The Semantic Drift of Quotations in Blogspace: A Case Study in Short‐Term Cultural Evolution

Abstract

We present an empirical case study that connects psycholinguistics with the field of cultural evolution, in order to test for the existence of cultural attractors in the evolution of quotations. Such attractors have been proposed as a useful concept for understanding cultural evolution in relation with individual cognition, but their existence has been hard to test. We focus on the transformation of quotations when they are copied from blog to blog or media website: by coding words with a number of well-studied lexical features, we show that the way words are substituted in quotations is consistent (a) with the hypothesis of cultural attractors and (b) with known effects of the word features. In particular, words known to be harder to recall in lists have a higher tendency to be substituted, and words easier to recall are produced instead. Our results support the hypothesis that cultural attractors can result from the combination of individual cognitive biases in the interpretation and reproduction of representations.

Keystrokes: A practical exploration of semantic drift in timed word association tasks

“…Indeed, similar results were found in several other COVID related studies, reinforcing the speed with which semantic drift can erode and reform the semantic landscape [26-28]. But semantic drift is far from limited to large scale, environmental change, and is as normal an occurrence in the blogosphere [29] as it is in many a regular conversation [30]. One aspect to semantic drift that has received little attention, is the role of the aforementioned cognitive processes themselves on the meaning of words generated via different systems over time, where earlier associations may result from language associations common to a majority of speakers (for example, antonyms, synonyms, hierarchies), while later associations may be the result of conceptual processing via situated generation [11].…”

Socio-semantic and other dualities

“…This led to a burgeoning of semantic network analyses in communication science (Corman, Kuhn, McPhee and Dooley, 2002;van Atteveldt, 2008;Nerghes, Hellsten and Groenewegen, 2015;Nerghes, Lee, Groenewegen and Hellsten, 2015), which already had the tradition of associating semantic networks with particular social groups (Danowski, 1982;Monge and Eisenberg, 1987;Rice and Danowski, 1993;Doerfel, 1998;Doerfel and Barnett, 1999). Simultaneously, rapid progress was made in the adjacent fields of cognitive science (Borge-Holthoefer and Arenas, 2010;Teixeira, Aguiar, Carvalho, Dantas, Cunha, Morais, Pereira and Miranda, 2010), linguistics (Steyvers and Tenenbaum, 2005), and computational social sciences (Taramasco, Cointet and Roth, 2010;Lerique and Roth, 2018). Researchers derived relations between words from various types of textual data and proposed natural language processing extensions, such as syntax-based approaches to enhancing word associations with information about parts of speech, grammatical relations between words, and ontologies (Corman, Kuhn, McPhee and Dooley, 2002;van Atteveldt, 2008;Sudhahar, Veltri and Cristianini, 2015;Evans and Aceves, 2016;Nivre, De Marneffe, Ginter, Goldberg, Hajic, Manning, McDonald, Petrov, Pyysalo and Silveira, 2016).…”

Four misunderstandings about cultural attraction

“…Random variation of vocal pitch is unlikely to play any significant role in explaining the cultural success of a given folktale. In contrast, attraction toward portraits with directrather than averted-gaze; 19 toward vertical and horizontal lines in the design of orthographic systems; 92 toward the collocation of blood and action in medical treatments; 18 toward procedural rather than realist interpretations of the infinitesimal in the evolution of calculus; 93 or toward misquotations with words more memorable than those in the original quote 94 -with this short list we are highlighting just a small handful of the numerous empirical studies inspired by CAT-are all types of biased transformation that have effects at the cultural level, and whose effects can only be detected with relatively fine-grained description of the relevant data.…”