Conducting Language Production Research Online: A Web-based Study of Semantic Context and Name Agreement Effects in Multi-Word Production

Abstract

Few web-based experiments have explored spoken language production, perhaps due to concerns of data quality, especially for measuring onset latencies. The present study highlights how speech production research can be done outside of the laboratory by measuring utterance durations and speech fluency in a multiple-object naming task when examining two effects related to lexical selection: semantic context and name agreement. A web-based modified blocked-cyclic naming paradigm was created, in which participants named a total of sixteen simultaneously presented pictures on each trial. The pictures were either four tokens from the same semantic category (homogeneous context), or four tokens from different semantic categories (heterogeneous context). Name agreement of the pictures was varied orthogonally (high, low). In addition to onset latency, five dependent variables were measured to index naming performance: accuracy, utterance duration, total pause time, the number of chunks (word groups pronounced without intervening pauses), and first chunk length. Bayesian analyses showed effects of semantic context and name agreement for some of the dependent measures, but no interaction. We discuss the methodological implications of the current study and make best practice recommendations for spoken language production research in an online environment.

Individual differences in online research: Comparing lab-based and online administration of a psycholinguistic battery of linguistic and domain-general skills

“…Standard psycholinguistic effects have also been replicated in online studies, including effects of word frequency, age of acquisition and name agreement on the speed of picture naming, lexical decision, and self-paced reading, among others (Corps & Meyer, 2023;Fairs & Strijkers, 2021;He et al, 2021;Hilbig, 2016).…”

Lexical selection in language production

“…However, because such statistics and their associations with dominant name production latencies are so often interpreted as evidence for competitive lexical selection under the competitor activation account (e.g., Alario et al, 2004;Bonin et al, 2002;Bose & Schafer, 2017;Britt et al, 2016;Cheng et al, 2010;Griffin, 2001;He et al, 2021;Johnson, 1992;Kan & Thompson-Schill, 2004;Karimi & Diaz, 2020;LaGrone & Spieler, 2006;Madden et al, 2019;Nozari & Hepner, 2019;Paivio et al, 1989;Shao et al, 2014;Székely et al, 2003;Vitkovitch & Tyrrell, 1995), the first challenge in this paper will be to assess whether an absolute selection threshold is sufficient to generate associations between (1) dominant name probability, (2) number of names, (3) entropy, and (4) dominant name production latencies, per the target activation account.…”

“…More generally, as discussed in the Introduction, although researchers' interpretation of name agreement metrics in picture naming has, in recent decades, shifted to emphasize an assumption of competition among multiple appropriate responses (e.g., Alario et al, 2004;Bonin et al, 2002;Bose & Schafer, 2017;Britt et al, 2016;Cheng et al, 2010;Griffin, 2001;He et al, 2021;Johnson, 1992;Kan & Thompson-Schill, 2004;Karimi & Diaz, 2020;LaGrone & Spieler, 2006;Madden et al, 2019;Nozari & Hepner, 2019;Paivio et al, 1989;Shao et al, 2014;Székely et al, 2003;Vitkovitch & Tyrrell, 1995), they have invariably used measures of 'codability' that were originally proposed to estimate either the strength of the dominant name itself or dispersion amongst weak alternatives (see R. W. Brown & Lenneberg, 1954;Lachman, 1973). Under the earlier, conceptual, definition that motivated the metrics, 'codability' merely described how likely a stimulus was to strongly activate at least one response.…”

Timing in Conversation

“…Compared to silence or noise baselines or to speech that participants cannot understand (e.g. Chinese speech for native speakers of Dutch, He, Meyer, Creemers, & Brehm, 2021 ), the presentation of distractor words in the participants’ own language slows down picture naming. Moreover, with suitable timing of the distractors, semantically related distractors (e.g., “cat” for the picture of a dog) slow down naming more than unrelated ones (e.g.…”