Home / A. Sentience and moral status / Part DIR
"Animals don't have language."
language is a means of communicating experiences, not a condition for having them. Those who can't speak can feel pain, many animals communicate pain and fear in ways science already knows how to measure and, taken seriously, the argument would exclude babies and people who have lost speech from moral protection.
Short answer
Language is for telling what we feel; it isn't what makes us feel. A baby doesn't talk and feels pain; someone who has lost speech after a stroke still feels everything. Besides, animals communicate pain: scientists created scales that measure pain in mice from their facial expressions, and similar scales are now used for other species, precisely to decide when to give painkillers.
For children
When a baby cries, they can't talk yet, but the family understands they're hungry or in pain. Animals also let us know when something's wrong: a dog whimpers, a cat hides, a cow moos calling for her calf. Not being able to talk like us doesn't mean not feeling. It means we need to pay more attention.
"If a person could no longer speak because of a stroke, would it be acceptable to hurt them?"
Como pode aparecer
"They can't talk."
"If they suffered, they'd complain."
"You don't know what the cow feels; she can't tell you."
"Show me a chicken explaining what she feels."
"Mooing isn't language."
"How do you know a fish feels pain if it doesn't even scream?"
"Without language there are no beliefs; without beliefs there are no desires or interests."
"Attributing mental states to those who can't speak is unverifiable projection."
"Animals have no voice in this debate." (a truth that increases our responsibility rather than reducing it)
The strongest version
"Animals don't have language, so you can't say they suffer or have interests."
Strongest version (steelman). The most philosophically rigorous formulation is R. G. Frey's, in Interests and Rights: The Case Against Animals (1980), supported by Donald Davidson's thesis (1982) that there's no thought without language:
- P1Having interests requires having desires.
- P2Having desires requires having beliefs (desiring something requires believing it isn't yet the case).
- P3Having beliefs requires language (to believe is to hold a sentence true).
- P4Animals don't have language.
- C1So animals don't have interests.
- P5Only those with interests can be morally harmed.
- C2So we don't harm animals by using them.
There's also an epistemological version, more common day to day: "without language, we have no way of knowing what they feel". It doesn't deny that animals feel, but denies we can know it, and concludes we needn't take that feeling into account.
Frey's version is valid. It fails on soundness: P2 and P3 are false or unjustified, P4 depends on how language is defined, and P1 ignores that pain harms without needing beliefs. The epistemological version makes another mistake: it turns difficulty of knowing into license to ignore.
Full answer
- a
Reconstruction and diagnosis
As in A1 and A2, one word carries the argument: "language". It can be understood in a broad sense (communication with signals that carry information) or a strict sense (recursive grammar, able to produce infinitely many new sentences, like human languages).
Sense of "language" What happens to P4 What happens to P3 Broad: meaningful communication False: many animals communicate about predators, food, pain and fear Loses its force: animals would have enough "language" to have beliefs Strict: language with recursive grammar True of all animals studied Excludes babies and people with severe aphasia, and Frey himself admits this The second sense is the one Frey uses, and he accepts the consequence: he saw no reason to attribute beliefs to any creature without language, "including human infants". So the argument doesn't separate humans from animals. It separates speakers from non-speakers.
- b
Where the argument breaks
1. Pain harms without needing beliefs (P1 and P2).
The desire-plus-belief model describes complex preferences, like wanting a promotion. But feeling pain doesn't require believing the sentence "I'm in pain". Pain is bad for whoever feels it simply by being felt. That's the core of sentientism: the interest in not suffering arises from the experience itself, not from a propositional attitude about it.
2. Beliefs don't require sentences (P3).
The thesis that believing is "holding a sentence true" is a particular philosophical theory, not a consensus. Cognitive science routinely attributes non-linguistic representations to babies and animals: expectations about objects, maps of space, memory of where food is. A chicken who chooses the feeder with painkiller when lame (see c) acts like someone who "knows" something about her own state, without needing words.
3. Species overlap, again.
Babies don't have language. People with global aphasia, usually after a stroke, lose the ability to produce and understand language, but remain conscious, feel pain and have emotions. If language were a condition for having interests, these people would lose them the moment the stroke struck. No one accepts that.
4. The empirical premise is more complicated than it looks (P4).
No animal studied has a language with the grammar and productivity of human languages, and it would be dishonest to say otherwise. But animal communication is rich: vervet monkeys give different alarms for leopards, eagles and snakes; Japanese tits combine calls in ways where order changes the response; and pain is communicated through facial expressions researchers have learned to measure.
5. The epistemological version proves too much.
We never have direct access to anyone's mind, not even other humans'. We infer mental states from behavior, physiology and similarity. Speech itself is a behavior, and it can lie. In the case of animals, the inference rests on several converging lines of evidence: pain receptors, opioid systems similar to ours, guarding of the injured part, learning to avoid what hurts, willingness to pay costs to avoid pain and active seeking of painkillers (Sneddon et al., 2014). Faced with a realistic possibility of sentience, the New York Declaration (2024) concludes it's irresponsible to ignore it. Uncertainty, when suffering is at stake and the cost of caution is low, calls for more care, not less.
- c
Data and sources
Finding Source Reliability Use in the argument Mice show facial expressions of pain codable on a standardized scale, with high accuracy and reliability Langford et al., 2010 High; the method was later adapted for other species Animal pain is communicated and measurable without language Lame chickens ate more feed containing the painkiller carprofen than healthy chickens Danbury et al., 2000 Moderate Behavior seeking pain relief Criteria for assessing pain in animals: response to painkillers, guarding of wounds, avoidance learning, costs paid to avoid pain Sneddon et al., 2014 High (reference review) A framework for inferring pain without verbal report Vervet monkeys react differently to alarms for leopard, eagle and snake Seyfarth, Cheney and Marler, 1980 Moderate: a reassessment by Price et al., 2015 confirmed the acoustic distinction, but suggests the calls also express emotional states, not "words" Communication carrying information, without being human language Japanese tits respond differently depending on the order of two combined calls Suzuki, Wheatcroft and Griesser, 2016 Moderate and debated (critical comments have been published) Combining signals with an order effect Projects teaching language to great apes (Koko, Kanzi and others) Popular and academic literature Low for the claim of full "language" Don't use as proof - d
The strongest counter-reply: consciousness requires thinking about one's own thinking
Besides Frey, there's a more sophisticated reply, from higher-order theories of consciousness. According to them, an experience is conscious only if the subject has a thought about that experience. Peter Carruthers (1989) used this idea to suggest that if animals lack such thoughts, their pains might not be conscious: there'd be reaction to pain without anyone feeling it.
- e
Why the counter-reply also fails
It proves too much.
Human babies also lack, in their first months, thoughts about their own thoughts. The same theory would imply their pains aren't conscious, the belief that for decades justified operating on newborns with minimal anesthesia (see A1, block 10).
It's a minority, disputed theory.
Higher-order theories aren't a consensus in consciousness science, and much of affective neuroscience locates the basic experience of pain and fear in ancient structures shared with other vertebrates. The New York Declaration, signed by hundreds of researchers, points the opposite way.
It violates parsimony.
Explaining pain receptors, opioid chemistry like ours, guarding of wounds and seeking painkillers by postulating "pain no one feels" only in those who can't speak requires a tailor-made exception. The simplest hypothesis is that someone is feeling.
Even the author backed away from the practical consequences.
In later work (Carruthers, 2004), he suggested that even without consciousness in his theory's sense, the frustration of an animal's desires could be an appropriate object of moral concern.
The cost of error is asymmetric.
If we treat as sentient animals that may not feel, we lose little. If we treat as insensible animals that do feel, we cause suffering on an immense scale. Under uncertainty, the rational choice is caution.
Reasoning errors
Logical
Is the form valid?
Equivocation
Non sequitur
Proving too much
Confusing medium and content
Epistemological
Does the evidence support it?
Appeal to ignorance
Selective skepticism
Demanding impossible evidence
Ignoring the asymmetry of risk
Moral
Is the criterion coherent?
Irrelevant criterion
Exploiting vulnerability
Biases
Motivated skepticism
Much more evidence is demanded to accept an unwelcome conclusion (animals suffer) than a welcome one
Ditto and Lopez, 1992
Motivated denial of mind
Less experience is attributed to animals used as food
Bastian et al., 2012; replicated by Jacobs et al., 2024
Two-dimensional mind perception
People perceive minds along two dimensions, the capacity to act and the capacity to feel; lacking speech lowers the first, but the argument uses it to deny the second
Communicative anthropocentrism
Pain is expected to be expressed as in humans (a scream, a word); silent species, such as fish, seem not to feel
The basis of argument A6
Voiceless-victim bias
Victims who can't complain get less attention and less empathy
A known pattern regarding vulnerable groups
Extrapolations
The principle being tested is: "whoever lacks language has no interests we need to respect".
If it applied to…BabiesTap to see
Hurting them wouldn't be a harm. A baby doesn't speak or understand sentences; if interests depend on language, their crying would be mere noise, with no one being harmed
Frey himself extends his thesis to human infants. In medical practice, the belief that newborns didn't really feel pain led to surgery with minimal anesthesia until the 1980s
If it applied to…People with global aphasiaTap to see
They'd lose their interests the moment the stroke struck. Since they stop producing and understanding language, by the criterion they'd also stop being able to be harmed
Clinically, they remain conscious, feel pain and have emotions
If it applied to…Patients who can't report painTap to see
Their pain could be ignored. People with advanced dementia or without verbal communication can't say they're in pain; the argument would remove the obligation to treat it
Medicine does the opposite: it uses behavioral scales, such as PAINAD for advanced dementia, because it knows pain exists without report
If it applied to…Those who don't speak our languageTap to see
They'd be treated as less able to think and feel. If the language we recognize is the measure of mind, those who communicate differently seem to have less mind
The Greek word bárbaros arose to describe peoples whose speech sounded incomprehensible, and became a synonym for inferior and uncivilized
If it applied to…Artificial intelligenceTap to see
A system that produces fluent language would have more interests and more moral status than a baby. If language is the criterion, verbal fluency would count for more than the capacity to feel
Language models show that producing sentences isn't, by itself, evidence of sentience. The criterion fails both ways: it excludes those who feel and includes those who merely talk
Contradictions
Takes the animal to the vet because "he's in pain", a conclusion drawn from exactly the kind of evidence the argument says is insufficient.
No one expects a newborn to explain what they feel before believing they feel it.
New pain medicines are tested on animals precisely because they're assumed to feel pain much as we do. You can't use that similarity in the lab and deny it on the plate.
Says "there's no way to know what they feel" and then acts with certainty that they don't suffer.
What the argument gets right
What is true
Language vastly enlarges mental life. With it, we plan the future, build narratives about ourselves, anticipate suffering and are hit by harms that exist only for linguistic beings, such as verbal humiliation or broken promises. Language is also a condition for full moral agency and political participation: animals can't represent themselves, argue or vote. Finally, verbal report is a valuable source of evidence, and without it assessing pain is genuinely harder and more error-prone.
But
But each of these points leads to the opposite of the argument's conclusion. If it's harder to know what they feel, we need more care, not less. If they can't represent themselves, someone needs to speak for them. Lack of a voice doesn't reduce the responsibility of those who have one: it increases it.
Connections
Usually comes before
Usually comes after
Go deeper in Questão de Ética
Chapters from the open-access collection A Journey Through Animal Ethics that discuss this argument in more detail:
- Could the Possession of Capacities or Relationships… Vol. I, ch. 4
References
Provisional author-date format. The links indicate the sources checked in writing this chapter.
Philosophy
- Brute experience. *The Journal of Philosophy*, 86(5).
- Suffering without subjectivity. *Philosophical Studies*, 121.
- Rational animals. Dialectica, 36(4).
- Discourse on Method, part V.
- *Interests and Rights: The Case Against Animals*. Clarendon Press.
- The New York Declaration on Animal Consciousness (2024).
Animal science
- Self-selection of the analgesic drug carprofen by lame broiler chickens. *Veterinary Record*, 146(11).
- Coding of facial expressions of pain in the laboratory mouse. *Nature Methods*, 7.
- Vervets revisited: a quantitative analysis of alarm call structure and context specificity. *Scientific Reports*, 5.
- Monkey responses to three different alarm calls. *Science*, 210.
- Defining and assessing animal pain. *Animal Behaviour*, 97.
- Experimental evidence for compositional syntax in bird calls. *Nature Communications*, 7.
Psychology and medicine
- Pain and its effects in the human neonate and fetus. New England Journal of Medicine, 317.
- Motivated skepticism. Journal of Personality and Social Psychology, 63(4).
- Dimensions of mind perception. *Science*, 315.
- Replication registered report of Bastian et al. (2012). *International Review of Social Psychology*.
- Development and psychometric evaluation of the Pain Assessment in Advanced Dementia (PAINAD) scale. Journal of the American Medical Directors Association, 4(1).