Questions & explanations
1. Compare the Rational Speech Act model with a simpler literal model for interpreting 'some'.
A literal model interprets 'some' as meaning 'at least one,' with no implicature. So if a speaker says 'Some students passed,' the literal model would consider it true even if all passed. The RSA model, however, reasons that the speaker could have said 'all' but chose 'some,' so the listener infers that not all passed. The RSA model thus derives the implicature, while the literal model does not. In experiments, human listeners often behave like the RSA model, showing that computational pragmatics better captures real interpretation. The literal model is simpler but fails to explain why people often reject 'some' when all are true.
2. What is the role of 'felicity conditions' in Searle's analysis of indirect speech acts?
Felicity conditions are the necessary conditions for a speech act to be successful. For an indirect request, the literal question (secondary act) must have its own felicity conditions (e.g., the speaker genuinely doesn't know the answer). But the primary request also has conditions (e.g., the hearer is able to do the action). The indirectness works because the literal question often implies that the preparatory conditions for the request are met. For example, asking 'Can you do X?' implies that the hearer is able, which is a felicity condition for a request. So the literal act helps satisfy the conditions for the primary act.
3. What is the main difference between high-context and low-context cultures in communication?
High-context cultures rely heavily on shared background and nonverbal cues, so much meaning is implied rather than stated directly. Low-context cultures prefer explicit, clear verbal messages where little is left to implication. For example, in a high-context culture like Japan, a simple 'maybe' can mean 'no' based on the situation. In a low-context culture like Germany, people usually say 'no' directly. This difference affects how implicatures—meanings that are hinted but not said—are understood. A speaker from a high-context culture may expect the listener to infer more, while a low-context listener may miss the hint.
4. How does the Rational Speech Act (RSA) model resolve implicatures?
The Rational Speech Act model treats communication as a game where speakers choose words to convey meaning efficiently, and listeners infer the intended meaning by reasoning about the speaker's choices. For implicatures, the model uses probabilities: a speaker says 'some' when they want to convey 'some but not all' because if they meant 'all,' they would have said 'all.' The listener then reasons: if the speaker said 'some,' they probably didn't mean 'all.' This Bayesian reasoning allows the model to compute the implicature that 'some' implies 'not all.' RSA has been successful in predicting human implicature judgments.
5. Compare the grammaticalization of 'this' and 'that' into articles: which one typically becomes a definite article and which an indefinite article?
Typically, a distal demonstrative like 'that' becomes a definite article, while a proximal demonstrative like 'this' often becomes an indefinite article or a presentative marker. For example, in English, 'that' is the source of the definite article 'the' in some historical stages, though English actually uses a different word. In many languages, the distal form marks something already known (definite), whereas the proximal form introduces new information (indefinite). This pattern is seen in the development from Latin 'ille' (that) to definite articles, and from 'ipse' (self/same) to some indefinite markers.
6. What is a speech act?
A speech act is an utterance that performs an action, like making a request, giving a promise, or apologizing. For example, saying 'I promise to come' is not just stating something; it is actually making a promise. Speech acts have three parts: the literal meaning (locution), the speaker's intention (illocution), and the effect on the listener (perlocution). In discourse, we analyze how these acts fit together in conversation. A single sentence can be a speech act, like 'Could you pass the salt?' which is a request, not a question. Understanding speech acts helps us see how language does things in the world.
7. Why might a researcher choose a picture selection task over a truth-value judgment task for testing children's implicatures?
Picture selection tasks are often easier for children because they don't require verbal responses or complex judgments. Children can simply point to a picture. Truth-value judgment tasks require understanding 'true' and 'false,' which may be abstract for young children. Also, picture selection allows children to choose among alternatives, which can show their interpretation more clearly. For example, a child might pick a picture showing only some apples red when hearing 'Some apples are red,' indicating they understand the implicature. This method reduces language demands and is more engaging for children.
8. What is a limitation of current computational models of implicature resolution?
Current models often assume that speakers and listeners are perfectly rational and have complete knowledge of each other's goals. In reality, people have limited cognitive resources and may not always reason optimally. Models also struggle with complex social factors like politeness or irony, which affect implicatures. For example, a sarcastic 'Great job' implies the opposite, but models may not capture the tone. Additionally, models are usually tested on simple sentences like 'some' vs. 'all,' but real-world implicatures are more varied. Improving models to handle natural conversation remains a challenge.
9. How does a word like 'that' change over time to become a grammatical marker like 'the'?
In many languages, a demonstrative like 'that' can weaken over time and become a definite article. For example, in Latin, the word 'ille' (that) evolved into the definite articles in Romance languages like French 'le' and Spanish 'el'. This happens because speakers start using the demonstrative in contexts where pointing is not needed, just to show the listener knows which thing is meant. Over centuries, the pointing meaning fades, and the word becomes a grammatical marker. This process is called grammaticalization. It shows how deictic words can lose their original force and take on new grammatical jobs.
10. What are the two conditions for a stimulus to be optimally relevant?
A stimulus is optimally relevant if (a) it is relevant enough to be worth the hearer's effort to process it, and (b) it is the most relevant one the speaker could have given, given her abilities and preferences. Condition (a) ensures the hearer gets enough benefit to justify paying attention. Condition (b) means the speaker chooses the best way to communicate, considering what she can do and what she wants. Together, they explain how speakers design utterances to be efficient. For example, if I say 'It's raining,' it's relevant enough to tell you to take an umbrella, and it's the simplest way to say it.
11. Give an example of a computational model that uses game theory to resolve implicatures.
One example is the Game-Theoretic Pragmatics model. In this model, a speaker and listener play a coordination game. The speaker chooses an utterance to achieve a goal, and the listener interprets it based on what the speaker would likely say. For instance, if the speaker says 'Some of the lights are on,' the listener considers that if all lights were on, the speaker would have said 'all.' So the listener infers that not all are on. The model uses concepts like optimality and equilibrium to predict which implicature will be derived. This approach formalizes how context and speaker goals affect meaning.
12. Compare truth-value judgment and picture selection tasks for studying implicatures.
Both tasks test how people interpret sentences with implicatures, but they differ in format. In truth-value judgment, participants say true or false to a sentence about a single scenario. In picture selection, they choose among multiple pictures. Truth-value judgment is simpler and directly tests truth assessment, while picture selection requires matching a sentence to a visual scene. Picture selection can test more complex scenarios with multiple options. Both can reveal whether participants compute implicatures, but picture selection may be better for children or non-readers because it uses images.