Department of Linguistics, University of Pennsylvania.
Cogn Sci. 2020 Apr;44(4):e12829. doi: 10.1111/cogs.12829.
Individuals shift their language to converge with interlocutors. Recent work has suggested that convergence can target not only observed but also expected linguistic behavior, cued by social information. However, it remains uncertain how expectations and observed behavior interact, particularly when they contradict each other. We investigated this using a cooperative map task experiment, in which pairs of participants communicated online by typing messages to each other in a miniature "alien" language that exhibited variation between alien species. The overall task comprised three phases, in each of which participants were told that they would be paired with a different partner. One member of the pair was given explicit linguistic expectations in each phase, while the software controlled whether or not observed behavior from their partner would be consistent or inconsistent with these expectations. The other participant was given no such expectations, allowing us to control for the role of expectation. Participants converged to both observed and expected linguistic behavior, and convergence was boosted when observation and expectation were aligned. When expected and observed behavior were misaligned, participants updated their expectations, though convergence levels did not drop. Furthermore, participants generalized what they learned about one partner to apparent novel partners of the same alien species. We also discuss individual variation in convergence patterns and the lack of a relationship between linguistic convergence and success at the map task. Findings are consistent with observations outside the laboratory that language users converge toward expected linguistic behavior. They also have broader implications for understanding linguistic accommodation and the influence of social information on linguistic processing and production.
个体调整自己的语言以与对话者趋同。最近的研究表明,趋同不仅可以针对观察到的语言行为,还可以针对社会信息提示的预期语言行为。然而,目前还不确定期望和观察到的行为如何相互作用,特别是当它们相互矛盾时。我们使用合作地图任务实验研究了这一点,在该实验中,参与者通过在线输入彼此的消息,在一个微型的“外星”语言中进行交流,该语言在不同的外星物种之间存在差异。整个任务由三个阶段组成,在每个阶段,参与者都被告知将与不同的伙伴配对。在每个阶段,其中一个参与者都被给予明确的语言期望,而软件则控制他们的伙伴的观察行为是否与这些期望一致或不一致。另一个参与者则没有得到这样的期望,使我们能够控制期望的作用。参与者既与观察到的语言行为趋同,也与预期的语言行为趋同,当观察和期望一致时,趋同程度会提高。当预期和观察到的行为不一致时,参与者会更新他们的期望,尽管趋同水平不会下降。此外,参与者将他们从一个伙伴那里学到的东西推广到相同外星物种的明显新伙伴身上。我们还讨论了趋同模式的个体差异以及语言趋同与地图任务成功之间缺乏关系。研究结果与实验室外的观察结果一致,即语言使用者会朝着预期的语言行为趋同。它们还对理解语言适应以及社会信息对语言处理和产生的影响具有更广泛的意义。