (2018) argue that a slur means exactly what its neutral counterpart means. What makes it a slur is a metalinguistic fact: the social prohibition on using it and the “tone” which reflects its history of use.
Nunberg (2018) pushes this further with the concept of linguistic metadata. Words carry information about their history—who used them, in what contexts, with what attitudes—and this metadata is what distinguishes a slur from the neutral term. Racists don’t use slurs because they’re derogatory; slurs are derogatory because racists use them.
Díaz-Legaspe, Liu and Stainton (2019) arguie that competent users of slurring expressions will recognize various “register features” — not just whether something is a slur or not, but where it falls on scales of offensiveness, whether it’s regional, generational, and so on.
These three accounts converge on a shared insight: the history of a word—its metadata, pattern of use, and associations—shapes what you can do with it. Crucially, this is fine-grained and language-specific. If this is true of slurs, it’s true of everything. Every word carries metadata: history, register, phonetic associations, etc. What you do with a word depends on features beyond mere semantic content.
2.2 Different languages conventionalize different things
The closing paragraph does a great job of summing up the subsection: “In summary, languages differ widely in terms of the range of conventional tools they make available. We have looked at examples involving derivational morphology and scales of politeness, but different languages feature conventional devices that track a tremendous variety of different properties.24 Since differences in the inventory of conventional devices amount to differences in the space of possible modes of action, this means that different languages make a tremendous variety of different speech acts possible.” (Nowak, 2020, p. 844)
“Conventional tools” means the linguistic resources a language makes available to its speakers because the speakers of that language collectively agree on and practice them.
- The morphological operations a language allows (Spanish has augmentative suffixes; English doesn’t). Thus, in Spanish, -azo can be used to make toro (“bull”) into torazo (“great big bull” or “mighty bull”). English can translate the meaning but can’t reproduce the act of expanding the word-form.
- Grammaticalized politeness distinctions (e.g., Russian, French, Korean, etc.) force speakers to mark social positioning.
- Register and frequency markers: In Russian, certain words are immediately recognizable to educated speakers as belonging to Pushkin’s literary register — they’re archaic, elevated, and almost never appear in everyday speech. When Russian commentators mocked Putin’s decision to erect a giant statue of Saint Vladimir in Moscow, they described it using these distinctly Pushkinian words. The humor lives entirely in the word choices and the metadata they carry, which is why it’s untranslatable: English words for the same content don’t evoke Pushkin, so the joke simply doesn’t exist in English.
- Honorific systems, phonetic properties, lexical distinctions, etc.