Probing cross-linguistic differences in time representations using arithmetic word problems - Archive ouverte HAL
Pré-Publication, Document De Travail Année : 2024

Probing cross-linguistic differences in time representations using arithmetic word problems

Résumé

This project sets out to investigate whether the language we speak influences the way we think – also known as the Sapir-Whorf hypothesis of linguistic relativism (Whorf, 1939). More specifically, we aim to build on previous findings which suggest that time perception may differ depending on the language spoken (Casasanto et al., 2004, 2007). Does language shape our perceptions about the world around us? Does language limit or expand our reasoning? Starting from Whorf’s contributions in the first half of the 20th century, all the way to the works of Casasanto, Boroditsky and their peers at the onset of the 21st century, the field of linguistic relativism has seen its fair share of breakthroughs and setbacks. Through this study, we plan to bring together two distinct research fields- linguistic relativism and mathematical reasoning. The Whorfian hypothesis suggests that the way a language is structured and the words it contains can affect how people see and understand the world around them (Whorf, 1939). For example, languages that have specific words for certain concepts might encourage speakers to think more deeply about those aspects of reality and impact how they mentally perceive those concepts. Over the 20th century, researchers from different fields sought to back the Whorfian hypothesis. For instance, Brown and Lenneberg (1954) studied colour perception, Miller and Stigler (1987) looked into how number systems vary across languages, and Choi and Bowerman (1991) explored the differences between languages and their potential impact. However, counter-evidence to the Whorfian hypothesis emerged in Heider's (1972) findings on universal colour perception. Pinker (1994) further challenged the Whorfian hypothesis, expressing strong scepticism towards experiments testing simplistic versions of the hypothesis. Similarly, Devitt and Sterelny (1999) argued that changes in the social environment do tend to influence one’s vocabulary and language usage, without necessarily impacting thought processes. Pullum (1991) supported these arguments while rejecting the exaggerated claims regarding the number of words for snow in Inuit and Yupik languages. Simultaneously he stressed the need for strong empirical evidence. He also argued that such linguistic variations even when they do exist would have minimal impact on thought. As the 21st century unfolded, researchers began embracing a nuanced interpretation or a ‘weaker version’ of the Whorfian hypothesis - linguistic relativism. This perspective removes the ‘deterministic/constraining’ element of the hypothesis and suggests that while language may not entirely determine thought, variations among languages could indeed influence differences in the thoughts of speakers to a certain degree (Lucy, 1997). For instance, Casasanto et al. (2004) suggest that English and Indonesian speakers tend to use distance-related expressions to talk about time (e.g. ‘long’ time), while Spanish and Greek speakers tend to use amount-related expressions instead (e.g. ‘much/lot of’ time) and found evidence supporting linguistic influence on time perception. Using arithmetic word problems on durations, we intend to evaluate whether these cross-linguistic differences in time perception impact the way individuals calculate durations, further probing the scope of linguistic relativism. In former studies involving French and English speakers, Gros et al. (2021) examined arithmetic word problems with a common mathematical structure capable of accommodating both 1-step and 3-step solving strategies. These problems aimed to elicit different solving strategies based on mental encodings formed by participants, depending on the problem type (such as durations, heights, collections, prices, etc.). It was shown that duration problems elicit an ordinal encoding, favouring a 1-step solving strategy, whereas problems involving collections, prices, or weights prompted a cardinal encoding, leading to a 3-step solving strategy. English and French are both languages which predominantly use distance-related expressions to express time (long time/ longtemps). This raises the question of whether replicating this similar ordinal encoding for duration problems might pose a challenge in languages using amount-related expressions to talk about time. Building on this bridge between linguistic relativism and mathematical reasoning, we propose to contrast the performance of native speakers of 4 different languages on arithmetic word problems involving durations. We seek to examine whether the use of distance-related or amount-related expressions concerning time in one’s native language influences their choice of a specific strategy to solve arithmetic word problems involving durations. We will examine two contrasting perspectives: one advocating linguistic relativism (the relativists) and the other challenging the notion that language significantly shapes thought (the universalists). Martin (1986) and Pullum (1991) emphasize the necessity of robust empirical evidence in research, citing the infamous vocabulary hoax surrounding words for snow as an example and cautioning against the dissemination of misinformation. Pinker’s (1994) critique highlights the scarcity of scientific evidence and underscores the importance of seeking evidence through non-linguistic tasks, as suggested by Gentner & Goldin-Meadow (2003). Motivated by these viewpoints, this experiment employs arithmetic word problems to discern evidence that corroborates either of the perspectives (the relativists and the universalists).
Fichier non déposé

Dates et versions

hal-04678539 , version 1 (27-08-2024)

Identifiants

Citer

Ann Mary John, Mônica Macedo-Rouet, Hippolyte Gros. Probing cross-linguistic differences in time representations using arithmetic word problems. 2024. ⟨hal-04678539⟩
5 Consultations
0 Téléchargements

Altmetric

Partager

More