EFFECTS OF EMOTIONAL EXPERIENCE IN ABSTRACT AND CONCRETE WORD PROCESSING by P. Ian Newcombe B.Sc., University of Northern British Columbia, 2012 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN PSYCHOLOGY UNIVERSITY OF NORTHERN BRITISH COLUMBIA February 2016 © P. Ian Newcombe, 2016 iii TABLE OF CONTENTS Abstract II Table of Contents Ill List of Tables V Classical and Grounded Cognitive Science Frameworks Evidence Supporting Grounded Cognition The Feedback Activation Framework and Visual Word Recognition Evidence for a Grounded Account of Abstract Word Processing Situated Conceptualization 6 Chapter One The Present Study Primary Hypotheses EE Effects in Experiments 1A and 1B EE Effects in Experiments 2A and 2B EE Effects in Experiments 3A and 3B Secondary Hypotheses 16 17 17 18 19 19 Chapter Two General Methods Participants Stimuli Procedure Lexical Decision Task Semantic Categorization Task Semantic Lexical Decision Task Data Analyses 20 20 21 21 23 23 23 24 Chapter Three Results and Discussion Experiment 1 Data sets Omnibus Analysis Experiment IA -Abstract LDT Experiment 1B - Concrete LDT 26 26 26 26 27 Chapter Four Results and Discussion Experiment 2 Data sets Omnibus Analysis Experiment 2A - Abstract SCT Experiment 2B - Concrete SCT 28 28 28 29 29 Chapter Five Results and Discussion Experiment 3 Data sets 31 31 Introduction 1 3 9 14 iv Omnibus Analysis Experiment 3A - Abstract SLDT Experiment 3B - Concrete SLDT 31 32 32 Chapter Six Secondary Hypotheses Discussion 33 Chapter Seven General Discussion Experiments 1A and 1B Experiments 2A and 2B Experiments 3A and 3B Secondary Hypotheses Semantic Feedback Framework Semantic Representation Framework Positivity, Negativity, Context Availability, Arousal and Emotional Experience Conclusions 35 36 37 38 40 41 43 45 47 References 46 Tables 52 Appendices Footnotes Abstract Stimuli Concrete Stimuli Nonword Stimuli 69 70 71 72 V LIST OF TABLES Table 1 Descriptive statistics and behavioral data for the 75 high EE abstract words, 75 low EE abstract words, 75 high EE concrete words and 75 low EE concrete words 57 Table 2 Zero-order Correlations Between the Criterion Variables and the Predictor Variables for Abstract Words 58 Table 3 Zero-order Correlations Between the Criterion Variables and the Predictor Variables for Concrete Words 59 Table 4 Results of Interaction Tests in the Omnibus Latency Analyses Between Abstract and Concrete Words in LDT, SCI, and SLDT 60 Table 5 Results of Interaction Tests in the Omnibus Error Analyses Between Abstract and Concrete Words in LDT, SCI, and SLDT 60 Table 6 Results of Hierarchical Multiple Regression Analyses for the Experiment IA: Abstract LDT 60 Table 7 Results of Hierarchical Multiple Regression Analyses for the Experiment lB: Concrete LDT 60 Table 8 Results of Hierarchical Multiple Regression Analyses for the Experiment 2A: Abstract SCI 60 Table 9 Results of Hierarchical Multiple Regression Analyses for the Experiment 2B : Concrete SCI 60 Table 10 Results of Hierarchical Multiple Regression Analyses for the Experiment 3A: Abstract SLDT 60 Table 11 Results of Hierarchical Multiple Regression Analyses for the Experiment 3B : Concrete SLDT 60 SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Introduction Classical and Grounded Cognitive Science Frameworks Classical cognitive science proposes that cognitive representations are abstract and amodal in nature. That is, perceptual input is coded and stored in memory as abstract, amodal symbols, analogous to a computer converting user input into binary code in its storage (i .e., memory) systems. As such, an important feature of classical cognitive science is that perception and action (i .e. , sensorimotor) representations are orthogonal to cognitive representations. Classical theories in cognition have provided many structural frameworks of cognitive systems (e.g. , see Fodor, 1975; Marr & Vision, 1982; Neisser, 2014; KarmiloffSmith, 1979), yet have failed to adequately address the symbol grounding problem (Hamad, 1990). Originally described by Searle ( 1980), the symbol grounding problem highlights the inability to assign a symbol meaning using only other abstract symbols. For example, if one were to see a road sign containing a word in a foreign language, it would be impossible to learn its meaning using only a foreign dictionary. The definition of the word in the dictionary would consist of more incomprehensible symbols, which would then require one to define each of them, and so forth. Thus, amodal theories appear to lack the ability to account for how abstract, amodal symbols derive meaning, because the symbols are not grounded in sensorimotor experience (Glenberg & Robertson, 2000). In contrast to classical cognitive science, grounded (or embodied) cognitive science proposes that sensorimotor systems are fundamental to cognition. Storage of sensorimotor information resides in modality-specific cortical regions dedicated to sensory systems such as vision, olfaction, and gustation (among others), as well as systems dedicated to processing action information. For example, activation patterns arising in sensorimotor cortices during SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 2 physical interaction with a table (i.e., online processing) are fundamentally the same kinds of activation patterns that arise in sensorimotor cortices when reading the word table 1 (i .e. , offline processing). An important and influential grounded cognition framework is Perceptual Symbol Systems (PSS; Barsalou, 1999), which proposes that the acquirement, storage, and retrieval of conceptual knowledge employs neural systems dedicated to processing different types of perceptual and action information. PSS offers one theoretical way to adequately address the symbol grounding problem because it proposes that the conceptual information contained in symbols is largely grounded in perception and action. Each time a physical interaction with a [TABLE] occurs, a unique sensorimotor cortical activation pattern is experienced. This is due to [TABLES] not being identical; they have different features such as NUMBER OF LEGS, COLOUR, or SIZE. Consequently, cortical activity will vary to some degree, in accordance with variations in sensorimotor experience. As hypothesized by Damasio (1989), conjunctive neurons save patterns of cortical activation during perception and action into memory. With each new sensorimotor experience, the saved cortical activation patterns are adjusted in regions called convergence zones, which strengthen the general features present across multiple bodily experiences. After many physical experiences with [TABLES) , conjunctive neurons activate a neural activation pattern that represents a generalized version of [TABLE], and they do so by activating multiple sensorimotor modalities simultaneously. Importantly, this generalized activation also occurs when [TABLE] is required for offline cognition, such as in reading about or imagining a table. When reading table for example, conjunctive neurons activate multiple neural patterns, such as motor patterns (representing past experiences of one's physical interactions with tables), auditory patterns (representing past experiences of how SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 3 things sound when placed on tables), and visual patterns (representing past experiences of what tables look like). The combined activation of multiple modalities contains the knowledge required for conceptual processing. Barsalou (1999) calls this cumulative activation of sensorimotor knowledge in multiple modalities a simulation. According to PSS, simulation is the central mechanism used by cognitive processes. In summary, classical cognitive science emphasizes the use of amodal symbol manipulation, devoid of the need or use of information stored in sensorimotor systems during conceptual processing. Classical theories fail to address how abstract symbols derive meaning and thus cannot account for the symbol grounding problem. Grounded cognitive science frameworks such as PSS emphasize the multimodal activation of sensorimotor systems (i.e., simulations) during conceptual processing. PSS is a framework that suggests a possible solution to the symbol grounding problem : symbols may be grounded in bodily, sensorimotor experience. Symbols are not amodal in nature, but represented, and hence grounded, in multimodal sensorimotor cortical regions. Evidence Supporting Grounded Cognition Recent research supports the notion that sensorimotor information influences cognitive processing. For the purposes of this thesis, the focus will be on several effects of sensorimotor information on linguistic processing. Evidence from three domains of linguistic processing will be discussed: sentence processing, metaphor comprehension, and single word recognition. In a study investigating the interaction between physical motor movements and sentence comprehension, Glenberg and Kaschak (2002) had participants read either coherent SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 4 sentences (e.g., y ou give John the flowers or John gives you the flowers) or nonsense sentences ( e.g., the lamp jumps with salsa). The task was to differentiate between coherent and nonsense sentences. To make a response, participants were instructed to push a button that required either a forward arm movement (i .e., away from the participant's body) or a backward arm movement (i.e. , toward the participant' s body) . Results showed that if the action implied by the sentence (e.g., away from one' s body such as in the sentence: you give John the flowers) was congruent with the required arm movement for the response (e.g. , away from one ' s body), participants were significantly faster at identifying the sentence, as compared to incongruent sentences (e.g. , away from one ' s body in response to the sentence: John gives you the flowers). The authors described this effect as the action-sentence compatibility effect. According to PSS, reading a sentence activates a simulation that, in this case, includes the motor cortical activation responsible for forward arm movement (e.g. , giving the flowers to John) or its opposite (e.g., John giving you the flowers). Such motor activation facilitates motor movements that are compatible with the sentence, thus leading to faster responding. Effects of perceptual manipulations have also been observed in sentence comprehension. In a study by Yaxley and Zwaan (2007), participants performed a sentence- picture verification task, whereby they were tasked with verifying whether a presented picture was mentioned in a preceding sentence. Two independent variables were manipulated. First, the sentences differed with regards to clarity of visual experience, with sentences referring either to high clarity (e.g., through the clean goggles, the skier could easily identify the moose) or low clarity (e.g., through the fogged goggles the skier could hardly identify the moose). Second, the pictures differed as to degree of resolution, with SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 5 either high resolution pictures or low resolution pictures (e.g., of a moose). Participants verified the picture as being mentioned in the sentence faster if sentence clarity and picture resolution were congruent (e.g., high clarity sentence and high resolution picture) as compared to incongruent conditions (e.g., high clarity sentence and low resolution picture). Importantly, the perceptual congruency effect was observed not only for high clarity sentences and high resolution pictures, but also for low clarity sentences and low resolution pictures. The results suggest that visual simulation is also an important mechanism that occurs during sentence comprehension. (For a review of simulation processes involved in sentence comprehension, see Zwaan & Madden, 2005.) Another area of interest for grounded cognitive researchers is metaphor comprehension. Metaphors are an intriguing area of study for researchers due to the frequency and fluidity of their use in language. Wilson and Gibbs (2007) demonstrated that a mapping of sensorimotor experience onto abstract metaphors occurs in order to understand them. After performing or imagining performing a specific action (e.g., spitting), participants were faster at identifying a metaphor if it matched the previously performed or imagined action (e.g. , spit out the truth) as compared to mismatched (e.g., grasp the concept) or no action conditions (e.g. , take a long nap). The constraint of sensorimotor experience prevents literal interpretations of metaphors to occur (i.e. , one cannot literally spit out the truth), yet the grounded nature of the mind maps such abstract concepts onto sensorimotor modalities. (For a review of simulation processes involved in metaphor generation and comprehension, see Gibbs, 2005 ; Lakoff & Johnson, 1980; 1999.) Finally, research has been conducted to explore the usefulness of grounded cognitive frameworks in explaining the effects of several dimensions of sensorimotor knowledge SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 6 observed in visual word recognition. The first dimension is imageability, which refers to the ease with which a word ' s mental image can be evoked. Research consistently reports a facilitatory imageability effect in visual word recognition studies: Words rated higher in imageability (e.g., keyboard) are responded to faster and more accurately than words rated lower in imageability (e.g. , assert) (Cortese & Fugett, 2004; Strain & Herdman, 1999; Strain, Patterson, & Seidenberg, 1995). The facilitatory imageability effect can be readily explained by PSS such that greater imageability is associated with greater available sensory knowledge, leading to richer sensory simulations that facilitate responding. The second dimension is body-object interaction (BOI), which refers to the ease with which a human body can physically interact with a word ' s referent. Words rated higher in BOI (e.g., chair) are responded to faster and more accurately than words rated lower in BOI (e.g., ship) in a variety of visual word recognition tasks (Siakaluk, Pexman, Aguilera, Owen, & Sears 2008; Siakaluk, Pexman, Sears, Wilson, Lochheed, & Owen, 2008; Wellsby, Siakaluk, Owen, & Pexman, 2011; Bennett, Burnett, Siakaluk, & Pexman, 2011). The BOI effect can be readily explained by PSS such that greater BOI is associated with greater available motor knowledge, leading to richer motor simulations that facilitate responding. The Semantic Feedback Activation Framework and Visual Word Recognition Visual word recognition tasks are among the powerful tools for investigating grounded cognitive theory, yet in order to adequately explain how sensorimotor semantic variables (e.g. , imageability and BOI) may influence conceptual processing, a framework for the role of semantic knowledge in visual word recognition must be described. The most prominent such framework is the semantic feedback activation framework (Hino & Lupker, 1996). According to this framework, three types of units exist within the lexical processing SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 7 system: orthographic units that process the spellings of words; phonological units that process the sounds of words ; and semantic units that process the meanings of words. The different types of units are interconnected and may influence one another through feedforward and feedback mechanisms (as outlined below). Although the activation of phonological units is important for responding in some tasks such as naming (Hino & Lupker), the focus of the current discussion will be on mechanisms involved with the activation of orthographic and semantic units. According to the semantic feedback activation framework, two unique mechanisms account for the influence of semantics during visual word recognition: semantic feedback and semantic processing. These mechanisms interact with both task demands and variations in semantic richness of the stimuli. A widely used task used to measure the influence of semantics is the lexical decision task (LDT) . In LDT, participants are presented with Jetter strings that are either real words (e.g. , table) or non words (e.g. ,flaig) , and are instructed to make one response to the real words and a different response to the nonwords. Because the decision criterion is whether the letter string is a real word or not, correct responding primarily depends on orthographic information. According to the semantic feedback activation framework, correct LDT responses to words occurs when orthographic units settle on a correct representation, with faster settling being associated with faster responses. Importantly, if the presented word is relatively more semantically rich, a greater amount of semantic units will activate, consequently sending greater semantic feedback to the orthographic units. Therefore, according to the semantic feedback activation framework, words that are relatively more semantically rich will send greater feedback from the semantic SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 8 units to the orthographic units, quickening orthographic settling and facilitating LDT response times. As noted above, words rated higher in imageability are responded to faster in LDT than words rated lower in imageability. According to the semantic feedback activation framework, imageability contributes to the sensory semantic richness of a word, which increases the amount of semantic feedback, and consequently facilitates LDT responding. Similarly, words rated higher in BOI are responded to faster in LDT as compared to words rated lower in BOI, because BOI contributes to the motor semantic richness of a word that increases the amount of semantic feedback, and consequently facilitates LDT responding. Another task used to measure the influence of semantics in visual word recognition is the semantic categorization task (SCT) . The SCT requires participants to make some type of semantic decision about visually presented words. The words are selected a priori to be categorized on some criterion (e.g., concreteness, such that half of the words are concrete words and the other half are abstract words), and participants then categorize the words using that criterion. For SCT, responding primarily depends on semantic information such that settling among the semantic units is the basis for responding (as opposed to settling among the orthographic units for LDT). This is the second mechanism by which semantics can influence the visual word recognition system and is called semantic processing. Thus, words that are relatively more semantically rich activate a greater amount of semantic units, and this greater activation among the semantic units quickens their settling and is associated with faster responding in the SCT. Imageability and BOI effects have also been found in SCT (Siakaluk et al., 2008a; Bennett et al. , 2011). In these studies, higher rated imageability words and higher rated BOI SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 9 words are associated with greater semantic richness (sensory and motor, respectively) and therefore elicit greater semantic processing, which in turn facilitates the settling of semantic units and leads to faster SCT responding. According to the semantic feedback activation framework, semantic feedback and semantic processing are proposed to be independent mechanisms, and it can therefore be hypothesized that their independent influences could be measured in a single task. Siakaluk et al. (2008b) devised such a task and called it the semantic lexical decision task (SLOT). In SLOT, participants are required to make two judgments regarding a presented letter string: first, to determine if the letter string is a real word (as in LDT), and second, if the letter string is a word, to determine if it belongs to a given semantic category (as in SCT). Should a semantic variable (e.g., imageability or BOJ) influence both semantic feedback and semantic processing independently, then perhaps the effect size may be greater in SLOT than in either LDT or SCT. Indeed, Siakaluk et al. (2008b) found such a result when assessing the effects of BO I in SCT and SLOT using the semantic decision category of high versus low imageability (i.e., is the word easily imageable?). Although there were facilitatory effects of BOI in both tasks, the effect was significantly larger in SLOT than in SCT. Evidence for a Grounded Account of Abstract Word Processing Although imageability and BOI effects provide evidence in support of grounded cognition in lexical processing, both variables measure sensorimotor experience and are therefore measures of concrete conceptual processing. Although concrete concepts undoubtedly underlie a large portion of conceptual knowledge, if grounded cognitive theory is to be considered comprehensive, it must also account for how abstract conceptual knowledge is processed. Grounded cognition researchers have proposed potential ways in SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 10 which grounded cognition frameworks may account for the processing of abstract concepts. Some examples include the grounding of abstract concepts onto physical representations, such as, for example, grounding the meaning of good and bad onto the right and left sides of the body (Casasanto, 2009); the meaning of anger in the physical sensation of heat (Wilkowski, Meier, Robinson, Carter, & Feltman, 2009); or the meanings of abstract metaphors (e.g., push the argument) onto sensorimotor experience (e.g., motor movement of physically pushing an object; Lakoff & Johnson, 1999; Gibbs, 2005 Wilson & Gibbs, 2007). Another mechanism may be that abstract concepts are grounded intrinsically in emotion bodily states (Vermeulen, Niedenthal, & Luminet, 2007). According to PSS , conceptual knowledge can be acquired not only from sensorimotor states but also from emotion (i .e., affective) states. Emotion knowledge may therefore contribute to mental simulations (Barsalou, 1999; Parisi, 2011 ; Vermeulen et al., 2007; Niendenthal , 2007) of abstract concepts. A recent study reported a modality-switching cost in linguistic processing of emotion phrases (Vermeulen et al., 2007). Emotion phrases such as a victim can be struck, are responded to faster if primed by a concept in the same modality (i.e. , negative affect) such as an orphan is helpless as compared to a prime in a different modality such as vision (e.g. , a spider is black). The observed modality switching cost is conceptually important because it provides support for the contribution of emotion knowledge to semantic processing independent of other modalities such as vision (as seen in imageability effects), or action (as seen in BOI effects). Therefore, one can propose that emotion knowledge is integrated into simulations in a similar fashion as sensorimotor information. SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 11 Vigliocco, Meteyard, Andrews, and Kousta (2009) proposed a (grounded cognition) semantic representation framework, which provides an account for how both concrete and abstract concepts are processed. According to their framework, conceptual processing is dependent on two types of information: firstly, knowledge gained through bodily experience, which, importantly, includes not only sensorimotor knowledge but also emotion knowledge; and secondly, knowledge gained through linguistic experience. By linguistic experience, Vigliocco et al. refer to the statistical regularities of syntactic and word co-occurrence information. Knowledge gained through both bodily and linguistic experience are active during conceptual processing of concrete and abstract concepts; however, the degree to which the conceptual processing draws on knowledge from the various sources depends on the nature of the concept. More specifically, processing of concrete concepts such as [TABLE] would use more sensorimotor knowledge as compared to emotion or linguistic knowledge, whereas abstract concepts such as [JUSTICE] would use more emotion and linguistic knowledge as compared to sensorimotor knowledge. Indeed, several studies have investigated the effects of emotion knowledge in abstract conceptual processing by specifically examining the effects of valence (i .e. , how positive or negative a concept is) and arousal (i .e., how calm or excited a concept is) in LDT. However, these findings are somewhat inconsistent. Estes and Adelman (2008), for example, reported slower lexical decision latencies to negative words (e.g., disappointment) as compared to positive words (e.g., pride; similar results were also reported by Larsen, Mercer, Balota, & Strube, 2008, and Kuperman, Estes, Brysbaert, & Warriner, 2014). The automatic vigilance hypothesis (Pratto & John, 1991) has been used to explain the above findings , such that for survival purposes, it is beneficial to attend to negative stimuli , which elicits more in-depth, SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 12 and hence longer attentional processing (resulting in longer lexical decision latencies) than that elicited from positive or neutral stimuli. Alternatively, some studies have found that both positive and negative words are responded to faster than neutral words (Kousta, Vinson, & Vigliocco, 2009; Vinson, Ponari, & Vigliocco, 2014; Yap & Seow, 2013). The consistent facilitatory effects observed for positive words across these studies suggests that positive emotion enriches the simulation process, speeding up lexical processing. Similar conflicting results have been found for arousal: Estes and Adelman (2008) and Recio, Conrad, Hansen and Jacobs (2014) reported faster LDT responses for words high in arousal, as compared to words low in arousal. However, Kuperman et al. (2014) observed faster LDT responses for low arousal words as compared to high arousal words . Although these studies lend support to the notion that emotion knowledge is important for the processing of abstract concepts, the inconsistency in the results suggests that further research is required to clarify the influence emotion knowledge plays in visual word recognition. Newcombe, Campbell, Siakaluk, and Pexman (2012) also addressed the question of the influence of sensorimotor and emotion knowledge in both concrete and abstract lexical processing using imageability, BOI, and a novel variable they called emotional experience (EE), which measures the ease with which words elicit or evoke emotion knowledge. Some abstract words may evoke emotion knowledge more easily (e.g. ,justice) than others (e.g., moment). According to PSS , a word that is rated higher in EE (e.g.,justice) implies that there is a richer emotion semantic representation of that word in memory, and thus richer emotion simulations occur when that word is processed. More specifically, when a higher rated EE word is processed, conjunctive neurons activate a richer network of neurons in emotion processing cortical regions. This activation in tum allows for richer emotion simulations that SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 13 facilitate conceptual processing. Indeed, in SCT using a decision criterion of "is this word abstract?", Newcombe et al. reported facilitatory EE effects such that responses were faster to abstract words rated higher in EE than to abstract words rated lower in EE. However, and somewhat surprisingly, in SCT using a decision criterion of "is this word concrete?", Newcombe et al. reported inhibitory EE effects such that concrete words rated higher in EE were responded to slower than concrete words rated lower in EE. In a subsequent study, Moffat, Siakaluk, Sidhu, and Pexman (2015) used the same two procedures as Newcombe et al., but had participants respond verbally rather than through button presses. Consistent with Newcombe et al. ' s findings , Moffat et al. reported facilitatory EE effects for abstract words, and inhibitory EE effects for concrete words. Newcombe et al. (2012) also reported that both imageability and BOI exerted facilitatory effects for concrete words (i.e., concrete words rated higher in imageability and concrete words rated higher in BOI were responded to faster than concrete words rated lower on these two dimensions). Although imageability exerted no effect for abstract words, Newcombe et al. reported inhibitory effects of BOI for abstract words (i.e. , abstract words rated higher in EE were responded to more slowly than abstract words rated lower in EE). The above results support Vigliocco et al. ' s (2009) semantic representation framework in the following ways. First, the facilitatory effects of EE on abstract word processing is consistent with the notion that abstract concepts use emotion knowledge during conceptual processing. Second, the facilitatory nature of imageability and BOI during concrete word processing is consistent with the notion that sensorimotor knowledge is associated with concrete conceptual processing. The inhibitory effects of EE on concrete SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 14 words and the inhibitory effects of BOI on abstract words were initially somewhat surprising, but can be made intelligible with the following explanation. If emotion knowledge is indicative of abstractness, then concrete words rated higher in EE should initially be more inconsistent with the decision criterion of " is the word concrete?", and hence additional processing time is required to resolve this ambiguity before a response is made, resulting in inhibitory effects of EE for concrete words. Similarly, if motor knowledge is indicative of concreteness, then abstract words rated higher in BOI should initially be more inconsistent with the decision criterion of "is the word abstract?", and additional processing time is required to resolve this ambiguity before a response is made, resulting in inhibitory effects of BOI for abstract words. In summary, the above findings suggest that the nature of the word being processed (i.e. , concrete or abstract) and the nature of the type of semantic knowledge that is being brought to bear (i.e., sensorimotor or emotion) will interact and differentially influence processing in visual word recognition . The facilitatory effects of imageability and BOI in concrete SCT and the inhibitory effects of these two dimensions in abstract SCT suggest that sensorimotor knowledge is indicative of concreteness. On the other hand, emotion knowledge, as measured by EE, appears to be indicative of abstractness. The facilitatory effects of EE in abstract SCT and the inhibitory effects of EE in concrete SCT support this notion. Situated Conceptualization Situated conceptualization is an extension of PSS , which highlights the importance of features of situations and environments to the development of conceptual knowledge. SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 15 Barsalou (2005 , 2009) proposed that concepts (both concrete and abstract) are nested within a wide array of other concepts across real world environments. For example, consider the processing of the concrete concept [TABLE]. Understanding the meaning of this concept does not, indeed cannot, occur in isolation from other concepts, such as co-occurrences with other objects (e.g., [CHAIR], [COMPUTER]), events (e.g., [WORKING], [EATING]), environments (e.g. , [OFFICE] , [KITCHEN] , or introspective states, importantly including emotion states (e .g., [RELIEF FROM PUTTING A HEAVY OBJECT ON THE TABLE] , [EATING WHILE HUNGRY AT THE KITCHEN TABLE]). Not only does situated conceptualization account for the processing of concrete concepts such as [TABLE] , it also accounts for the grounding of abstract concepts. For example, consider the concept [JUSTICE]. A situated conceptualization will include cooccurrences with objects (e.g., a [GAVEL] or [JAIL CELL]), events (e.g., [BEING PLACED UNDER ARREST] or [RECEIVING A SPANKING]), environments (e.g. , a [COURTROOM] or a [PRINCIPAL ' s OFFICE]), or introspective states, importantly including emotion states (e.g., [SATISFACTION] or [DISAPPOINTMENT]). Each component of situated conceptualization (i.e. , aspects of a situation) facilitates the processing of concrete and abstract concepts, yet their influence varies depending on the nature of the concept. Again, it has been proposed that concrete concepts are more heavily reliant on sensorimotor information, and abstract concepts are more heavily reliant on emotion information (Vigliocco et al., 2009). Indeed, when asked to list features and properties of abstract concepts (e.g., HOPE), participants list a greater variety of introspective properties (e.g. , happen, good, want), as compared to concrete concepts (e.g., SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 16 TREE), to which they are more likely to list physical properties (e.g., branches, growing, coloured; Wiemer-Hastings & Xu, 2005). Thus, on one hand it appears as though abstract concepts rely more heavily on a dimension of situated context that contains entities or properties located within the agent (e.g., emotion). On the other hand, concrete concepts appear to rely more heavily on a different dimension of the situated context, one which contains entities or properties located outside the agent (e.g. , through sensorimotor interaction with the world). Moffat et al. (2015) suggested that EE is a dimension of emotion knowledge, and thus located within the agent. They further suggested that a variable called context availability (CA) captures contextual information external to the agent. Originally developed by Schwanenflugel, Harnishfeger, and Stowe (1983), CA measures the ease with which words can be placed into situated contexts. Boat, for example, can be more easily placed into situated contexts as compared to allegory. Therefore, boat would be given a higher CA rating than allegory. Moffat et al. found that EE and CA accounted for significant amounts of unique variability in SCT, suggesting that these two variables are tapping into different types of situated experiential knowledge. Importantly, these results indicate that EE and CA measure unique dimensions of situated conceptualization and must be accounted for when investigating semantic processing. The Present Study The goal of the primary analyses of the present research was to compare abstract and concrete conceptual processing in three different word recognition tasks. Therefore, three sets of two experiments were conducted. The first set of experiments consisted of two SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 17 LDT's: Experiment IA used abstract words as the critical experimental stimuli, and Experiment lB used concrete words as the critical experimental stimuli. The second set of experiments consisted of two SCT's: Experiment 2A used abstractness as the semantic decision (i.e., "is the word abstract?"), and Experiment 2B used concreteness as the semantic decision (i.e., "is the word concrete?"). The third set of experiments consisted of two SLDT's: Experiment 3A used abstractness as the semantic decision, and Experiment 3B used concreteness as the semantic decision. The hypotheses for each of these pairs of experiments are provided below. As noted, the dimension of EE may influence the two mechanisms in the visual word recognition system. The first mechanism is semantic feedback, which is responsible for semantic effects in LDT. The second mechanism is semantic processing, which is responsible for semantic effects in SCT. It is proposed that both mechanisms should separately influence responding in SLDT, which allows for a secondary set of hypotheses to be made for the present research. That is, it is possible to determine if separate influences of these two semantic mechanisms will lead to different effects of EE in SLDT as compared to LDT or SCT. The secondary hypotheses will be further explained below. Primary Hypotheses Two important variables that will be manipulated in order to examine the effects of EE in LDT, SCT, and SLDT are: (1) the nature of the words being responded to (i.e. , abstract words or concrete words) and (2) task demands (i .e. , deciding if the stimulus is a word, deciding if the stimulus is abstract or concrete, or both). EE Effects in Experiments lA and lB SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 18 Experiments 1A and I B will first be examined with a test of interaction, which indicates whether the effects of EE are similar or different for the abstract and concrete words in LDT. It is hypothesized that there should be a significant interaction. For Experiment lA, facilitatory EE effects should be observed (i.e. , faster and more accurate responding should be associated with abstract words rated higher in EE), because emotion knowledge is a key aspect underlying the processing of abstract words (Vigliocco et al. , 2009). This result would be due to the greater semantic feedback from semantic units to orthographic units for abstract words rated higher in EE as compared to abstract words rated lower in EE. For Experiment lB, null effects of EE should be observed for one or more of the three following reasons. First, the control variables entered prior to EE in the multiple regression analyses may account for a sufficiently large amount of response latency and error variability that too little variability is left over to be accounted for by EE. Second, concrete words are generally considered to be more highly familiar than abstract words, and more familiar words do not necessarily require the recruitment of semantic knowledge to be processed (Forster, 1976). Third, according to Vigliocco et al. ' s semantic processing framework, emotion knowledge is not as important to the processing of concrete words as it is for abstract words, and as such, the influence of EE on concrete words in LDT may be negligable. EE Effects in Experiments 2A and 2B Experiments 2A and 2B will first be examined with a test of interaction, which indicates whether the effects of EE are similar or different for the abstract and concrete words in SCT. It is hypothesized that there should be a significant interaction. For Experiment 2A, there should be facilitatory effects of EE (i .e., faster and more accurate SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 19 responding should be associated with abstract words rated higher in EE), but inhibitory effects of EE for Experiment 2B (i.e., slower and Jess accurate responding should be associated with concrete words rated higher in EE). Importantly, both the facilitatory and inhibitory EE effects would be due to greater semantic processing within semantic units for words rated higher in EE. EE Effects in Experiments 3A and 3B Experiments 3A and 3B will first be examined with a test of interaction, which indicates whether the effects of EE are similar or different for the abstract and concrete words in SLOT. It is hypothesized that there should be a significant interaction. As noted, both the semantic feedback and the semantic processing mechanisms are involved in responding in this task, because participants are required to make both lexical and semantic decisions. As such, the two following hypotheses were made. First, in Experiment 3A, facilitatory effects of EE should be observed, because EE should influence both the semantic feedback and semantic processing mechanisms. Second, in Experiment 3B, inhibitory effects of EE should be observed, because EE should exert no effect on the lexical decision component, while exerting an inhibitory effect on the semantic decision component. Secondary Hypotheses In addition to the primary hypotheses outlined above, the following set of secondary hypotheses are proposed. The first set compares the size of the effects of EE for abstract words across Experiments IA, 2A, and 3A. More specifically, it is hypothesized that there should be smaller effects of EE in Experiments I A and 2A than in Experiment 3A. This is because there is only one semantic mechanism involved in Experiments IA and 2A, whereas SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 20 both semantic mechanisms are involved in Experiment 3A. To the extent that the two semantic mechanisms are independent of each other, and, importantly, their effects are measureable throughout the duration of the experimental trials, their combined effects should result in larger EE effects in Experiment 3A. The second set compares the size of the effects of EE for concrete words across Experiments I B, 2B , and 3B. More specifically, it is hypothesized that there should be no effects of EE in Experiment IB , but similar effects of EE in Experiments 2B and 3B, for the following reasons. First, in Experiment IB , EE should not exert a measurable influence through the semantic feedback mechanism for concrete words, because of the reasons outlined above. Second, in Experiment 2B, EE should exert an inhibitory influence through the semantic processing mechanism for concrete words. Third, in Experiment 3B, to the extent that the two semantic mechanisms are independent of each other, and, again, their effects are measureable throughout the duration of the experimental trials, their combined effects should result in inhibitory EE effects of roughly equal size to those of Experiment 2B. 2 General Methods Participants Separate groups of 30 undergraduate students from the University of Northern British Columbia (UNBC) participated in each experiment. All participants were native English speakers, and all had normal or corrected-to-normal vision. Upon completion of the experiment, each participant received one bonus credit, which was allocated to an applicable course of their choosing as per the guidelines used in the Department of Psychology. SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 21 Stimuli The 75 highest EE and 75 lowest EE abstract words and the 75 highest EE and 75 lowest EE concrete words were selected from Newcombe et al. (2012). Abstract words had concreteness and imageability ratings of 4.0 or less, whereas concrete words had concreteness and imageability ratings of 5.0 or higher. Control variables included HAL logfrequency, Levenshtein orthographic distance, number of letters, phonemes, syllables, and morphemes (English Lexicon Project; Balota et al., 2007), age of acquisition (Kuperman, Stadthagen-Gonzalez, & Brysbaert, 2012), concreteness and imageability (Friendly et al., 1992; Paivio, Yuille, Madigan, 1968), inverse N-count (Shaoul & Westbury, 2010), semantic diversity (Hoffman, Lambon Ralph, & Rogers, 2013), BOI (Newcombe et al. , 2012), negativity, positivity, and arousal (Warriner, Kuperman, & Brysbaert (2013), and CA (Moffat et al., 2015). As noted, EE ratings were taken from Newcombe et al. The 150 nonwords used in the LDTs and SLDTs were created by replacing one to three letters of real English words while remaining pronounceable (see Appendix C for a complete list). Procedure Recent research has indicated that blocking stimuli by a variable of interest increases the salience of that variable and allows for the effect to be more easily detectible (for examples of blocking by EE, see Moffat et al., 2015 ; Siakaluk, Knol, & Pexman, 2014). Blocking refers to presenting one type of stimuli in one group of trials, and then presenting a different type of stimuli in a different group of trials during the experiment. For the present study, this was done by including the 75 highest rated EE words being presented in block 1 and the 75 lowest rated EE words being presented in block 2 (within each block the words were randomized separately for each participant). Furthermore, the two blocks of trials were SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 22 randomly presented between participants (i.e. , half of the participants randomly received block 1 first and block 2 second, and the other half of the participants receive the reverse order of presentation). A pilot study revealed no effect of EE in LDT when the words rated higher in EE were intermixed with words rated lower in EE (i.e., EE was not blocked as outlined above). Therefore, in all the experiments presented below, the stimuli were blocked by the dimension of EE. In every experiment, stimuli were presented to participants one at a time in the centre of a computer monitor in white, Times New Roman, size 24 font on a black background. For LDT and SCT, stimuli remained on the screen until either a response was made or a time limit of 2,500 ms was reached. Because of the increased complexity of SLDT, stimuli remained on the screen until either a response was made or a time limit of 3,000 ms was reached. Presented stimuli were preceded by a fixation marker appearing in the middle of the screen for 1,000 ms. For LDT and SCT, participants had an opportunity to take a break after 150 trials. For SLDT, participants had an opportunity to take a break after 225 trials. Responses were made by pressing a key on a computer keyboard and response latencies were recorded by DirectRT software (http://www.empirisoft.com/DirectRT.aspx). Prior to LDT and SCT, participants completed practice trials consisting of20 stimuli (for Experiment IA, IO abstract words and IO nonwords; for Experiment 1B, 10 concrete words and IO nonwords; for Experiments 2A and 28, IO abstract words and 10 concrete words). Prior to SLDT, participants completed practice trials consisting of 30 stimuli (IO abstract words, IO concrete words, and IO nonwords). During the practice trials, participants were monitored to insure they were doing the task correctly. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 23 LDT. Experiments lA and lB each consisted of 300 stimuli. In Experiment lA, half of the stimuli were abstract words (e.g. , justice) and the other half were nonwords (e.g., fraig) . In Experiment lB, half of the stimuli were concrete words (e.g. , oven) and the other half were nonwords. For each stimulus, participants were instructed to decide whether or not it was a real word. If the stimulus was a real word, participants pressed the "?" key. If the stimulus was a nonword, they did not press any key, and the stimulus was removed after the time-out limit was reached. SCT. Experiments 2A and 2B each consisted of 300 stimuli. For both experiments, half of the stimuli were abstract words (e.g. ,justice) and the other half were concrete words (e.g., table) . In Experiment 2A, participants were instructed to decide whether or not the word was abstract. If the word was abstract, participants pressed the "?" key. If the word was concrete, they did not press any key, and the stimulus was removed after the time-out limit was reached. In Experiment 2B, the same procedure was followed , but participants made key presses to the concrete words, and no key presses to the abstract words. SLDT. Experiments 3A and 3B each consisted of the abstract words, concrete words, and nonwords, for a total of 450 stimuli. In Experiment 3A, participants were instructed to decide whether or not the stimulus was a word, and if it was a word whether or not it was an abstract word. If the stimulus was not a word, participants pressed the "z" key; if the stimulus was an abstract word, participants pressed the "?" key; finall y, if the stimulus was a concrete word, they did not press any key, and the stimulus was removed after the time-out limit was reached. For Experiment 3C, if the stimulus was not a word, participants pressed the "z" key; if the stimulus was a concrete word, participants pressed the "?" key; and if the stimulus was SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 24 an abstract word, they did not press any key, and the stimulus was removed after the time-out limit was reached. Data Analyses Preliminary data cleanup was done in the following manner. Any participant who made errors in excess of 30% was removed from the analysis. Additional participants were added if necessary to insure a total of 30 participants in each experiment. 3 Additionally, for each experiment, any word with an error rate in excess of 30% was removed from the analyses of that experiment. The following procedure was used in identifying outliers. First, any response latency less than 250 ms or greater than 2,000 ms for Experiments I and 2, or less than 250 ms or greater than 2,500 ms for Experiment 3, were removed from the analyses. Second, for each participant, any response latency exceeding 2.5 SDs above or below their mean was considered an outlier and removed from the analyses. Response errors were also removed from the correct latency data sets. Following data cleanup, the raw response latencies were z-transformed for the response latency analyses. Separate analyses were conducted on response latencies and errors. As noted, for each experiment, a test of interaction was conducted. Following each test of interaction, hierarchical multiple regression analyses were conducted on the response latency and error data from each experiment separately. This statistical procedure provides the following analyses. First, it allows for all variability associated with the control predictor variables (e.g., frequency, age of acquisition, OLD, concreteness, etc.) to be accounted for in the first step of the analysis. Second, in the second step of the analysis the change in the amount of accounted for variability (!1R 2 and sr2 both represent this amount of accounted for SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 25 variability) due to EE can be measured. Third, at step 2 of the analysis, the sign of sr indicates whether the effect of EE is facilitatory or inhibitory. The focus of discussion will be on the second step in each analysis as described below. The order of entry for the predictor variables for each experiment were as follows . For step 1 the following predictor variables were entered: Log HAL frequency, age of acquisition, orthographic Levenshtein distance, length, phonemes, syllables, and morphemes, inverse N-Count, semantic diversity, concreteness, imageability, BOI, arousal , negative and positive valence, and CA. Importantly, the inclusion of these predictor variables at step 1 removes any response latency and response error variance associated with them. For step 2, EE was entered. This analysis set-up allows for the most stringent tests of the effects of EE on abstract and concrete word processing through the mechanisms of semantic feedback and semantic processing. Means and standard deviations for the predictor and criterion variables of the present study, for the abstract and concrete words, are shown in Table I. Due to differential error rates across experiments, different stimuli were removed from each experiment. Thus each experiment could have a unique corresponding table of zero-order correlations for predictor and criterion variables. However, the values in each table are negligibly different. Therefore, only two tables will be presented here: Table 2 contains zero-order correlations for all abstract stimuli , and Table 3 contains zero-order correlations for all concrete stimuli. The Results sections will occur in the following manner. First, the primary hypotheses for each task will be presented, followed by the secondary hypotheses. Discussion of the primary hypotheses will be restricted to the effects of EE (at step 2 in each SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 26 of the analyses); other significant effects of interest will be discussed in more detail in the General Discussion section. Experiment 1: Results and Discussion Data Sets For Experiment IA, two words (one high EE word, and one low EE word) were removed from analyses due to experimenter coding error. Thus, a total of 74 high EE and 74 low EE abstract words were included in the analyses. A total of 3 .11 % of the data were considered outliers and removed from the data set. For Experiment I B, a total of 3.69% of the data were considered outliers and removed from the data set. Omnibus Analysis As shown in Tables 4 and 5, no significant L1R2 was found at step 2 of the omnibus interaction regression analyses for either response latencies (L1R 2 = .00, p = .827) or response errors (L1R 2 = .00, p = .686). These results do not support the hypothesis that there would be different effects of EE for abstract and concrete words in LDT, but instead indicates that they were similar. Follow up tests were conducted to determine the magnitude and direction of the sr for each word type to better understand the effects of EE in LDT. Experiment lA: Abstract LDT. Refer to Table 6 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings . First, at step 2, there was a significant !1.R 2 (.04, p < .00 I), indicating that EE accounted for a significant amount of unique response latency variability above and beyond that associated with all the preceding predictor variables entered into the analysis. Second, the sr associated SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 27 with EE at step 2 was negative (-.19). This finding shows that higher EE ratings for abstract words were associated with faster response latencies. Regarding the response error analysis, the addition of EE at step 2 accounted for no additional variability (~R 2 =.00, sr = -.01 , both ps = .818). This finding indicates that higher EE ratings for abstract words were not associated with response errors, after the effects of the other predictor variables were statistically controlled. Experiment lB: Concrete LDT. Refer to Table 7 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings. First, at step 2, there was a significant ~R2 (.0 I , p = .051 ), indicating that EE accounted for a significant amount of unique response latency variability. Second, the sr associated with EE at step 2 was negative (-.08), indicating that higher EE ratings for concrete words were associated with faster response latencies in LDT. Regarding the response error analysis, the addition of EE at step 2 accounted for no additional variability (~R 2 =.00, sr = .02, both ps = .742), indicating that higher EE ratings for concrete words were not associated with response errors in LDT, after the effects of the other predictor variables were statistically controlled. In summary, the results of Experiments IA and IB suggest that EE has a significant facilitatory influence on LDT response latencies for both abstract and concrete words. According to the semantic feedback framework, these facilitatory effects arise through the semantic feedback mechanism. These results help extend Vigliocco et al. ' s (2009) semantic representation framework regarding the effects of emotion knowledge for abstract and concrete words in the following manner. In LDT, the task demands are to determine if the SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 28 stimulus is a word or not. EE facilitates this process for both abstract and concrete words because emotion knowledge is relevant to LDT task demands for both types of words. Although the interaction was not statistically significantly different, the effect of EE was numerically larger for abstract as compared to concrete words, which is qualitatively consistent with Vigliocco et al.'s framework. Experiment 2: Results and Discussion Data Sets For Experiment 2A, two words (one high EE word, and one low EE word) were removed from analyses due to experimenter coding error. Any word that exceeded an error rate of 30% across participants was removed from analysis. By this method, an additional seven low EE words, and one high EE word were removed (refer to the Appendix for the specific words that were removed). Thus, a total of 72 high EE and 67 low EE abstract words were included in the analyses . A total of 2.43% of the data were considered outliers and removed from the data set. For Experiment 2B, seven high EE concrete words, and one low EE concrete word were removed (refer to the Appendix for the specific words that were removed) due to error rates exceeding 30%. Thus, 68 high EE and 74 low EE concrete words were included in the analyses. A total of 4.39% of the data were considered outliers and removed from the data set. Omnibus Analysis As shown in Tables 4 and 5, a significanU IR 2 was found at step 2 of the omnibus interaction regression analyses for both response latencies (L1R 2 = .05, p < .001) and response errors (L1R 2 = .03, p = .001 ). These results support the hypothesis that the pattern of EE effects is different for abstract and concrete words in SCT. To provide additional insight into SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 29 the effects of emotion knowledge in SCT, follow-up hierarchical multiple regression analyses were conducted to understand the specific magnitude and direction of the sr for each word type. Experiment 2A: Abstract SCT. Refer to Table 8 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings . First, at step 2, there was a significant t...R 2 (.06, p < .001 ), indicating that EE accounted for a significant amount of unique response latency variability. Second, the sr associated with EE at step 2 was negative (-.25), indicating that higher EE ratings for abstract words were associated with faster response latencies in SCT. Regarding the response error analysis, at step 2 there was a significant t...R 2 (.03 , p = .020), indicating that EE accounted for a significant amount of unique response error variability. The sr associated with EE at step 3 was negative (-.17), indicating that higher EE ratings for abstract words were associated with fewer response errors in SCT. Experiment 2B - Concrete SCT. Refer to Table 9 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings . First, at step 2, there was a significant t...R 2 (.03 , p = .002), indicating that EE accounted for a significant amount of unique response latency variability. Second, the sr associated with EE at step 2 was positive (.16), indicating that higher EE ratings for concrete words were associated with slower response latencies. Regarding the response error analysis, at step 2 SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 30 there was a significant !iR 2 (.02, p = .007), indicating that EE accounted for a significant amount of unique response error variability. The sr associated with EE at step 2 was positive ( .14 ), indicating that higher EE ratings for concrete words were associated with greater response errors in SCT. In summary, the task demands of SCT require participants to make a decision regarding the semantic category of a presented word (i.e., is this word abstract/concrete?). The results of Experiment 2A suggest that EE is a variable that facilitates this decision if the task is to identify abstract words. The results of Experiment 2B suggest that EE inhibits this decision if the task is to identify concrete words. These results suggest that EE is indicative of abstractness for two reasons. First, when task demands (i.e., "is the word abstract?") are congruent with what EE is indicative of (i.e., abstractness), then responding is facilitated for abstract words rated higher in EE. Conversely, when task demands (i.e. , "is the word concrete?") are incongruent with what EE is indicative of (i.e., abstractness), then responding is inhibited for concrete words rated higher in EE. Vigliocco et al. ' s (2009) semantic representation framework suggests that emotion knowledge is important to the representation of abstract words, but may not be as important when processing concrete words. The results of Experiment 2 extend the semantic representation framework by suggesting that emotion knowledge is not necessarily more important for abstract words than for concrete words, but rather exerts qualitatively different effects on these two types of words. Taken together, the facilitatory effect of EE for abstract words in Experiment 2A and the inhibitory effect of EE for concrete words in Experiment 2B suggest that emotion knowledge is indicative of abstractness in SCT. Thus, the findings from SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 31 Experiment 2 help clarify the influence of emotion knowledge, such that this type of knowledge, as measured by EE, exerts different effects depending on both the nature of the word and task demands. Experiment 3: Results and Discussion Data sets For Experiment 3A, any word that exceeded an error rate of 30% across participants was removed from analysis. By this method, three high EE abstract words and twelve low EE abstract words were removed (refer to the Appendix for the specific words that were removed). Thus, a total of 72 high EE and 63 low EE abstract words were included in the analyses. A total of 4.99% of the data were considered outliers and removed from the data set. For Experiment 3B, nine high EE concrete words, and two low EE concrete word were removed (refer to the Appendix for the specific words that were removed) due to error rates exceeding 30%. Thus, 66 high EE and 73 low EE concrete words were included in the analyses. A total of 4.31 % of the data were considered outliers and removed from the data set. Omnibus Analysis As shown in Tables 4 and 5, a significantLIR 2 was found at step 2 of the omnibus interaction regression analyses for both response latencies (.10, p < .001) and response errors (.12, p < .001). These results support the hypothesis that the pattern of EE effects is different for abstract and concrete words in SLDT. To provide additional insight into the effects of emotion knowledge in SLDT, follow-up hierarchical multiple regression analyses were conducted to understand the specific magnitude and direction of the sr for each word type. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 32 Experiment 3A: Abstract SLDT. Refer to Table 10 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings. First, at step 2, there was a significant t-..R 2 (.08, p < .001), indicating that EE accounted for a significant amount of unique response latency variability. Second, the sr associated with EE at step 2 was negative (-.29), indicating that higher EE ratings for abstract words were associated with faster response latencies in SLDT. Regarding the response error analysis, at step 2 there was a significant f...R 2 (.13, p < .001 ), indicating that EE accounted for a significant amount of unique response error variability. The sr associated with EE at step 2 was negative (-.35), indicating that higher EE ratings for abstract words were associated with fewer response errors in SLDT. Experiment 3B: Concrete SLDT. Refer to Table 11 for the results of the hierarchical multiple regression for response latencies and response errors. Regarding the response latency analysis, there are two important findings. First, at step 2, there was a significant f...R 2 (.04, p < .001 ), indicating that EE accounted for a significant amount of unique response latency variability. Second, the sr associated with EE at step 2 was positive (.21 ), indicating that higher EE ratings for concrete words were associated with slower response latencies in SLDT. Regarding the response error analysis, at step 2 there was a significant f...R 2 (.09,p < .001), indicating that EE accounted for a significant amount of unique response error variability. The sr associated with EE at step 2 was positive (.30), indicating that higher EE ratings for concrete words were associated with greater response errors in SLDT. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 33 In summary, the task demands in SLDT require participants to decide whether presented Jetter strings are real English words (i.e. , is this a word?), and if so, whether they are abstract words (Experiment 3A) or concrete words (Experiment 3B). The results of Experiment 3 are consistent with those of Experiment 2, and can be accounted for in the same way. That is, they suggest that EE is indicative of abstractness, such that when the nature of the word is congruent with task demands (i.e., Experiment 3A) faster and more accurate responses result. Conversely, when the nature of the word is incongruent with task demands (i.e. , Experiment 3B), slower and less accurate responses result. Secondary Hypotheses To the extent that the semantic feedback and semantic processing mechanisms are independent of each other, and that their effects are measureable throughout the duration of the experimental trials in SLDT, it was hypothesized that their effects should be combined. For the abstract words this reasoning led to the hypothesis that there should be larger facilitatory EE effects in Experiment 3A than in Experiments IA or 2A. Neither the response latency nor the response error multiple regression analyses strictly supported this hypothesis. For the abstract words, the effects of EE did not combine for response latencies (i.e., sr' s = -.19 and -.25 for the LDT and SCT did not combine in the observed sr of -.29 for the SLDT), or response errors (i.e., sr' s = -.01 and -.17 for the LDT and SCT did not combine in the observed sr of -.35 for the SLDT). For the concrete words it was hypothesized that roughly equivalent inhibitory EE effects should be observed in Experiments 2B and 3B (with a null effect in Experiment 1B). Neither the response latency nor the response error multiple regression analyses strictly supported this hypothesis. For the concrete words, the effects of SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 34 EE did not combine for response latencies (i.e., sr' s = -.08 and .16 for the LDT and SCT did not combine in the observed sr of -.21 for the SLDT), or response errors (i.e., sr's = .02 and .14 for the LDT and SCT did not combine in the observed sr of .30 for the SLDT). One possible reason why the above secondary hypotheses were not strictly supported by the data is the following. In SLDT, there are two task demands. The lexical component involves the semantic feedback mechanism, whereas the semantic component involves the semantic processing mechanism. It may be that the EE effects due to the lexical component of SLDT (i.e., due to semantic feedback) were no longer measureable by the time responses were made, and thus any observed EE effects in SLDT were primarily due to the semantic component (i.e., due to semantic processing). A second possible reason the secondary hypotheses were not supported in the response latency analyses may have to do with the following. The results of the SLDT error analyses provide important information regarding the interaction of EE and task demands . For abstract words, there was no effect of EE in LDT (sr = -.01 , p = .816), but there was a facilitatory effect of EE in SCT (sr = -.17, p < .01 ), and a notably larger facilitatory effect of EE in SLDT (sr = -.35 , p < .001). This analysis suggests that EE has a greater influence on correct responding in SLDT as compared to LDT and SCT. More specifically, in SLDT, as abstract words decrease in EE, the more likely they are to be responded to incorrectly, and thus removed from the response latency analysis. Importantly, there is a strong positive relationship between errors and response latencies in SLDT (r = .67, p < .01 ; see Table 2). If lower rated EE abstract words were prone to produce more errors (due to low abstract EE words being less indicative of abstractness), the latencies of correct responses to those lower SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 35 rated EE abstract words would be relatively higher as compared to higher rated EE abstract words (due to high EE abstract words being more indicative of abstractness). For example, the lower rated EE abstract word,jeopardy, produced nine errors and a mean response latency of 1,499 msec, whereas the higher rated EE abstract word, honour, produced no errors and a mean response latency of 997 msec . Therefore, it is suggested that if the stimuli prone to produce more errors (i .e., low EE abstract words) had more correct trials included in the SLDT latency analysis, a larger sr may have been observed in SLDT, and the hypothesized combined effect of EE on the semantic feedback and semantic processing mechanisms may have been observed. General Discussion In contrast to classical cognitive science, grounded cognition suggests that cognition is fundamentally based in perception and action (e.g., PSS ; Barsalou, 1999). Further developments in grounded cognitive theory have led researchers to suggest the idea of situated conceptualization (Barsalou, 2005 , 2009), an extension of PSS that offers an account of how conceptual knowledge may be acquired during bodily experience within different environmental contexts. Specifically, internal contextual elements unique to environmental situations, including emotion states, are largely used to ground abstract concepts, whereas external contextual elements, including sensorimotor interactions with objects and agents, are largely used to ground concrete concepts. A large body of research supports the idea that concrete concepts are grounded in sensorimotor knowledge, yet there is less research that has investigated the grounding of abstract concepts. Vigliocco et al. (2009) suggested that abstract concepts may largely use SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 36 emotion knowledge as a platform for grounding. More specifically, Vigliocco et al. ' s semantic representation framework suggests that emotion knowledge is more heavily involved in abstract conceptual processing as compared to concrete conceptual processing, which largely involves sensorimotor knowledge. To test the notion that abstract conceptual processing relies on emotion knowledge, the influence of EE on the two mechanisms of the semantic feedback framework was measured. The first mechanism is semantic feedback, which occurs when semantic unit activation feeds back onto orthographic units, and is proposed to be the principle semantic mechanism underlying the influence of semantics on performance in LDTs (as examined in Experiments IA and IB). The second mechanism is semantic activation, which is the degree of semantic unit activation, and is proposed to be the principle semantic mechanism underlying the influence of semantics on performance in SCTs (as examined in Experiments 2A and 2B). Both mechanisms were examined in a single task, the SLOT, in Experiments 3A and 3B. Experiments lA and lB: LDT Significant facilitatory effects of EE were observed in both Experiments I A and I B. The results from Experiment IA support the primary hypothesis that abstract words recruit emotion knowledge, in this case as measured by EE, during LDT processing, in which greater semantic feedback facilitates LDT responding. This finding is consistent with Vigliocco et al. ' s (2009) semantic representation framework, which proposes that emotion knowledge underlies the representation of abstract words. In Experiment I B, the facilitatory effect of EE on concrete word identification suggests that emotion knowledge is also relevant SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 37 to concrete word processing. Taken together, Experiments IA and 1B indicate that EE is an important dimension of emotion knowledge that is used when task demands require decisions as to whether the stimulus is a word or not, regardless of word type. However, the influence is larger for abstract words than for concrete words (i1R2 = 3.61 % and 0.64%, respectively) . Experiments 2A and 2B: SCT A significant interaction was observed between Experiments 2A and 2B for both latency and errors, and is consistent with the primary hypotheses : the effects of EE on SCT were different for abstract words as compared to concrete words. More specifically, significant facilitatory effects were observed for Experiment 2A, whereas significant inhibitory effects were observed for Experiment 2B. This set of results suggests that when categorizing words along an abstract-concrete continuum (i .e., judging whether they are either abstract or concrete), EE is a dimension of emotion knowledge that influences, in fundamentally different ways, those categorizations. Abstract words rated higher in EE are responded to faster and more accurately than abstract words rated lower in EE. Experiment 2A supports the semantic representation framework (Vigliocco et al. , 2009), such that emotion knowledge facilitates the processing of abstract words. Experiment 2B extends the semantic representation framework by suggesting that EE is also used in the processing of concrete words, but inhibits, rather than facilitates, the processing of such words. This is because relatively high EE ratings for concrete words provides evidence to the word recognition system that that word may be abstract, because EE is indicative of abstractness. Thus, the concreteness decision is more difficult for concrete words with relatively high EE ratings, leading to either longer response latencies or higher SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 38 error rates, because more time is needed to resolve the ambiguity (or an error is made before such a resolution) that the word contains evidence that is indicative of it being concrete (i .e., higher ratings of concreteness, imageability, BOI, etc.) and other evidence that is indicative of it being abstract (i .e., higher EE ratings). Taken together, Experiments 2A and 2B indicate that EE is an important dimension of emotion knowledge that is used when task demands require decisions as to whether words are abstract or concrete. That is, EE is indicative of abstractness. On one hand, when EE is congruent with task demands (i.e., in Experiment 2A) facilitatory effects are observed. On the other hand, when EE is incongruent with task demands (i.e., Experiment 2B) inhibitory effects are observed. Experiments 3A and 3B: SLDT The distinguishing feature of Experiments 3A and 3B is that the task demands are a combination of Experiments 1 and 2. By first making a lexical decision (i.e. , is the stimulus a word?), followed, if necessary, by making a semantic decision (i .e., "is the word abstract?"; "is the word concrete?"), the SLDT is in effect tapping into both the semantic feedback and semantic activation mechanisms. In Experiments 3A and 3B, consistent with the primary hypotheses, a significant interaction was observed, suggesting that the effects of EE in SLDT were different for abstract words as compared to concrete words. More specifically, facilitatory EE effects were observed for the abstract SLDT, and inhibitory EE effects were observed for the concrete SLDT. As with Experiment 2, the SLDT results indicate that EE is indicative of abstractness. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 39 In Experiment 3A, abstract words rated higher in EE were responded to faster and more accurately than abstract words rated lower in EE. This result supports the semantic representation framework (Vigliocco et al., 2009), such that emotion knowledge is indeed used, and facilitates the processing of abstract words. In Experiment 3B, concrete words rated higher in EE were responded to slower and less accurately as compared to concrete words rated lower in EE. Experiment 3B extends the semantic representation framework by suggesting that EE does indeed influence the processing of concrete words, by inhibiting, rather than facilitating, the processing of such words. To reiterate, if higher EE ratings associated with concrete words provide evidence to the word recognition system that that word may be abstract, because EE is indicative of abstractness. Thus, the concreteness decision is more difficult for concrete words with relatively high EE ratings, leading to either longer response latencies or higher error rates, as was the case in Experiment 2B. Taken together, Experiments 3A and 3B indicate that EE is an important dimension of emotion knowledge that is used when part of the task demands require decisions as to whether words are abstract or concrete (i.e., the semantic categorization component of the SLDT). That is, EE is indicative of abstractness. On one hand, when EE is congruent with task demands (i .e. , in Experiment 3A) facilitatory effects are observed. On the other hand, when EE is incongruent with task demands (i.e. , Experiment 3B) inhibitory effects are observed. Interestingly, these results arise even when a lexicality decision is required prior to the semantic decision being made, which suggests that EE exerts strong effects on semantic processing regardless of the exact nature of the task demands. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 40 Secondary Hypotheses Recall that the secondary hypotheses suggested that to the extent that the semantic feedback and semantic processing mechanisms are independent of each other, and that their effects are measureable throughout the duration of the experimental trials of the SLDT, the effects of EE in LDT and SCT should be combined in SLDT. For abstract words, the expectation was that facilitatory EE effects should be larger in SLDT than in LDT or SCT. For concrete words, however, the expectation was that inhibitory EE effects should be roughly similar in SLDT and SCT, but both larger than in LDT. The secondary hypotheses for both types of words were not strictly supported by the data. As discussed above, a possible explanation is that the observed EE effects in SLDT were due to semantic processing, because by the time responses were made any EE effects attributable to semantic feedback were no longer measurable and therefore not represented in the response latency data. The error analyses provide additional insight regarding the interaction of EE and task demands, suggesting that EE has a greater influence on correct responding in SLDT as compared to LDT and SCT. The hypothesized combined effect of the semantic feedback and semantic processing mechanisms in SLDT on the effects of EE for response latencies may have been observed if a greater number of low EE abstract words had been included in the multiple regression latency analyses. The notably larger EE effects observed in the SLDT error analyses for both abstract (larger facilitatory effects) and concrete words (larger inhibitory effects) provides some evidence that under more difficult task demands, the saliency of EE effects is increased. SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 41 The non-additive effects of EE in the SLDT of the present thesis are inconsistent with the reported additive effects of BOI in the same task by Siakaluk et al. (2008b). There may be one or more possible reasons for this inconsistency. First, there may be differences in the two variables themselves. That is, perhaps BOI exerts stronger effects on semantic feedback than EE, and thus additive BOI effects are measurable in SLDT, whereas additive EE effects are not. Second, different procedures were used to measure BOI and EE effects in the two studies. More specifically, Siakaluk et al. (2008b) experimentally controlled for the effects of confound variables such that their high and low BOI word lists differed only on BOI, and this factorial design allowed for a relatively strong investigation of the effects of this variable. In contrast, in the present thesis the effects of EE were measured only after any shared variance with the control variables and EE was statistically removed through the procedure of hierarchical multiple regression. Such a procedure may not provide for as strong an investigation of the effects of EE in SLDT as the procedure used by Siakaluk et al. (2008b) for BOI. Finally, the results of the secondary hyopthesis are suggestive that the SLDT is, in general, a more difficult task as compared to the LDT and SCT. This notion is important methodologically as it may prove to be an important tool for future research should the researchers desire a word recognition task requiring additional cognitive load. Semantic Feedback Framework According to the semantic feedback activation framework, two unique mechanisms account for the effects of EE in the present thesis: semantic feedback and semantic SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 42 processing. These mechanisms interact with variations of the semantic richness of the stimuli and the task demands. According to this framework, correct LDT responses to words occurs when orthographic units settle on a correct representation (or in PSS terms, settles on a correct simulation), with faster settling being associated with faster responses. Importantly, if the presented word is relativel y more semantically rich, a greater amount of semantic units will activate, consequently sending greater semantic feedback to the orthographic units. In Experiments IA and 18, it was observed that words (both abstract and concrete) rated higher in EE were responded to faster than words rated lower in EE. As such, according to the semantic feedback framework, EE increases the semantic richness of words (because higher EE is associated with greater emotional bodily experience), thus increasing the amount of semantic feedback, and consequently facilitating LDT responding. According to the framework, responses in SCT primarily depend on settling amongst the semantic units (i .e., semantic processing). Words that are relatively more semantically rich generate greater activation among the semantic units that can either quicken or delay settling on a particular semantic activation pattern (or simulation), which is then associated with faster or slower responding in the SCT, depending on task demands. In Experiments 2A and 28, both abstract and concrete words rated higher in EE would activate more semantic units than abstract and concrete words rated lower in EE. Of critical importance is the nature of the task demands. In Experiment 2A, the criterion was to decide whether words were abstract, and higher ratings of EE would be associated with greater semantic unit activation and faster and more accurate responses, because EE is indicative of abstractness and hence congruent with these task demands. This pattern of results were observed in Experiment 2A. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 43 In Experiment 2B, the criterion was to decide whether words were concrete, and higher ratings of EE would be associated with greater semantic unit activation and slower and less accurate responses, because EE is indicative of abstractness and hence incongruent with these task demands. This pattern of results were observed in Experiment 2B. In SLDT, responding depends both on the semantic feedback and semantic processing mechanisms . As noted in the above paragraph, the relatively greater amount of semantic unit activation for abstract words rated higher in EE facilitated responding (Experiment 3A), because semantic processing and task demands were congruent; whereas, the relatively greater amount of semantic unit activation for concrete words rated higher in EE inhibited responding (Experiment 3B), because semantic processing and task demands were incongruent. The lack of a combined effect due to the influence of EE on both semantic mechanisms is likely due to any EE effects attributable to the semantic feedback mechanism no longer being measureable by the time responses were made. Semantic Representation Framework The semantic representation framework (Vigliocco et al. , 2009) suggests that knowledge gained through emotion experience primarily underlies the meanings of abstract words, whereas sensorimotor experience primarily underlies the meanings of concrete words. The results of the present thesis extend the semantic representation framework by suggesting that EE is a dimension of emotion knowledge that underlies the representation and processing of both abstract and concrete words. The results of Experiments 2 and 3 support this notion. The facilitatory effects observed in Experiments 2A and 3A indicate that EE influences processing of abstract words in one direction when what it is indicative of (i .e. , SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 44 abstractness) is congruent with task demands (i.e. , deciding whether a word is abstract). These facilitatory EE effects are consistent with the semantic representation framework. Importantly, however, the inhibitory effects in Experiments 2B and 3B indicate that EE also influences processing of concrete words in a different direction when what it is indicative of (i.e. , abstractness) is incongruent with task demands (i.e., deciding whether a word is concrete). This second finding extends the semantic representation framework regarding how emotion knowledge, at least as measured by EE, interacts with the processing of concrete words. Taken together, the facilitatory EE effects for abstract words and the inhibitory EE effects for concrete words, for tasks whose demands include abstract/concrete categorizations, provide converging evidence that EE is indicative of abstractness. Thus, the results of the present thesis suggests that both abstract and concrete words that are rated relatively higher in EE are represented as being more abstract, relative to abstract and concrete words that are rated relatively lower in EE, an important extension to the semantic representation framework. Although, as noted above, that EE is indicative of abstractness for both abstract and concrete words, the degree of this influence was not quite equivalent for the two types of words. In Experiment 2, the l1R2 was 6.25% and 2.56% for abstract and concrete word latencies, and 2. 89% and 1.96% for abstract and concrete word errors, respectively. In Experiment 3, the i1R2 was 8.41 % and 4.41 % for abstract and concrete word latencies, and 12.25% and 3.61 % for abstract and concrete word errors, respectively. Thus, it can be seen that EE accounts for roughly two to two-and-a-half times more unique latency variability, and one-and-a-half to three times more unique error variability for abstract words than for concrete words. Evidently, EE is a dimension of emotion knowledge that influences both SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 45 abstract and concrete words, but the effects of EE are greater for abstract words as compared to concrete words. Although the effects of EE are putatively larger for abstract words than for concrete words, it is proposed that EE is still a reliable indicator of abstractness for concrete words for the following reason. Concrete words are associated with higher ratings of sensorimotor and contextual variables that are indicative of concreteness, such as concreteness, imageability, BOI, and CA (see Table 1). These variables exert facilitatory effects because they are indicative of concreteness and are congruent with the task demands of deciding if words are concrete (see Tables 9 and 11). Regardless of the many just noted sources of evidence that concrete words are in fact concrete, significant inhibitory effects of EE for concrete words were still observed in SCT and SLDT. Thus, although, in the present thesis, EE may influence concrete words to a lesser degree than abstract words, the inhibitory EE effects observed in SCT and SLDT is an important piece of converging evidence that EE is indicative of abstractness. Positivity, Negativity, Arousal, Context Availability, and Emotional Experience It is important to note that the statistical analyses used for each experiment were chosen with the intention of providing the control predictor variables the greatest opportunity to account for response latency and error variability prior to the inclusion of EE in the analyses. Using this statistical procedure, tests for the effects of EE were very stringent. This is particularly evident in the LDTs, where large amounts of variability were accounted for in the preliminary step (69% in the abstract LDT and 77% in the concrete LDT). By adding EE to the analyses after the control predictor variables, it is impressive that EE influenced LDT latency performance at all, especially for the concrete words. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 46 One set of predictor variables that are of particular interest are the emotion variables commonly used in research examining the effects of emotion knowledge (i.e., positivity, negativity, and arousal) in visual word recognition. Recall the inconsistency in the literature: some studies have shown that negative words are responded to slower on average than positive words (e.g., Estes & Adelman, 2008), whereas other studies have shown that both positive and negative words are responded to faster than neutral words (e.g. , Kousta et al. , 2009). The purpose of the present study was not to address this inconsistency directly, however, with the inclusion of negativity, positivity, and arousal in the preliminary step of the multiple regression analyses, the significant effects of EE suggest that EE is to some extent measuring a dimension of emotion knowledge that is different from positivity, negativity, or arousal. An important aspect of EE distinguishing it from positivity, negativity, and arousal was discussed in Siakaluk et al. (2014), who suggested that although concepts (specifically abstract concepts) may have specific emotion features, such as being positive or negative, there are core emotion features that may be shared across different contexts or situations. For example, the abstract concept [COURAGE] may exist in many contextual situations (e.g., soldiers having courage in battle, public speakers having courage to speak in front of a crowd, finding the courage to ask for a raise, etc.). Although there are clear differences in each of these situations, there may be core aspects of [COURAGE] that is nonetheless shared in each of them. More specifically, one of the core aspects may be doing something that is risky in the given situation, and a second core aspect may be that not every individual would be willing to act in such a way (among other potential core aspects). As individuals encounter an increasing number of situations where [COURAGE] is experienced, more knowledge SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 47 becomes available for future processing of [COURAGE]. This emotion knowledge, based on core aspects shared across situations relating to [COURAGE] , is what EE is hypothesized to capture. Another (somewhat related) predictor variable of interest is CA. According to the notion of situated conceptualization, two categories of contextual knowledge may be involved in conceptual processing: contextual knowledge internal to the agent and contextual knowledge external to the agent. Although EE is suggested to be a dimension of contextual knowledge internal to the agent, specifically as a measure of emotion knowledge derived through bodily experience with the environment (as explained in the paragraph above), CA has been suggested to measure contextual knowledge external to the agent (Moffat et al. , 2015). The inclusion of CA as a control predictor variable in the analyses is theoretically important, because CA putatively measures external contextual knowledge, whereas EE putatively measures internal contextual knowledge. Two important findings from the multiple regression analyses regarding the effects of CA and EE (see Tables 6-11) are of note. First, in both the abstract and concrete LDTs, there were facilitatory effects of CA and EE, suggesting that regardless of word type, both CA (a dimension of contextual knowledge external to the agent) and EE (a dimension of contextual knowledge internal to the agent) facilitate the identification of real words in LDTs . Second, when identifying abstract words in the SCT and SLDT, there were facilitatory EE effects but no effects of CA. In contrast, when identifying concrete words in SCT and SLDT, there were inhibitory EE effects but facilitatory CA effects. Taken together, this pattern of effects suggests that EE and CA are SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 48 indeed tapping into different dimensions of contextual knowledge derived from bodily experience. Conclusions The results from Experiment I suggest that when task demands require a lexical decision, EE facilitates responding for both abstract and concrete words, because EE represents evidence that the stimulus is a word. Thus, in LDT, EE exerts facilitatory effects via the semantic feedback mechanism. The results from Experiments 2 and 3 suggest that when task demands require a semantic decision regarding the abstractness of a word, EE facilitates responding; however, when task demands require a semantic decision regarding the concreteness of a word, EE inhibits responding. These latter two separate, yet converging findings from Experiments 2 and 3 lead to the inference that EE is indicative of abstractness. Thus, in SCT and SLDT in which abstract or concrete semantic decisions are required, EE exerts either facilitatory or inhibitory effects via the semantic processing mechanism as a result of task demands. The findings from the present thesis help extend the influence of emotion knowledge as currently stipulated in the semantic representation framework (Vigliocco et al., 2009) in such a way that considerations of task demands are vitally important. Finally, EE is suggested to capture core, internal emotion knowledge derived across different situated contexts. SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 49 References Balota, D. A., Yap, M. J., Hutchison, K. A. , Cortese, M. J. , Kessler, B., Loftis, B., ... & Treiman, R. (2007). The English lexicon project. Behavior Research Methods , 39, 445-459. Barsalou, L. W. (1999). Perceptions of perceptual symbols. Behavioral and Brain Sciences , 22, 637-660. Barsalou, L. W. (2005). Situated Conceptualization. In Cohen, H. & Lefebvre, C. (Eds.), Handbook of Categorization in Cognitive Science (pp. 619-650). Oxford: Elsevier. Barsalou, L. W. (2009). Simulation, situated conceptualization, and prediction. Philosophical Transactions of the Royal Society B: Biological Sciences, 364, 1281-1289. Bennett, S. D., Burnett, A. N. , Siakaluk, P. D. , & Pexman, P. M. (2011). Imageability and body-object interaction ratings for 599 multisyllabic nouns. Behavior Research Methods, 43 , 1100-1109. Casasanto, D. (2009). Embodiment of abstract concepts: good and bad in right-and lefthanders. Journal of Experimental Psychology: General, 138, 351. Cortese, M. J., & Fugett, A. (2004). Imageability ratings for 3,000 monosyllabic words. Behavior Research Methods, Instruments, & Computers , 36, 384-387. Damasio, A. R. (1989). Time-locked multiregional retroactivation: A systems-level proposal for the neural substrates of recall and recognition. Cognition, 33 , 25-62. Estes, Z., & Adelman, J. S. (2008). Automatic vigilance for negative words in lexical decision and naming: Comment on Larsen, Mercer, and Balota (2006). Emotion, 8, 441-444. Friendly, M., Franklin, P. E., Hoffman, D. , & Rubin, D. C. (1982). The Toronto Word Pool: SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 50 Norms for imagery, concreteness, orthographic variables, and grammatical usage for 1,080 words. Behavior Research Methods & Instrumentation, 14, 375-399. Fodor, J. A. (1975). The language of thought. Harvard University Press. Forster, K. I. (1976). Accessing the mental lexicon. New approaches to language mechanisms, 30, 231-256. Gibbs Jr, R. W. (2005). Embodiment in Metaphorical Imagination. In R.A. Zwann & D. Pecher (Eds.), Grounding Cognition (115-128). Cambridge, New York: Cambridge University Press. Glenberg, A. M., & Kaschak, M. P. (2002). Grounding language in action. Psy chonomic Bulletin & Review, 9, 558-565 . Glenberg, A. M., & Robertson, D. A. (2000). Symbol grounding and meaning: A comparison of high-dimensional and embodied theories of meaning. Journal of memory and language, 43(3), 379-401. Hamad, S. (1990). The symbol grounding problem. Physica D : Nonlinear Phenomena, 42(1), 335-346. Hino, Y., & Lupker, S. J. (1996). Effects of polysemy in lexical decision and naming: An alternative to lexical access accounts. Journal of Experimental Psy chology: Human Perception and Performance, 22(6) , 1331. Hoffman, P., Lambon Ralph, M.A., & Rogers, T. T. (2013). Semantic diversity: A measure of semantic ambiguity based on variability in the contextual usage of words . Behavior Research Methods, 45, 718-730. Karmiloff-Smith, A. (1979). Micro-and macrodevelopmental changes in language acquisition and other representational systems. Cognitive science , 3(2), 91-117. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 51 Kousta, S. T ., Vinson, D. P. , & Vigliocco, G. (2009). Emotion words, regardless of polarity, have a processing advantage over neutral words. Cognition, 112(3), 4 73-481 . Kuperman, V. , Stadthagen-Gonzalez, H. , & Brysbaert, M . (2012). Age-of-acquisition ratings for 30,000 English words. Behavior Research Methods, 44(4) , 978-990. Kuperman, V. , Estes, Z. , Brysbaert, M ., & Warriner, A. B. (2014). Journal of Experimental Psy chology: General. Advance online publication. doi:10.1037/a0035669 Johnson, M. , & Lakoff, G. (1980). Metaphors we live by. Chicago : U of Chicago P. Lakoff, G. , & Johnson, M. (1999). Philosophy in the flesh : The embodied mind and its challenge to western thought. Basic books. Larsen, R. J. , Mercer, K. A. , Balota, D. A., & Strube, M. J. (2008). Not all negative words slow down lexical decision and naming speed: Importance of word arousal. Emotion, 8, 445-452 . Marr, D ., & Vision, A. (1982). A computational investigation into the human representation and processing of visual information. WH San Francisco: Freeman and Company. Moffat, M., Siakaluk, P. D. , Sidhu, D. M., & Pexman, P. M. (2014). Situated conceptualization and semantic processing: effects of emotional experience and context availability in semantic categorization and naming tasks. Psychonomic bulletin & review, 1-12. Neisser, U. (2014). Cognitive Psychology: Classic Edition. Psychology Press. Newcombe, P. I., Campbell, C. , Siakaluk, P. D. , & Pexman, P. M . (2012). Effects of emotional and sensorimotor knowledge in semantic processing of concrete and abstract nouns. Frontiers in human neuroscience, 6. Niedenthal, P. M . (2007). Embodying emotion. Science , 316(5827) , 1002-1005 . SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 52 Paivio, A., Yuille, J. C., & Madigan, S. A. (1968). Concreteness, imagery, and meaningfulness values for 925 nouns. Journal of experimental psychology, 76(1 p2), 1. Parisi , D. (2011). The other half of the embodied mind. Embodied and grounded cognition. Pratto, F., & John, 0. P. (2005). Automatic Vigilance: The Aention-Grabbing Power of negative Social Information. Social cognition: key readings, 250. Recio, G., Conrad, M. , Hansen, L.B., & Jacobs, A. M. (2014). On pleasure and thrill : the interplay between arousal and valence during visual word recognition. Brain and language, 134, 34-43. Searle, J. R. (1980). Minds, brains, and programs. Behavioral and brain sciences, 3(03), 417424. Schwanenflugel, P. J. , Harnishfeger, K. K ., & Stowe, R. W. (1988). Context availability and lexical decisions for abstract and concrete words. Journal of Memory and Language, 27, 499-520. Shaoul, C. , & Westbury, C. (2010b). Neighborhood Density Measures for 5 7, 153 English words. Edmonton, AB : University of Alberta (downloaded from http://www.psych.ualberta.ca/- westburylab/downloads/westburylab.arcs.ncounts.htm I) Siakaluk, P. D. , Pexman, P. M., Aguilera, L., Owen, W. J., & Sears, C.R. (2008). Evidence for the activation of sensorimotor information during visual word recognition: The body-object interaction effect. Cognition, 106(1), 433-443 . SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 53 Siakaluk, P. D., Pexman , P. M., Sears, C.R., Wilson, K., Locheed, K., & Owen, W. J. (2008). The benefits of sensorimotor knowledge: Body-object interaction facilitates semantic processing. Cognitive Science, 32(3), 591-605. Siakaluk, P. D. , Knol, N. , & Pexman, P. M. (2014). Effects of emotional experience for abstract words in the Stroop task. Cognitive science , 38(8), 1698-1717. Strain, E., & Herdman, C. M . (1999). Imageability effects in word naming: an individual differences analysis. Canadian Journal of Experimental Psychology/Revue canadienne de psychologie experimentale, 53(4), 347. Strain, E. , Patterson, K. , & Seidenberg, M. S. (1995). Semantic effects in single-word naming. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(5), 1140. Vermeulen, N., Niedenthal, P. M ., & Luminet, 0. (2007). Switching between sensory and affective systems incurs processing costs. Cognitive Science, 31 (1 ), 183-192. Vigliocco, G., Meteyard, L. , Andrews, M. , & Kousta, S. (2009). Toward a theory of semantic representation. Language and Cognition, 1(2), 219-24 7. Vinson, D., Ponari, M. , & Vigliocco, G. (2014). How does emotional content affect lexical processing?. Cognition & emotion, 28(4), 737-746. Warriner, A. B., Kuperman, V. , & Brysbaert, M. (2013). Norms of valence, arousal, and dominance for 13,915 English lemmas. Behavior Research Methods, 45, 1191-1207. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 54 Wellsby, M ., Siakaluk, P. D., Owen, W. J. , & Pexman, P. M. (2011). Embodied semantic processing: the body-object interaction effect in a non-manual task. Language and Cognition, 3(1), 1-14. Wiemer-Hastings, K. , & Xu, X. (2005). Content Differences for Abstract and Concrete Concept. Cognitive Science, 29(5) , 719-736. Wilkowski, B. M., Meier, B. P., Robinson, M. D. , Carter, M. S., & Feltman, R. (2009). "Hotheaded" is more than an expression: The embodied representation of anger in terms of heat. Emotion, 9(4), 464. Wilson, N. L., & Gibbs, R. W. (2007). Real and imagined body movement primes metaphor comprehension. Cognitive Science, 31(4), 721-731. Yap, M. J ., & Seow, C. S. (2014 ). The influence of emotion on lexical processing: Insights from RT distributional analysis. Psy chonomic bulletin & review, 21(2), 526-533. Yaxley, R.H. , & Zwaan, R. A. (2007). Simulating visibility during language comprehension. Cognition, 105(1 ), 229-236. Zwaan, R. A., & Madden, C. J. (2005). Embodied sentence comprehension. Grounding cognition: The role ofperception and action in memory, language, and thinking, 224245. 55 SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Table I Descriptive statistics and behavioral data for the 75 high EE abstract words, 75 low EE abstract words, 75 high EE concrete words and 75 low EE concrete words Abstract Words Low EE High EE Variable M Concrete Words High EE SD M SD Low EE M SD M SD Log frequency (HAL) 9.12 1.57 9.09 1.73 9.40 1.54 7.79 1.78 Age of acquisition 8.91 2.26 9.79 1.97 6.42 1.98 6.66 1.89 Orthographic Levenshtein distance 2.51 .67 2.42 .59 2.38 .80 2.82 1.02 Letters 7.16 1.79 7.04 1.53 6.69 1.59 7.57 1.86 Phonemes 5.95 1.62 5.97 1.52 5.31 1.53 5.97 1.72 Syllables 2.39 .85 2.36 .71 2.08 .54 2.36 .67 Morphemes 1.56 .68 1.68 .60 1.29 .49 1.56 .72 NCOUNT-INV .16 .36 .24 .42 .09 .28 .34 .47 Semantic Diversity 1.80 .24 1.89 .24 1.62 .50 1.45 .24 Concreteness 2.41 .67 2.84 .68 6.05 .58 6.23 .53 Imageability 3.28 .50 2.77 .50 5.79 .77 5.81 .48 Body-Object Interaction 2.10 .32 2.02 .44 4.75 1.03 5.05 .84 Arousal 4.64 .79 3.89 .58 4.41 .92 3.78 .89 Negativity -.63 .99 -.22 .42 -.41 .8 5 -.11 .27 Positivity .90 .89 .40 .45 1.17 .85 .52 .52 Context Availability 4.64 .63 4.17 .77 5.84 .81 5.81 .60 Emotional experience 4.56 .60 2.30 .42 3.06 .77 1.54 .16 Response latency LDT 633.49 65.32 683.36 76.74 652.60 74.68 646.35 78.83 1.56 4.76 2.44 5.9 1 2.04 4.17 .62 3.53 921.92 88 .24 1031.35 96.35 765.45 86.49 778.92 95.48 Categorization Errors SCT 5.22 4.90 8.28 6.78 2.20 4.52 4.90 6.75 Response latency SLOT 1157.01 119.27 1314.94 131.92 971.67 116.88 986.97 136.67 4.26 5.68 11 .00 7.64 4.05 6.70 7.73 9.02 Categorization Errors LDT Response latency SCT Categorization Errors SLOT Note. LDT= lexical decision task; SCT = semantic categorization task; SLDT = semantic lexical decision task; NCOUNT-INV = inverse of number of word neighbours plus I. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 56 Table 2 Zero-order Correlations Between the Criterion Variables and the Predictor Variables for Abstract Words Measure I 2 3 4 l 6 7 8 9 10 II 12 13 14 ll 16 17 18 19 20 21 22 I. CL LDT 2. Errors LDT .65· 3 . CLSCT scr 49• .32* 4. Errors . 13 . 14 l . CLSLDT .58* _43• .n• .43• 6. Errors SLOT .50* .44* .10• .38* .67* 7. Frequency - .611 "' -.51* -.2 1' 03 -.24* -.20• 8. AoA .66* .45• .21• -.0 1 .3 6* .29* -.54· 9 . OLD .29 . ll .Ol - 04 . 12 .Ol -.32• .29* 10. Letters .40* .2 1• . 16 .04 . ll .09 -.36* .37* .110• I I. Phonemes .3 8* .2 1• .22 03 .20• . IO -.34* ..13* .68* 12. Syllab les .3l .22• .19 -.0 1 .20 • . II -.30* ..t i * .60* .n• .n• 13. Morphe mes .29* . ll .20• .02 . 11• . 19* -.24* .32* .45• .66* .57* .59* 14. INV+ I .52* .51* .19* -.07 .32 .2 8* -.64* .17* . 10 .07 .O& .Ol 15. SemDi,· -.44* -.37* -. 13 -.07 -. 10 -.0& .ss• .3 1* -Al* -.25 * -.31* -.Jo• -.36 * -.30* -.38* 16. Concreteness -.09 -. 13 .35* .H* .23* .29* .20• -.Ol -.24 * -. 19* -. 13 -.20 • -. l l -.19* .24* 17. lmagcabil ity -.24* -.11 -. 19* -.Ol -.2.5* -. 11 .03 -.30* . II .04 -.Ol -.03 -. 10 -.13 -.02 18. BOI -.09 -. l l -.09 .25* -. 13 -.03 -.02 -.17* .01 -.03 .0 1 -.0& -.04 -.04 .03 .22• 19. Aroussal -.18• .02 -.29" -. 14 -.3 1· -. 19· -.03 -. l l -.08 .0 1 .0 1 -.04 -.11 0. 18 -.03 -. 13 .4 1• .23" 20. Negativity -.02 -.0& . 12 19' -.0 1 -.0 1 .22• -.03 -.0 1 -.02 .0 1 . 13 .00 -. 18" .0 1 -.0& -.26" -.Ol -.26" 2 1. Positi\'ity -.26* -. 12 -.2K" -. 10 -.39" -.27' .3 1" -.2k" .0 1 -.07 -.O'J .0 1 -. 18" -.24" .07 •. 22• . 16 -.02 .20" .47" 22. CA -.60" -.58" -.27' . 10 -.34" -. 29' .4 1" -.l9· -.06 -. 13 -. 13 -. 12 •.2 1• -.36" .23" . II .47• .37" . 19" -.0& .22 • 23. EE -.35" -.0& -.ll ' -.32" -.51" -.l l ' .0 1 -.23' .07 .0 1 -.04 -.02 -.11 -.09 -.0& -.33' .5 1" . 10 .SR" -.30" .3X" .47• .113* .o7 .37• .35" *p < .05 Note . CL= categorization latency; Frequency= HAL log-frequency; AoA = age of acquisition ; OLD = orthographic Levenshtein distance; INV+ l = inverse of number of word neighbours plus 1; SemDiv = semantic diversity; BOI = body-object interaction; CA = context availability; EE= emotional experience. SITU A TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 57 Table 3 Zero-order Correlations Between the Criterion Variables and the Predictor Variables for Concrete Words Measure I 2 3 4 l 6 7 8 9 10 II 12 13 14 15 16 17 18 19 20 21 22 I. C L LDT 2. Errors LDT .46• 3. CL SCT .5 1· .24• 4. Errors SCT .22 • . 19· .65· 5. C L SLDT .47• _23• .68· 6 . Errors SLOT .2R• .n• .Rt• .58' .72' .6R• 7. Frequency -.67• •.40• -.11 . II -. 14 - .Ol 8. AoA .63• _35• .66' .3 1' _51• .38' -.34' 9 . OLD .60' .22• . II -.07 . 12 .Ol -.66' .32' 10. Let ters .57' .17' . II -.04 .13 .01 -.6 1' .28' 89 ' I I. Phonemes .S6' . 14 .17' -.03 .ll .03 -.53' .27' .80' .84' 12. Syllab les .52' .07 .10 -.03 .l l .02 -.44' . 17' .72' .71' 13. Morphemes .29' -.02 .01' . _O,l .03 -. 11 •.44• .09 .44' .51' .44' 14. INV+ I .47' .36' -.03' -.11 -.06 -.06 -.70' . 15 .64' .61' .53' .4-J' .39' 15. ScmDiv -.27' - .Ol .02 .22' -.01 .06 .40' -.11 -.2S' -.23' -.25' -.2-1' -.26' -.27' 16. Concreteness -.38' -.20' -.55' -.50' -.53' -.46"' . 12 -AO' -.35' -.39' -.42' -.33' -.23' -. ll 17. lmagcability -.48' -.28' -.65 ' -.5 1' -.55' -.48' .17' -.63' -.24' -.24· -.25• •. 19• -.04 -.11 -.Ol .64· -. 14 -.53· -.65· -.54· _.45• -.05 -.32· .OR -.Ol .03 .OR .06 . 14 -.11 ...is• .78' .3-l' -.11 18. 801 -.n• 19. Aroussal -.09 .0 1 .OR .Ol .06 .01 .03 .07 .Ol .01 .06 .00 -.07 -.04 . 15 •. 11• .Ol -. 11• 20. Ncgati,•ity -. 14 -.02 -. 13 -.Ol -.11• -.06 .19• -.23· -.25· -.23· -.26· -.20• -.09 -.12 -.26• _34• .28• .20• 21. Positi\'ity -.0 1 -.21• .03 -. 12 .04 .36· -.35· -.2t1• -.26· -.21• -.26* . _24• •. 19• .02 . 16 .3 1* .16 .08 _44• 22. CA -.42· __49• -.4 2· -.60· -.41* -.ll ' -.49* .20• -.58· -.19· -. l l -. 14 -. 13 -.03 -.09 -.36· .42• .62* .34* -.02' .41' -.38· -.09 -. 10 .23· .19• .31· .39• -.07 -. 18· -. 18* -.16 -.17· -.22• -.34· .21• -.21• ._04• .Jo• 23 . EE -. 13 .4 1• -.32· 4 1' .35' -.38· -.0 1 *p < .05 Note. CL= categorization latency; Frequency= HAL log-frequency; AoA = age of acquisition; OLD= orthographic Levenshtein distance; INV+ l = inverse of number of word neighbours plus I; SemDiv = semantic diversity; BOI = body-object interaction; CA= context availability; EE= emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 58 Table 4 Results of Interaction Tests in the Omnibus Latency Analyses Between Abstract and Concrete Words in LDT, SCT, and SLDT Variable B Step 1 (Control variables) Step 2 LDT Word type x EE LDT Step 1 (Control variables) Step 2 SCT Word type x EE SCT Step 1 (Control variables) Step 2 SLDT Word type x EE SLDT 0.0 0.17 0.24 SEE 0.03 0.04 0.04 fJ .03 .82 1.21 .01 .21 *** .31 *** .00 .71 *** .71 *** .05*** .32*** .37*** .10*** .29*** .39*** *p < .05, **p < .01 , ***p < .001 Note. LDT= lexical decision task; SCT = semantic categorization task; SLDT = semantic lexical decision task; EE= emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Table 5 Results of Interaction Tests in the Omnibus Error Analyses Between Abstract and Concrete Words in LDT, SCT, and SLDT Variable B Step 1 (Control variables) Step 2 LDT Word type x EE LDT Step 1 (Control variables) Step 2 SCT Word type x EE SCT Step 1 (Control variables) Step 2 SLDT Word type x EE SLDT .17 2.45 5.78 SEB .43 .71 0.87 p .07 .71 1.34 sr .02 .18** .34*** l1R 2 R2 .00 .45*** .45*** .03** .22*** .25*** .12*** .21 *** .33*** *p < .05 , **p < .01 , ***p < .001 Note . LDT= lexical decision task; SCT = semantic categorization task; SLDT = semantic lexical decision task; EE = emotional experience. 59 SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 60 Table 6 Results of Hierarchical Multiple Regression Analyses for the Experiment 1A: Abstract LDT Latencies Variable Step 1 Freq AoA OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageabi Iity BOI Arousal Negativity Positivity CA Step 2 EE B SEB /J sr -.07 .04 -.12 .09 .01 -.01 -.04 .04 .25 -.10 -.06 .11 -.06 .00 .03 -.14 .02 .02 .06 .03 .03 .05 .05 .04 .08 .12 .05 .07 .03 .04 .04 .04 -.26 .22 -.17 .35 .02 -.02 -.06 .06 .22 -.06 -.08 .10 -.12 .00 .06 -.24 -.16** .14** -.o9t .14** .01 -.01 -.04 .05 .15** -.04 -.06 .08 -.10tt -.00 .04 -.16** -0.12 0.03 -.33 -.19*** L1R2 R2 .69*** .04*** .73 *** L1R2 R2 Errors Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI B SEB /J sr -.30 .04 -.76 .18 .18 .84 -.65 .32 3.46 -1.97 .51 -.13 .34 .24 .95 .51 .42 .74 .75 .55 1.19 1.79 .81 1.05 -.09 .02 -.09 .06 .05 .12 -.08 .04 .25 -.09 .05 -.01 -.06 .01 -.05 .02 .03 .07 -.05 .04 .18** -.07 .04 -.01 .40*** SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE .31 -.67 .85 -3.50 .51 .57 .62 .68 .05 -.10 .12 -.49 .04 -.07 .09 -.31 *** -0.11 0.46 -.03 -.01 .00 61 .40*** t = .055, tt = .053 , *p < .05, **p < .01 , ***p < .001 Note. Freq= HAL log-frequency; AoA = age of acquisition; LOD = Levenshtein orthographic distance; NCOUNT-INV = inverse of number of word neighbours plus 1; BOI = body-object interaction; CA = context availability; EE= emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 62 Table 7 Results of Hierarchical Multiple Regression Analyses for the Experiment 1B: Concrete LDT Latencies Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI Arousal Negativity Positivity CA Step 2 EE B SEB p sr -.10 .05 -.02 .02 .03 .14 -.04 -.02 -.05 -.07 -.01 -.05 -.01 .15 -.07 -.15 .02 .01 .05 .03 .02 .05 .04 .06 .07 .06 .05 .02 .02 .04 .03 .04 -.41 .23 -.04 .08 .10 .20 -.06 -.03 -.05 -.07 -.02 -.11 -.02 .22 -.12 -.25 -.23*** .15** -.01 .03 .05 .12** -.05 -.02 -.03 -.05 -.01 -.09* -.01 .15*** -.09* -.15** -.06 0.03 -.13 -.o8t LJR2 R2 .75*** .01 t .78*** LJR2 R2 Errors Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI B SEB p sr -.73 .03 -.23 -.04 .OJ -.80 -1.29 -1.03 2.37 -1.09 .39 -.0 I .25 .21 .69 .37 .34 .71 .52 .81 .99 .93 .70 .35 -.34 .02 -.05 -.02 .01 -.13 -.21 -.15 .25 -.11 .06 .00 -.19** .01 -.02 -.01 .00 -.07 -.16* -.08 .16* -.08 .04 .00 .43*** SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE .25 .97 .71 -2.67 .33 .58 .43 .63 .06 .16 .14 -.49 .05 .11 .11 -.28*** 0.14 0.43 .03 .02 .00 63 .43*** t = .051 , *p < .05, **p < .01 , ***p < .001 Note. Freq= HAL log-frequency; AoA = age of acquisition; LOD = Levenshtein orthographic distance; NCOUNT-INV = inverse of number of word neighbours plus I ; BOI = body-object interaction; CA = context availability; EE = emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 64 Table 8 Results of Hierarchical Multiple Regression Analyses for the Experiment 2A : Abstract SCT Latencies Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI Arousal Negativity Positivity CA Step 2 EE E SEE p sr -.01 -.01 -.10 .02 .02 .05 .04 .23 .20 -.10 -.04 -.05 -.04 .11 -.09 -.04 .03 .02 .08 .04 .03 .06 .06 .04 .10 .14 .06 .11 .04 .04 .05 .06 -.06 -.07 -.16 .11 .10 .10 .06 .44 .21 -.07 -.05 -.04 -.09 .24 -.18 -.08 -.04 -.04 -.09 .04 .05 .06 .04 .38*** .15* -.05 -.04 -.03 -.07 .18* -.13 -.05 -0.13 0.03 -.43 -.25*** L1R2 R2 .40*** .06*** .46*** L1R2 R2 Errors Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI E SEE p sr .77 .16 -.21 .11 .17 -.90 .36 1.80 4.81 -4.12 -1.78 6.39 .47 .33 1.30 .69 .57 1.01 1.04 .76 1.65 2.44 1.13 1.86 .21 .06 -.02 .03 .04 -.12 .04 .21 .30 -.17 -.17 .33 .12 .04 -.01 .01 .02 -.07 .03 .18* .22** -.13 -.12 .26** .30*** SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE -.77 2.15 -1.29 1.64 .70 .76 .83 0.96 -.10 .28 -.16 .20 -.08 .21 ** -.17 .13 -1.48 0.62 -.30 -.17* .03* 65 .33*** *p < .05, **p < .01 , ***p < .001 Note . Freq= HAL log-frequency; AoA = age of acquisition; LOD = Levenshtein orthographic distance; NCOUNT-INV = inverse of number of word neighbours plus I; BOI = body-object interaction; CA = context availability; EE = emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 66 Table 9 Results of Hierarchical Multiple Regression Analyses for the Experiment 2B: Concrete SCT Latencies Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI Arousal Negativity Positivity CA Step 2 EE B SEB /J sr -.02 .07 -.06 .00 .03 -.01 .00 -.12 -.10 .18 -.15 -.09 .03 .09 .03 -.11 .02 .02 .05 .03 .03 .06 .04 .06 .08 .11 .05 .03 .03 .05 .04 .05 -.10 .34 -.15 .00 .13 -.01 .01 -.16 -. I 0 .11 -.23 -.21 .08 .I2 .05 -.16 -.05 .21 *** -.06 .00 .06 -.01 .00 -. I 0 -.06 .08 -.14** -.16** .07 .09 .04 -.11 * .10 .03 .23 .16** l1R2 R2 .67*** .03** .70*** l1R2 R2 Errors Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI B SEB /J sr -.04 -.12 -1.08 .57 -.40 .28 .08 -1.61 .87 4.17 -3.19 -3.33 .31 .26 .81 .45 .40 .86 .62 .96 1.20 1.67 .84 .43 -.01 -.04 -.18 .18 -.12 .03 .01 -.15 .06 .17 -.33 -.51 -.0 I -.02 -.07 .07 -.05 .02 .01 -.09 .04 .13* -.20*** -.40*** .66*** SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE -.21 -.52 2.32 -1.04 .40 .78 .55 .81 -.03 -.05 .30 -.11 -.03 -.03 .22*** -.07 1.38 .50 .21 .14** .02** 67 .68*** *p < .05, **p < .01, ***p < .001 Note. Freq= HAL log-frequency; AoA = age of acquisition; LOD = Levenshtein orthographic distance; NCOUNT-INV = inverse of number of word neighbours plus 1; BOI = body-object interaction; CA = context availability; EE= emotional experience. SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 68 Table 10 Results of Hierarchical Multiple Regression Analyses for the Experiment 3A : Abstract SLDT Latencies Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI Arousal Negativity Positivity CA Step 2 EE B SEB /J sr .02 .01 .04 -.01 .01 .07 -.02 .14 .31 .05 -.04 .03 -.07 .04 -.11 -.06 .03 .02 .07 .04 .03 .06 .06 .04 .I0 .14 .06 .11 .04 .04 .05 .06 .07 .06 .07 -.04 .03 .15 -.04 .29 .33 .03 -.06 .03 -.16 .09 -.23 -.13 .04 .04 .04 -.02 .02 .09 -.03 .25** .24** .03 -.04 .02 -.13 .07 -.17* -.08 -.14 .03 -.51 -.29*** L1R2 R2 .39*** .08*** .48*** L1R2 R2 Errors Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI B SEB /J sr .04 .47 .78 -.40 -.24 .16 1.89 4.01 6.57 1.93 .64 2.07 .57 .42 1.58 .84 .72 1.29 1.28 .94 2.14 3.09 1.39 2.38 .01 .13 .07 -.09 -.05 .02 .16 .38 .32 .06 .05 .09 .0 l .08 .04 -.04 -.03 .01 .11 .32*** .23** .05 .04 .07 .33*** SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE -.90 1.04 -.77 -1.52 .86 .95 1.04 1.24 -.I 0 .11 -.08 -.14 -.08 .08 -.06 -.09 -3.73 .72 -.62 -.35*** .13*** 69 .46*** *p < .05, **p < .01 , ***p < .001 Note. Freq= HAL log-frequency; AoA = age of acquisition ; LOD = Levenshtein orthographic distance; NCOUNT-INV = inverse of number of word neighbours plus 1; BOI = body-object interaction; CA = context availability; EE = emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 70 Table 11 Results of Hierarchical Multiple Regression Analyses for the Experiment 3E: Concrete SLDT Latencies Variable Step 1 Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI Arousal Negativity Positivity CA Step 2 EE E SEE fJ sr -.05 .02 -.08 .05 -.03 .08 -.05 -.13 -.18 .10 -.10 -.13 .01 .01 .08 -.13 .02 .02 .05 .03 .03 .06 .04 .06 .08 .11 .06 .03 .03 .05 .04 .05 -.24 .13 -.21 .24 -.12 .14 -.09 -.19 -.21 .06 -.16 -.33 .03 .01 .17 -.22 -.13* .08 -.08 .09 -.05 .08 -.07 -.12* -.13* .05 -.09 -.26*** .02 .01 .12* -.14* .12 .03 .30 .21 *** JR2 R2 .60*** .04*** .64*** JR2 R2 Errors Variable Step I Freq Ao A OLD Letters Phonemes Syllables Morphemes Concreteness NCOUNT-INV Semantic Diversity Imageability BOI E SEE fJ sr -.83 .03 .67 -.22 -.50 -.02 -1 .32 -4.20 -2.65 2.83 -2.53 -2 .13 .52 .44 1.38 .77 .69 1.47 1.06 1.66 2.00 2.86 1.56 .74 -.19 .01 .08 -.05 -.10 .00 -. I 0 -.27 -.14 .08 -.18 -.24 -.10 .00 .03 -.02 -.05 .00 -.08 -.16* -.09 .06 -.11 -.19** .48*** SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Arousal Negativity Positivity CA Step 2 EE -.38 .06 2.54 -3.02 .69 1.34 .91 1.37 -.04 .00 .24 -.22 -.04 .00 .18** -.14* 4.04 .80 .44 .30*** .09*** 71 .57*** *p < .05 , **p < .01 , ***p < .001 Note. Freq= HAL log-frequency; AoA = age of acquisition; LOD = Levenshtein orthographic distance ; NCOUNT-INV = inverse of number of word neighbours plus 1; BOI = body-object interaction; CA= context availability; EE= emotional experience. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 72 Appendix A Abstract Words Used in the Experiments hourc degreec opinion accord accountbc hypothesis opportunity delay ignorance advance democracy outcome illusion pardon adversity desire be image patience devotion advice afterlife discipline immunity quality discretion impulse rating allegory inclinec amount disposition reaction inducement appeal distress reform approach duty instance regard aptitude effect instant relief arrayc effort intellect request emancipationc interest aspect reserve jeopardyh atrocity envy revenge attempt equity judgement review attitude export justice sensation attribute expression knowledge situation sobrietyc basis extent limit makerbc facilitybc belief soul malicec betrayal factor spirit menace ab blandness failure status capacity fallacy manner support comment fate meantime suppression comparison feature memory tendency compound be feeling mercy theory trifle be concept figment merit concern method trouble folli conflict forethought miracle truth contrast fortune moment unification control freedom mood upkeep courage future neglect value CrISIS gratitude nonsense venture custom greed nothing violation danger habit notion virtue deduction ab heredity obedience weakness deceit hinderance obsession wonder honourc decline offence decrease hope offer Note. a = removed from Experiment I A, b = removed from Experiment 2A, c = removed from experiment 3A. SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING Appendix B Concrete Words Used in the Experiments accordian circle forehead motor acid city forest mountain alligator closet freckles mouthpiece clothing aluminium garden navy arrow collar grasshopper number artist college headboard oven colonelb auditorium helmet painter author column highway painting 3 hotel baby comrade partner bacteria concert hunter penicillin barrel contract hurricane person basement corner husband picture illnessab basin cotton pitcher bedroom country island poem binoculars couple jacket police blacksmith cousm jellyfish projector blanket creature jewel propeller body cupboards jitterbug quarter building daughter journal railway bullet daylight kitchen rattlesnake businessab destroyerab ladyh rectangle laughter3b butter diamond refrigerator leaderb canal dinner sandwich disease ab candy leather scholar cannon dishwasher letter screwdriver canoe dollar lion servant luncheonb captain doorway sheriff carpet dragon marshmallow squirrel memberb castle eagle station caterpillar elbow merchant steamer cattle empire metal stomach cauliflower engine mirror student mistressb chestnut envelope summer eveningab chickenpox moisture 3 trapezoid chimney fabric monarch tuberculosis chinchilla farmer money winter chopstick finger monster cigarette flashbulb mother Note. a = removed from Experiment 2B, b = removed from experiment 3B. 73 SITUATED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING an ant an cits and ow arric babon bagin bangon bannin banny banpet barpit barrit basple bastle baundent bernness bloninet breeture bull er buttet cacel camm cancilal cany capou capple captics carrouds cery chellput chimter churiff coamin coctee col Ion combiss comder comeer Appendix C Nonwords Used in the Experiments hechway congart herter constave cop par hertway ho let cortlel hubtand costoft huslant cosuor is back cotpade isbant coustly jeersal cowl er kenchen cudy laithler daglives datcher leantain dethraxer loontain dirtner looster diseine machent donchter malap earchwoil maltress mas tore em pond middor engant envenyas m1mmer estine moby eubre morty fanser mubber filric murry murshant flanter flashcurp murvant nurger flooker footdell orshack pagor forebood forerood panner pellar fouther perpame frithing furcory piffure furmicame pluckles futter pluset gaby potike ganter praitee headboosh purbon purpon reimond roolway roonway rouler run arch runey runster scuckles sether shamney shurry slayon soating sotal spager spandow spanjet sparet spimant splilar spurry spyntal staroan stooser stopaff stub ant tagy tanchter tearer ushness veltenia vicket wan tow warpen wint 74 SITUA TED CONCEPTUALIZATION AND CONCEPTUAL PROCESSING 75 Footnotes 1 Following standard notation of The Handbook of Categorization in Cognitive Science (2005), words with square brackets in capital letters indicate a conceptual category (e.g., [TABLE]), words with capital letters indicate a conceptual feature (e.g., LEGS), and italicized words indicate the linguistic form (e.g., table). 2 Neither myself, my supervisor, or his colleagues could determine the procedure for conducting a within-subjects interaction analysis for multiple regression. Thus, these additional hypotheses will be analyzed using more qualitative methods, as no statistical analysis was conducted to examine the secondary hypotheses. Thus, caution was used in the interpretion of these hypotheses. 3 In Experiment 2B, one participant required replacement because their error rate was greater than 30%. In Experiment 3A, two participants required replacement because their error rate was greater than 30%. In Experiment 3B, two participants required replacement because their error rate was greater than 30%.