Scalar implicatures: working memory and a comparison with only

Paul Marty, Emmanuel ChemlaFrontiers in Psychology 2013 · DOI

Abstract

A Scalar Implicature (SI) arises when the use of a relatively weak sentence (e.g., some politicians are corrupt) implies the denial of an alternative, stronger sentence (e.g., not all politicians are corrupt). The cognitive effort associated with the processing of SIs involves central memory resources (De Neys and Schaeken, 2007; Dieussaert et al., 2011; Marty et al., 2013). The goal of this study is to locate this previous result within the current psycholinguistic debate, and to understand at which level of SI processing these resources are specifically involved. Using a dual-task approach, we show that (1) tapping participant's memory resources interferes with the derivation of SIs, whereas (2) it does not affect the interpretation of sentences involving similar competition mechanisms between a sentence and potential alternatives through the use of only (e.g., only some politicians are corrupt). We explain how these findings suggest that the central memory resources are not involved in the core process at the source of SIs, and discuss how this difference between SIs and only bears on recent linguistic debates on the division of labor between grammar and pragmatics.

Introduction: The Locus of Pragmatic Effort

In linguistics, a Scalar Implicature (SI) occurs when a speaker uses a weak term (like "some") to imply the negation of a stronger term ("all"). For example, while the sentence "Some snakes are reptiles" is logically true (since all snakes are indeed reptiles), most listeners find it "odd" or even "false" because they infer it means "some but not all."

Previous research has established that deriving these SIs is cognitively taxing, requiring more time and memory than literal interpretations. However, a major debate remained regarding the exact source of this cost. Is the effort spent on the Pragmatic Decision (deciding whether to move beyond the literal meaning) or the Mechanical Computation (generating and negating alternative sentences like "All snakes are reptiles")?

This study addresses this question by comparing implicit SIs to sentences containing the word only (e.g., "Only some snakes are reptiles"). Both require the same mechanical computation of alternatives, but only makes the "not all" meaning mandatory and explicit, effectively removing the pragmatic decision step.

Experimental Design: The Dual-Task Paradigm

The researchers employed a dual-task paradigm with 16 native French speakers to observe how competing cognitive demands affect language processing. Participants were required to perform two tasks simultaneously:

  1. Dot Memory Task: Participants had to remember a visual pattern of dots in a 3x3 grid. The difficulty was manipulated through two conditions: Low Load (3 dots in a simple line) and High Load (4 dots in a complex, non-linear arrangement).
  2. Truth-Value Judgment Task: While holding the dot pattern in memory, participants judged categorical sentences as True or False based on general knowledge.
Figure 1: Examples of dot patterns used to manipulate cognitive load: (A) Low Load and (B) High Load.

The critical comparison involved "Some-total" sentences (e.g., "Some snakes are reptiles"), which are True literally but False if an SI is derived, and "Only-total" sentences (e.g., "Only some snakes are reptiles"), which are False by definition.

Key Findings: Implicit vs. Explicit Constraints

The results revealed a clear distinction between implicit inferences and explicit grammatical constraints:

  • Memory Load Impairs SIs: Under high memory load, participants derived significantly fewer scalar implicatures. They gave "False" responses to "Some snakes are reptiles" only 34% of the time under high load, compared to 54% under low load. When cognitive resources were scarce, participants defaulted to the literal "at least one" meaning.
  • "Only" is Robust: Crucially, memory load had no effect on sentences containing only. Participants correctly identified "Only some snakes are reptiles" as False roughly 75-79% of the time, regardless of the difficulty of the memory task.

The statistically significant interaction between Load and Sentence Type confirms that the cognitive "tax" is specific to the implicit pragmatic inference rather than the mechanical logic of alternatives.

Discussion: Grammar vs. Pragmatics

These findings suggest a clear "division of labor" in language processing. The grammar (represented by the word only) handles the computation of alternatives efficiently and automatically. In contrast, pragmatics (the derivation of SIs) is sensitive to central memory resources because it involves a disambiguation choice—the decision to apply an inference when it isn't explicitly required.

This challenges the "Silent Operator" hypothesis, which suggests that SIs are handled by a hidden version of only in the brain. If they were processed identically, both should have been equally affected by memory load. Instead, the results indicate that the "heavy lifting" of language is not the logic itself, but the pragmatic decision to use it.

Limitations and Future Directions

While the results are compelling, the study's small sample size (16 participants) suggests a need for replication with a larger cohort. Additionally, the categorical sentences used are somewhat artificial; future research could explore whether these memory costs persist in naturalistic conversations with clearer communicative goals. Finally, investigating other "silent" operators, such as the covert version of "even," could further clarify the boundaries between grammar and pragmatics.

Summary updated

Abstract

A Scalar Implicature (SI) arises when the use of a relatively weak sentence (e.g., some politicians are corrupt) implies the denial of an alternative, stronger sentence (e.g., not all politicians are corrupt). The cognitive effort associated with the processing of SIs involves central memory resources (De Neys and Schaeken, 2007; Dieussaert et al., 2011; Marty et al., 2013). The goal of this study is to locate this previous result within the current psycholinguistic debate, and to understand at which level of SI processing these resources are specifically involved. Using a dual-task approach, we show that (1) tapping participant's memory resources interferes with the derivation of SIs, whereas (2) it does not affect the interpretation of sentences involving similar competition mechanisms between a sentence and potential alternatives through the use of only (e.g., only some politicians are corrupt). We explain how these findings suggest that the central memory resources are not involved in the core process at the source of SIs, and discuss how this difference between SIs and only bears on recent linguistic debates on the division of labor between grammar and pragmatics.
BibTeX
@inproceedings{marty2013scalar,
  author = {Paul Marty and Emmanuel Chemla},
  title = {{Scalar implicatures: working memory and a comparison with only}},
  booktitle = {Frontiers in Psychology},
  year = {2013},
  doi = {10.3389/fpsyg.2013.00403},
}
Figure