Trying to make it work: Compositional effects in the processing of compound “nonwords”

Abstract

Speakers of languages with synchronically productive compounding systems, such as English, are likely to encounter new compounds on a daily basis. These can only be useful for communication if speakers are able to rapidly compose their meanings. However, while compositional meanings can be obtained for some novel compounds such as bridgemill, this is far harder for others such as radiosauce; accordingly, processing speed should be affected by the ease of such a compositional process. To rigorously test this hypothesis, we employed a fully implemented computational model based on distributional semantics to quantitatively measure the degree of semantic compositionality of novel compounds. In two large-scale studies, we collected timed sensibility judgements and lexical decisions for hundreds of morphologically structured nonwords in English. Response times were predicted by the constituents’ semantic contribution to the compositional process, with slower rejections for more compositional nonwords. We found no indication of a difference in these compositional effects between the tasks, suggesting that speakers automatically engage in a compositional process whenever they encounter morphologically structured stimuli, even when it is not required by the task at hand. Such compositional effects in the processing of novel compounds have important implications for studies that employ such stimuli as filler material or “nonwords,” as response times for these items can differ greatly depending on their compositionality.

Nymph piss and gravy orgies: Local and global contrast effects in relational humor.

“…In contrast, the constituent-compound predictors were not as strong predictors of compound-word humor, even though they have been found to be important predictors for the perceived meaningfulness of compounds (Günther & Marelli, 2016) and in compound processing (Günther & Marelli, 2020). The small but significant effect of constituent1-compound similarity suggests that some nonnegligible amount of automatic meaning construction is occurring when processing the word pairs for humor.…”

“…In a two-word context, fluent readers reflexively attempt to construct meaning from the two words, and likely less so in a single-word context. This is supported by single-word psycholinguistic investigations that find that semantic variables are less crucial predictors of performance in single-word recognition tasks than in tasks that involve a pair of concepts or categories as in semantic categorization or classification (Goh et al, 2016; Yap et al, 2011), as well as research into how people process the meaningfulness of known and novel compounds (Günther & Marelli, 2016, 2020). This may suggest that in a two-word context, a person may hold stronger expectations about the semantics of the compound than in the single-word context such that semantic variables play a more important role in the violation of such expectations in compound-word humor than in single-word humor.…”

“…Here we examined whether the similarity between the first constituent and the entire compound (i.e., constituent1-compound), as well as the similarity between the second constituent and the entire compound (i.e., constituent2-compound), might improve our models from Study 1 and Study 2. Because these measures were obtained from the model by Günther and Marelli (2020), to be consistent the measure of semantic similarity between the two individual words was also derived from the same model rather than reusing the measure obtained from the Macroscope (Li et al, 2019).…”

“…Orthographic distance and Phonological distance: The orthographic distance between two words in a given word pair was the 1 A reviewer (Fritz Günther) noted that a potential point of concern with the current analysis was that our measures were collected from a variety of sources based on different language corpora, which will have different underlying distributional properties. To assess if our results might be an artifact of this we re-ran our analyses with word and letter frequencies obtained from the same source corpus (Günther & Marelli, 2020) and found that the pattern of results did not change. Another analysis that was conducted was to include orthotactic and phonotactic frequencies (i.e., probabilities of pairs of letters and sounds of words in the language) as additional predictors and again we found that it did not change the overall pattern of findings.…”

Distilling vector space model scores for the assessment of constructed responses with bifactor Inbuilt Rubric method and latent variables

“…All these conclusions are directly related to psychoeducational assessments of constructed responses. However, such general dimensions may provide substantive variance to distill the modeling of other cognitive processes working as a proxy of general semantic noise to distill compositional processes (e.g., Günther & Marelli, 2020;Marelli et al, 2017) or modulate similarity judgments of concepts (e.g., Ichien et al, 2021;Netisopakul et al, 2021).…”