If I were to offer you some of my cake, how much would you assume you could take? A slim slice? A generous wedge? Anything short of the whole thing? That tiny judgement call sits at the heart of a new study from the University of Florence, published in Quaderni di Linguistica e Studi Orientali, exploring why we sometimes infer “not all” from some – and why sometimes we don’t.
The study
The researchers worked with adult participants who completed tasks designed to trigger scalar implicatures — those subtle inferences we make when interpreting quantifiers like some, many or all. Participants read short scenarios & judged whether the statements they saw were fully accurate.
They also completed a set of executive function tasks that reveal how people manage competing information. These tasks distinguish between two broad cognitive control styles:
- a context‑maintaining style, where someone holds on to the current interpretation even when distractions appear
- a reactive style, where someone shifts attention quickly in response to new cues
Alongside this, the researchers checked how clearly participants identified the “question under discussion” (QUD) — the implicit question they believed the speaker was answering. For example, when hearing “Some students passed”, did they think the speaker was addressing “How many passed?” or simply “Did anyone pass?” That distinction turned out to matter a lot.
The findings
Participants with a context‑maintaining control style were more likely to derive scalar implicatures. They tended to stick with the idea that some contrasts with all, making the “not all” inference more consistently.
This effect became even stronger when participants correctly identified the QUD.
If they assumed the speaker was answering a quantity‑based question (“How many passed?”), they were far more likely to infer “not all”.
If they assumed the speaker was answering a yes/no question (“Did anyone pass?”), the implicature often disappeared.
Those with more flexible, reactive control strategies were less consistent in drawing these inferences. The authors argue that implicature derivation depends on both cognitive stability & pragmatic awareness.
This fits neatly with broader pragmatic theory: Grice’s maxims, relevance theory & work showing that implicatures arise from shared assumptions about intent & relevance. It also echoes cognitive research suggesting that maintaining a stable interpretive frame supports deeper inferencing.
To illustrate this with an example, imagine a colleague says:
“We covered some of the tasks in the workshop.”
If you assume the QUD is “How much did you cover?”, your mind naturally treats some as part of a scale (some → many → all).
In that frame, some contrasts with all, so you’re likely to infer “not all”.
If you assume the QUD is “Did you cover anything?”, you’re not thinking about quantity at all. You’re simply checking whether the answer is yes or no.
In that frame, some just means “yes, at least one”, so there’s no pressure to infer “not all”.
In other words, the difference isn’t in the sentence – it’s in the question you think the speaker is answering. That’s exactly what the study shows: implicatures aren’t automatic; they depend on the listener’s interpretation of what’s relevant.
It’s a small example, but it captures something central to language use: learners aren’t just decoding words, they’re inferring intent. Even if we don’t teach QUDs explicitly, being aware of how these hidden questions shape interpretation can sharpen the way we design tasks & discuss meaning.
Teacher takeaways?
- When teaching quantifiers, ask learners to identify the underlying question: “What is this sentence really answering?”
- Use short dialogues to contrast different QUDs & show how meaning shifts subtly.
- Treat implicatures as expectations rather than rules – help learners see that some may or may not imply “not all”, depending on what feels relevant in the moment.
This study is a reminder that meaning-making is a cognitive–pragmatic partnership. It’s not just what is said but the silent question we think we’re answering that shapes interpretation.
Where do you see learners struggling most with these kinds of inferences?
Leave a Reply