In recent years, sensorimotor effects in lexical–semantic processing have been a vibrant topic of research. This interest has been driven largely by the relevance of these effects to theories of semantic representation. In particular, embodied semantic theories claim that retrieval of word meaning, even in simple reading and semantic decision tasks, involves activation of the sensorimotor and perceptual systems (e.g., Barsalou, 2008; Glenberg, 2015). As such, sensorimotor effects in lexical–semantic processing have been used to test the viability of embodied accounts of semantic representation.
One common strategy researchers have used to study sensorimotor effects in lexical processing is to examine effects of words’ rated body–object interaction (BOI). BOI has been a focus of study because the dimension captures the relative availability of sensorimotor information (Hargreaves et al., 2012). Ratings of BOI are intended to measure the ease with which the human body can interact with a word’s referent (Siakaluk, Pexman, Aguilera, Owen, & Sears, 2008). Words whose referents afford relatively more BOI (e.g., toothbrush) are typically processed faster than words whose referents afford fewer opportunities for interaction (e.g., elephant) (Siakaluk, Pexman, Aguilera, et al., 2008; Siakaluk, Pexman, Sears, et al., 2008). Facilitation for high BOI words in lexical–semantic tasks is typically attributed to richer semantic representations of motoric interactions for those items (Pexman, 2012; Yap, Pexman, Wellsby, Hargreaves, & Huff, 2012).
The facilitation of lexical–semantic processing for words with high BOI ratings versus words with low BOI ratings (hereafter the “BOI effect”) is now well established, having been replicated and demonstrated across a variety of contexts. For example, the BOI effect has been observed in lexical decision tasks (LDT) (Siakaluk, Pexman, Aguilera, et al., 2008; Tillotson, Siakaluk, & Pexman, 2008; Van Havermaet & Wurm, 2014), semantic decision tasks (e.g., “Is it concrete?,” “Is it easily imageable?”; Bennett, Burnett, Siakaluk, & Pexman, 2011; Hansen, Siakaluk, & Pexman, 2012; Hargreaves, Leonard, et al., 2012; Siakaluk, Pexman, Sears, et al., 2008; Tousignant & Pexman, 2012; Yap et al., 2012), a semantic lexical decision task (i.e., “Is it a word?,” then “Is it easily imageable?”; Siakaluk, Pexman, Sears, et al., 2008), sentence reading (Phillips, Sears, & Pexman, 2012), and with child participants in auditory-word naming (6- to 7-year-old children; Inkster, Wellsby, Lloyd, & Pexman, 2016) and in printed-word naming (8- to 9-year-old children; Wellsby & Pexman, 2014). The BOI effect has also been observed using fMRI: High-BOI words were associated with higher levels of activation in the left inferior parietal lobule (a sensory association area involved in the planning of object-directed hand movements; van Elk, 2014) than were low-BOI words (Hargreaves, Leonard, et al., 2012). The fact that BOI effects are observed in lexical–semantic tasks suggests that children and adults routinely access information about their past sensorimotor experience with words’ referents when making decisions about word meaning. Thus, BOI effects have been taken as evidence that sensorimotor information is important to representations of word meaning, although they cannot adjudicate between models that assume sensorimotor information is necessary for meaning activation (e.g., Glenberg, 2015) and those that assume sensorimotor information is simply activated as a by-product of meaning activation (e.g., Mahon, 2015).
The BOI effect has been frequently observed, but the precise nature of the information captured by the BOI dimension is not well specified. That is, BOI is a rather coarse semantic dimension that seemingly captures a variety of types of sensorimotor information and does not specify the nature of the interaction. Of particular relevance to the present study is the fact that the contribution of specific kinds of motor experience to facilitatory BOI effects has not previously been examined. The imprecision of the BOI measure makes it challenging to draw theoretical conclusions, because it is not clear what aspects of bodily information are being activated when word meanings are retrieved. The goal of the present study was to identify which aspects of motor experience drive BOI effects in order to better understand the BOI effects observed thus far, and to help refine theories of semantic representation by specifying the types of motor information involved in lexical–semantic processing.
Candidate motor dimensions
Several more specific motor dimensions may drive the BOI effects observed in lexical–semantic processing. In particular, there are specific aspects of object manipulation that have primarily been explored in the cognitive neuroscience, neuropsychology, and/or object recognition literatures, in which the stimuli are line drawings or photographs (e.g., Boronat et al., 2005; Garcea & Mahon, 2012; Guérard, Lagacé, & Brodeur, 2015; Madan, Chen, & Singhal, 2016; Madan, Ng, & Singhal, 2018; Salmon, Matheson, & McMullen, 2014; Salmon, McMullen, & Filliter, 2010; Tobia & Madan, 2017). In the present work we considered whether these specific dimensions are related to BOI and whether they also explain word recognition behavior. We chose three candidate motor dimensions that have been shown in the object recognition literature to be related but also somewhat distinct aspects of objects’ motor attributes (Guérard et al., 2015).
Graspability ratings provide a measure of how easily a person can grasp an object with one hand (Salmon et al., 2010). Many of the words that are rated as being high in BOI refer to objects that seem easily graspable (e.g., scissors). Thus, it is plausible that the facilitation effect observed for high-BOI words in lexical–semantic tasks is actually driven by sensorimotor information captured by graspability. Whereas graspability is based on how easily the hand can interact with a word’s referent, BOI measures the ease with which any part of the body can interact with the word’s referent. Nonetheless, the fact that the hands are primary for most human interactions with objects means that there is likely a good deal of overlap in the two dimensions. Effects of graspability have been investigated in both object recognition (Grèzes & Decety, 2002; Guérard et al., 2015; Salmon et al., 2014) and word recognition (Amsel, Urbach, & Kutas, 2012; Díez-Álamo, Díez, Alonso, Vargas, & Fernandez, 2017) research. Two previous studies examined the relationship between graspability and BOI for word stimuli: Amsel et al. (2012) found a moderate correlation of r = .62 between the two dimensions for a set of 266 words, Díez-Álamo et al. found r = .75 for the same relationship in a set of 342 Spanish words. Thus, some evidence suggests that there is indeed overlap in the types of motoric information captured by these two measures, but neither previous study explored their relative contributions to lexical processes.
Ease of pantomime
Ease of pantomime refers to how easily one can pantomime an object’s functional use so that another individual could guess the identity of the object (Brodeur, Dionne-Dostie, Montreuil, & Lepage, 2010; Guérard et al., 2015; Magnié, Besson, Poncet, & Dolisi, 2003). For example, the referent for wharf would be difficult to pantomime, whereas doorknob would be easy (e.g., making a cupping shape with your hand and twisting at the wrist). This dimension depends on functional actions and so it likely taps how readily one can retrieve the conceptual knowledge associated with the object. Indeed, ease of pantomime is related to object naming latencies, with faster latencies for objects that are easier to pantomime (Guérard et al., 2015). Because BOI captures the ease with which the human body can interact with a word’s referent, which typically occurs through functional use of the referent, it is plausible that ease of pantomime may be related to BOI for word stimuli and may drive or help drive the BOI effect.
Number of actions
The number-of-actions dimension captures the number of functional actions that can typically be performed with an object (Guérard et al., 2015; Lagacé, Downing-Doucet, & Guérard, 2013). This dimension also has a relationship with object-naming latencies, such that objects that afford more actions are associated with faster object naming (Guérard et al., 2015). Interaction between a word’s referent and the human body typically occurs while using the referent for a functional purpose. For example, a word with a referent that affords few actions (e.g., fleck) is unlikely to be easy for the human body to interact with, whereas a word with a referent that affords many actions (e.g., baby: holding, cuddling, playing, etc.) is very easy for the human body to interact with. Thus, high-BOI words may afford relatively more actions and motor information regarding the number of actions a word’s referent affords may drive or help drive the BOI effect.
Other non-motor dimensions related to BOI
Several other semantic dimensions might be related to BOI. In particular, we expected that animacy, size, danger, and usefulness might be relevant. BOI attempts to capture the ease with which the human body can interact with a word’s referent and so it is possible that words with referents that are small, inanimate, nondangerous, and useful would be rated high on this dimension and that these factors may be related to the facilitation effect observed for BOI in lexical–semantic processing. Previous research suggests that each of these four dimensions may be related to semantic processing. For instance, there is evidence of faster recognition of animate than of inanimate objects (Proverbio, Del Zotto, & Zani, 2007) and of better memory for words referring to animate concepts than for words referring to inanimate concepts (e.g., Bonin, Gelin, & Bugaiska, 2014). There is also evidence that words referring to large objects are recognized faster than words referring to small objects (Sereno, O’Donnell, & Sereno, 2009), but at the same time, there is evidence that object size does not affect word recognition (Kang, Yap, Tse, & Kurby, 2011). Effects of danger, usefulness, and their interactions with BOI have been characterized in recent studies by Wurm and colleagues (e.g., Van Havermaet & Wurm, 2014, 2017; Wurm, 2007; Wurm & Seaman, 2008). Although some dangerous objects are frequent targets of interaction, it seemed possible that the caution required when interacting with such objects might lead to lower BOI values since those assess relative “ease” of interaction. We included each of these variables in the present study in order to assess their relationships with BOI, and to allow us to examine effects of the candidate motor dimensions independent of these other nonmotor factors.
Here we investigated several semantic dimensions that could be related to BOI effects observed in lexical–semantic processing, focusing in particular on the motor dimensions described by Guérard et al. (2015) as influential for object recognition. We collected new ratings so that we could examine effects for word stimuli. Our stimuli were 621 words that had been previously rated for BOI (Bennett et al., 2011; Tillotson et al., 2008). We used hierarchical regression analyses to examine relationships between our candidate dimensions and BOI ratings. We then examined whether our candidate dimensions predicted variance in semantic decision behavior using response data from the Calgary Semantic Decision (Calgary SDT) Project (Pexman, Heard, Lloyd, & Yap, 2017). In the Calgary SDT Project, the decision category was “is it concrete or abstract?.” We chose the behavioral data from this task to examine effects of the candidate motor dimensions because in this task a relatively large amount of variance is explained by semantic variables (Pexman et al., 2017). Less variance is typically explained by semantic variables in the lexical decision task (LDT; see Pexman, 2012, for a review), but because the LDT is so widely used, we also examined whether our candidate dimensions predicted variance in LDT behavior using response data from the English Lexicon Project (Balota et al., 2007). We predicted that the semantic variables would account for less variance in LDT than SDT.
If a single motor dimension underlies the BOI dimension, then a significant amount of unique variance in BOI ratings should be predicted by ratings on that motor dimension. If multiple motor dimensions underlie the BOI dimension, then ratings on several motor dimensions should significantly predict variance in BOI ratings. Finally, the motor dimension(s) found to predict unique variance in BOI ratings will likely also predict SDT and LDT response behavior, and may be a stronger predictor of SDT and/or LDT behavior than is BOI.