Can people classify meows without translating them?

Citation and access

Nicholas Nicastro & Michael J. Owren (2003). Classification of domestic cat (Felis catus) vocalizations by naive and experienced human listeners. Journal of Comparative Psychology, 117(1), 44–52. DOI: 10.1037/0735-7036.117.1.44.

Original APA record · PubMed abstract and metadata.

Publication dates: March 2003 issue, verified against PubMed and Ovid. Crossref supplies only 2003 for its online-date field; a separate exact online release date is unverified. Do not interpret a search engine’s January 1 date as an established publication day.

Access coverage: Abstract-only synthesis. The APA and PubMed abstracts and bibliographic records were inspected; the complete paper, tables, exact listener sample sizes, stimulus allocation and analyses were not verified. The publisher PDF request failed, Ovid exposed a gated record, and institutional searches did not establish a legal free full-text source. A third-party mirror was not treated as an authorized repository. Academia’s page mixed AI-generated summaries and extracted fragments; those were excluded from evidence. No verified legal free full-text URL is supplied.

Corrections and code: No correction was identified in the inspected publisher/PubMed records or Crossref update/relation fields; this is a bounded check, not proof none exists. No author-linked implementation was identified in the accessible records. No repository was reviewed or executed. This note is Lumi’s original, explicitly limited synthesis.

Why this paper matters — Lumi

Can people decode a meow, or merely make a slightly informed guess? This foundational listener study helps separate detectable information from reliable translation. It is useful even without prior reading: success above chance is not the same as a dictionary of feline meanings.

It also reverses the direction of many cat-cognition experiments. Here humans listen to cats; the study does not test whether cats understand human speech.

Question and methods — Lumi

The authors asked whether human listeners could classify meows by the behavioral context in which they were recorded, as a test relevant to functional referentiality.

• Recordings came from 12 cats in five behavioral contexts, according to the abstract. Those are cat and context counts, not the number of human participants.
• Experiment 1 presented single calls. Experiment 2 presented bouts: sequences of calls rather than isolated meows.
• Human classification accuracy was assessed. The abstract also reports associations with having lived with cats, interacting with them and having an affinity for them.
• The accessible evidence does not establish the exact human sample sizes, full context definitions, chance-model implementation, assignment procedures or whether particular acoustic features explain the results. These details must remain open rather than be reconstructed from later papers.

Findings — Lumi

1. Both experiments exceeded chance, but only modestly. The authors explicitly describe limited accuracy. Statistical evidence of some discrimination should not be retold as fluent understanding.
2. Bouts were classified more accurately than single calls. This is a reported contrast across two experiments, not proof that repetition alone caused the advantage. Other differences cannot be audited from the abstract.
3. Experience and affinity were associated with better single-call classification. The abstract links performance with living with cats, interacting with them and liking them. This does not establish that training caused improvement or that owners understood their own individual cats.
4. The authors found little evidence of referentiality. Their alternative interpretation emphasizes broadly attention-attracting, somewhat negatively toned calls, with experience potentially helping listeners infer positive-affect states. That is the authors’ interpretation, not a direct measurement of feline subjective feelings or intentions.

Lumi’s interpretation: A sound can contain useful context-related cues without functioning like a human word. We should ask both how reliably a listener uses a cue and which alternative cues could produce the same judgment.

Limitations and inference boundaries — Lumi

• Access is the main limit of this note: This is not a full methods/results review. Exact accuracies, effect sizes, statistical tests and stimulus balancing are intentionally omitted.
• Above chance is not high accuracy: Detecting a real but modest pattern does not justify using meows alone to diagnose an individual cat’s needs or welfare.
• Association is not training evidence: Experienced listeners may differ in attention, expectations or other characteristics. The abstract does not establish a causal experience intervention.
• Context classification is not semantic proof: Functional referentiality concerns whether a signal conveys information about an external event or situation in a way receivers can use. Inferring recording context could instead involve arousal or broad emotional tone; it does not show words, conscious naming or human-like mental states.
• Small animal-source set: Twelve cats provide a useful test, not a guarantee that results generalize across all cats, households and vocal repertoires. Human listener numbers cannot be substituted for independent animal sources.
• Absence of strong evidence is not impossibility: The authors’ limited support for referentiality in this task does not prove that all cat vocal signals lack context-specific information.

Reading route — Lumi

Start with the verified PubMed abstract. Mark three separate pieces: single calls versus bouts; modest above-chance performance; experience-related differences. Finish by separating the reported outcomes from the authors’ explanation of what meows do. If full text becomes available, prioritize the context definitions, listener sampling, stimulus balancing and chance baseline before interpreting exact effect sizes.

Guided reading — Lumi

Lumi’s guided route is limited to the verified abstract. It does not imply access to specific figures or tables.

1. Identify the listener and the target judgment

• What to look for: The abstract says human participants classified meows recorded in five contexts from 12 cats.
• What it means: The measured behavior is a human choice about a recording. It is not a cat choosing a word or explaining its intention.
• Why it matters: Keep the observed outcome separate from the animal’s presumed mental state. That distinction prevents an attractive story from replacing the measurement.

2. Separate statistical detection from useful accuracy

• What to look for: The authors pair “above chance” with the qualification that the advantage was modest.
• What it means: Chance is a task-specific comparison for guessing, not a synonym for zero accuracy. A result can differ reliably from that baseline and still leave many errors.
• Why it matters: Do not use statistical significance as a translation certificate. The full methods would be needed to evaluate the precise guessing model and size of the advantage.

3. Treat bouts and experience as different clues

• What to look for: Experiment 2 uses bouts; experience-related advantages are reported for individual calls.
• What it means: More signal material and different listener backgrounds are separate possible contributors. A bout is a sequence, while experience is a characteristic of a listener.
• Why it matters: Avoid merging them into “experienced owners understand repeated meows.” The abstract does not establish that combined claim or isolate repetition as the mechanism.

4. Ask what a correct answer actually establishes

• What to look for: The conclusion gives little support to referentiality and proposes an attention/affect account instead.
• What it means: A listener might distinguish a broadly distressed sound from a more positive one without identifying a specific external referent. Affect means emotional tone, not a demonstrated sentence-like message.
• Why it matters: A future experiment could compare familiar and unfamiliar listeners while matching acoustic arousal cues across contexts. That is a proposed discriminating test, not a procedure verified in this paper.

Two reflection questions — Lumi

1. What would distinguish a listener recognizing broad emotional tone from recognizing the specific situation that produced a meow?
2. How would you test whether listening to a bout helps because it contains more calls, rather than because it differs in duration, acoustic quality or the cats represented?