Abstract
Free full text
Tight and loose are not created equal: An asymmetry underlying the representation of fit in English and Korean speakers
Abstract
Research concerning the spatial dimension fit (tight versus loose) has been based on a tacit but untested assumption that the dimension fit is symmetrical, with tight- and loose-fitting relations highlighting the dimension fit with equal force. We propose a reformulation, documenting that adult speakers of English (Experiment 1) and Korean (Experiment 2) are sensitive to the dimension fit, but that their representation is asymmetric, with tight-fitting events highlighting fit with greater force than loose-fitting events. We propose that sensitivity to the dimension fit is more resilient than has previously been suggested, and that the asymmetry documented here provides a foundation upon which to pursue nuanced questions about the relationship between language and our underlying representations of space.
Tight and loose are not created equal: An asymmetry underlying the representation of fit Some of the most captivating questions in cognitive psychology consider how our most fundamental concepts (e.g., time, space) are represented and whether our representations are shaped by the language that we have acquired. In crafting answers to these questions, researchers have adopted very different perspectives. At one extreme lies the suggestion that human languages differ profoundly, that language exerts a strong influence on underlying conceptual representations, and that as a result, when cross-linguistic differences arise, they are accompanied by concomitant differences in underlying representations (Whorf, 1956). At the other extreme lies the suggestion that human languages differ little (if at all) in their representational capacities (Chomsky, 1986), that language exerts a minimal influence on nonlinguistic representations, and that as a result, speakers of different languages share strong convergences in their underlying representations (Gennari, Sloman, Malt, & Fitch, 2002; Li & Gleitman, 2002; Malt, Sloman, Gennari, Shi, & Wang, 1999; Munnich, Landau, & Dosher, 2001; Papafragou, Massey, & Gleitman, 2002). Between these lie a host of intermediate perspectives, which vary in their characterization of cross-linguistic differences and in their articulation of whether these differences influence the underlying non-linguistic representations (e.g., Boroditsky 2001; Davidoff, Davies, & Roberson, 1999; Hespos & Spelke, 2004; Imai & Gentner,1997; Levinson, 1996; Lucy, 1992; McDonough, Choi, & Mandler, 2003; Slobin, 1996).
Despite these differences in perspective, research programs in this area follow virtually the same logical approach. First, two (or more) languages are identified that differ in a potentially relevant way. Then, nonlinguistic tasks are designed to discover whether this difference in the domain of language has consequences on speakers’ conceptual representations.
The success of this approach rests upon the accuracy with which both the relevant linguistic phenomena and the underlying conceptual representations are characterized. In this paper, we focus on the representations underlying adults’ spatial concepts pertaining to fit. We begin with a review of recent linguistic and psychological evidence, pointing out that this work has rested upon a tacit but untested assumption that the dimension fit is symmetric, with tight-and loose-fitting events drawing attention to this dimension with equal force. We propose an alternative, suggesting that our underlying representation of fit is asymmetric, with tight-fitting events drawing attention to fit more powerfully than loose-fitting events. We then provide evidence for this proposal with adult speakers of English (Experiment 1) and Korean (Experiment 2).
Several researchers have described an intriguing cross-linguistic difference in how the concept fit is ‘packaged’ in the lexicon (Bowerman & Choi, 2001; Choi & Bowerman, 1991; Choi, McDonough, Bowerman, & Mandler, 1999; McDonough et al., 2003). Consider Figure 1. Speakers of English typically describe the events depicted in A and B as “in” and those depicted in C and D as “on”. For speakers of Korean, the events depicted in A and C are typically described as tight (the Korean term “kkita” is glossed in English as “to put together tightly”; “to interlock”) and those depicted in B and D as loose (the Korean term “nehta” is glossed as “to put loosely in or around”; the term “nohta” is glossed as “to put loosely on a horizontal surface”). Based on observations like these, the claim is that in describing events like these, speakers of English may, as an option, specify tightness-of-fit (by using an adverb, e.g., ‘loosely’; ‘snugly’), but that for speakers of Korean, specifying tightness-of-fit is not optional because information about fit is inherent in the very meanings of these verbs (e.g., Bowerman & Choi, 2001; but see Kawachi, 2007).
English and Korean terms for four distinct spatial relations.
Building upon these observations, researchers have asked whether this cross-linguistic difference is reflected in speakers’ non-linguistic representations of the dimension fit. The evidence to date suggests that it is. For example, pre-linguistic infants appear to distinguish among all four spatial relations depicted in Figure 1 (Casasola & Cohen, 2002; Casasola, Cohen, & Chiarello, 2003; Hespos & Spelke, 2004; McDonough et al., 2003). They thus appear to be poised to acquire the semantics of Korean, English or any other human language. However, the evidence from adults suggests that their performance on non-linguistic tasks may vary with the language they have acquired: adult speakers of Korean apparently attend to tightness-of-fit, but adult speakers of English apparently do not (Hespos & Spelke, 2004; McDonough et al., 2003).
These findings have been interpreted as evidence for ‘conceptual tuning’, and have been likened to a phenomenon in speech perception known as ‘perceptual tuning’: Infants are initially sensitive to all possible phonemic contrasts, but their ability to distinguish non-native phonemic contrasts diminishes as a function of exposure to their native language (e.g., Best, 1995; Kuhl, 1998; Werker & Tees, 1984). The possibility that there is ‘conceptual tuning’ in the domain of spatial relations is intriguing, but if we are to discover whether and how native language semantics ‘tune’ the conceptual representation of fit, it is essential that we capture this underlying representation with sufficient accuracy and precision. We therefore take a closer look at adults’ representation of fit.
Research on this topic has been based on an (untested) assumption that the representation underlying the dimension fit is symmetric, but there is reason to suspect that this may not be the case, and that tight-fitting events draw attention more forcefully to fit than do loose-fitting events. For example, when two entities fit together tightly, the spatial relation between them is quite specific: all or most of their surfaces are in snug contact. But when two objects fit together loosely, the spatial relation between them is, in fact, underspecified: their surfaces may or may not be in contact. Moving beyond observation, a review of the linguistic evidence offers further reason to suspect that there may be an asymmetry in our representations of fit. For example, for the spatial relations depicted in Figure 1, Korean speakers apparently require only a single word, kkita, to describe a tight-fitting event (whether it involves support or containment), but require (at least) two different words to describe a loose-fitting event, nehta for loose containment and nohta for loose support. This fact -- that in a language that marks fit explicitly within the lexicon, more terms are required to describe loose-fitting than tight-fitting events -- is consistent with the observation that tight-fitting events may provide more precise information about the spatial relation between participant objects than do loose-fitting events.
There are also hints that in non-linguistic tasks, tight-fitting events may guide attention toward fit with greater precision and force than do loose-fitting events. Despite broad interest in the topic of spatial representations, to the best of our knowledge, only two empirical investigations of adults’ sensitivity to fit have been reported (Hespos & Spelke, 2004; McDonough et al., 2003). Because both were based on a tacit assumption of underlying symmetry, the question that they asked was whether participants would attend to fit in a general sense, not whether they would be more attentive to fit in the context of tight- than loose-fitting relations. For example, McDonough et al. (2003) gathered English- and Korean-speaking adults’ responses to support and containment events involving tight- and loose-fit. But in reporting their results, they averaged over the tight- and loose-fitting events, providing a composite index of speakers’ sensitivity to fit. They reported that adult speakers of Korean, but not English, were sensitive to fit. On the basis of the averages that they report, it is not possible to consider the possibility that adult speakers of either language were more sensitive to fit in the context of tight-than loose-fitting events. However, Hespos and Spelke (2004) reported the results for tight- and loose-fitting events independently. Although English-speaking adults did not reliably demonstrate attention to fit in any condition, a careful examination of their results suggests that attention to fit was more pronounced when participants had been familiarized to tight- than to loose-fitting events. This is consistent with our suggestion that English-speaking adults may indeed represent the underlying dimension fit, and may do so in an asymmetric fashion.
In the current experiments we consider this possibility directly. We focus first on English-speaking adults because they provide the strongest test case: Although explicit mention of fit is optional when describing events in this language, we predict that adult speakers may nonetheless be sensitive to fit, with tight-fitting events guiding their attention to fit with greater force than loose-fitting events.
Experiment 1
To test the hypothesis that tight-fitting events guide attention to fit more powerfully than loose-fitting events, we adopted a similarity-judgment task patterned closely after Hespos and Spelke (2004). See Figure 2.
Experiment 1. Structure of the experimental design.
Method
Participants
Participants were 64 Northwestern University undergraduates (38 females), ranging from 17 to 22 years (M = 19.81). Seventy-nine percent were White, 15% Asian, and less than 1% Black. All were native English speakers.1
Procedure
Participants were seated in a quiet testing room, four feet from a stage on which the experimental events took place. An experimenter described the task and then moved behind a barrier for the remainder of the session.
Practice trials
To familiarize them with the structure of the task, participants first completed two practice trials, identical in structure to the experimental trials, but featuring single abstract shapes rather than spatial events (described below).
Experimental trials
Participants then completed four experimental trials, each composed of a familiarization and test phase (Figure 2). No feedback was provided. Trial order was determined by Latin-square.
Familiarization phase
Each familiarization phase featured one type of spatial event (Tight-In, Loose-In, Tight-On, or Loose-On). To begin each trial, the experimenter’s hands appeared from behind a backdrop. She placed one object (a container, pedestal or post) on the stage, then placed a second object in its specified relation to the first, and then removed it. For example, on the Tight-In trial, she placed a container on the stage, placed a cylinder tightly inside it, and then removed it. This sequence was repeated six times.
Test phase
Next, the experimenter presented a Loose-In and Tight-In test event. The same two test events appeared on all trials; they appeared in alternating order across trials. Participants used a 10-point scale to rate the similarity of each test event to the familiarization event that preceded it.
Predictions
If participants attended to fit during familiarization, then the test event that matched the familiarization event for fit should receive a higher similarity rating than the test event that did not. For example, if participants attended to fit during the Tight-In familiarization events, then at test they should provide a higher similarity rating for the Tight-In than the Loose-In test event. If participants did not attend to fit, their ratings for two test events should not differ.
Thus, if tight-fitting relations draw attention to fit more powerfully than loose-fitting relations, then (1) when participants view tight-fitting events during familiarization, they should provide a higher similarity rating for the Tight-In than the Loose-In test event, but (2) when they view loose-fitting events during familiarization, their ratings for the test events should not differ.
Results and Analysis
First trial
We first considered participants’ performance on their first trial only. This provides a point of comparison with Hespos and Spelke (2004) and with Experiment 2. We submitted similarity ratings from participants’ first trial to an analysis of variance, using Fit-at-familiarization (2: Tight versus Loose) and Location-at-familiarization (2: In versus On) as a between-participants factors and Test Event (2: Tight-In versus Loose-In) as a within-participants factor. See Figure 3. There was a main effect for Location-at-familiarization, F(1,62) = 29.20, p < .0001; ηp2 = .32, indicating that participants were sensitive to the distinction between support and containment: Similarity ratings for the two (In) test events were higher following familiarization events involving In (M = 6.98) than involving On (M = 5.19). A main effect for Test Event, F(1,62) = 31.18, p < .0001; ηp2 = .37, revealed that participants’ similarity ratings were higher for the Tight-In (M = 6.54) than the Loose-In test event (M = 5.66). This was qualified by a Test Event by Fit-at-familiarization interaction, F(1,61) = 43.23, p < .0001; ηp2 = .42. Tests of simple main effects revealed that when participants were familiarized to tight-fitting events, they attended to fit, providing higher similarity ratings for the Tight-In than the Loose-In test event (M = 6.85 and 5.03, respectively, F(1,61) = 79.33, p < .0001; ηp2 = .18). However, when participants were familiarized to loose-fitting events, their ratings for the two test events did not differ, F(1,61) = .21, p > .05; ηp2 = .04).
Experiment 1 (first trial data). Similarity judgments as a function of Fit- and Location-at-Familiarization; ****p < .0001.
All trials
An analysis of participants’ ratings on all four trials yielded the same main effects (Location-at-familiarization, F(1, 62) = 244.12, p < .0001; ηp2 = .80; Test Event, F(1,62) = 53.18, p < .0001; ηp2 = .46) and interaction (Test Event by Fit-at-familiarization, F(1,62) = 53.18, p < .0001; ηp2 = .55). However, this was qualified by an interaction between Test Event, Fit-at-familiarization and Location-at-familiarization, F(1,62) = 26.16, p < .001; ηp2 = .30. Tests of simple main effects revealed that when participants were familiarized to tight-fitting events, they were attentive to fit (Tight-In familiarization trial, F(1,62) = 79.09, p < .0001; ηp2 = .56; Tight-On familiarization trial, F(1,62) = 56.92, p < .0001; ηp2 = .48). However, when participants were familiarized to loose-fitting events, performance was inconsistent. When they viewed Loose-On familiarization events, participants rated the test events as comparable, F(1,62) = 2.88, p > .05; ηp2 = .04. Yet when they viewed Loose-In familiarization events, they attended to fit, their rating of Tight-In (M = 6.51) test event exceeded that of the Loose-In test event (M = 7.23), F(1,62) = 10.72, p < .01; ηp2 = .15.
Participants may have been most likely to attend to fit when Location remained constant throughout the trial; indeed, they were more attentive to fit during the Tight In than the Tight On trial, F(1,62) = 11.20, p < .01; ηp2 = .15. Thus, if participants were to attend to fit on any Loose trial, we would expect them to do so for the Loose In trial, and not the Loose On trial. Still, if our proposal is correct, and loose events do not effectively draw attention to fit, what accounts for participants’ unexpected attention to fit on the Loose-In trial? A strong possibility is that their performance on the Loose In trial was related to their exposure to tight-fitting relations on a preceding trial.
To address this issue, we selected participants who viewed the Loose-In familiarization event in their second trial (n = 16); half of these had completed the Tight-On trial just prior and half had completed the Loose-On trial. Performance on the Loose-In trial varied as a function of the preceding trial. Those who first viewed a Tight-On trial attended reliably to fit on their subsequent Loose-In trial, F(1,14) = 5.88, p < .05; ηp2 = .40, but those who first viewed a Loose-On trial did not, F(1,14) = 3,77, p > .05; ηp2 = .21.
This suggests that experience with a tight-fitting event is sufficiently powerful to draw participants’ attention to fit on a subsequent trial involving a loose-fitting event. Because participants in both conditions completed the same number of trials, it was not experience with the task in general that heightened their attention to fit. Moreover, experience with one loose-fitting event did not boost participants’ attention to fit on a subsequent loose-fitting event, but experience with one tight-fitting event did
Follow-up experiment
We sought to replicate this finding with an independent sample and to gain insights from participants’ justifications of their choices. Participants were 32 Northwestern University undergraduates (19 females), ranging from 18 to 20 years (M = 19.07). Participants were 90% White and 10% Asian. All were native English speakers.1 The procedure and materials were identical to Experiment 1, except that all participants completed only two trials. Participants were randomly assigned to view either the Tight-On or Loose-On events on their first trial. For all participants, the second trial was the Loose-In trial. After completing the second trial, participants were asked to justify their similarity ratings. Justifications were recorded verbatim and then coded by two independent coders for reference to “tightness,” “snugness,” and “fit,” etc.
The results replicate the finding that tight-fitting events draw attention to fit more powerfully than do loose-fitting events, and that attention to fit carries over from a Tight to a subsequent Loose familiarization trial. We submitted participants’ similarity ratings to an ANOVA, using Condition (Tight-first vs. Loose-first) as a between-participants factor, and Trial Position (First vs. Second trial) and Test Event (Tight-In vs. Loose-In) as within-participants factors. There was a main effect for Trial Position, F(1,30) = 45.26, p < .001; ηp2 = .60; a Condition by Test Event interaction, F(1,30) = 11.08, p < .01; ηp2 = .16; and a Trial Position by Test Event interaction, F(1,30) = 8.20, p < .01; ηp2 = .27. These effects were all mediated by a Condition by Trial Position by Test Event interaction, F(1,30) = 8.20, p < .01; ηp2 = .22.
We pursued this interaction by conducting tests of simple effects, focusing on performance on each trial. As predicted, on their first trials, participants in the Tight-first condition attended to fit, F(1,30) = 20.43, p < .0001; ηp2 = .41, but those in the Loose-first condition did not, F(1,30) = .82, p > .05; ηp2 = .03. This converges perfectly with the first-trial analysis of Experiment 1. Also as predicted, participants’ performance on their second (Loose-In) trial was influenced by their preceding experience. Participants in the Tight-first condition attended reliably to fit on their subsequent Loose-In trial, F(1,30) = 6.70, p < .05; ηp2 = .18; participants in the Loose-first condition did not, F(1,30) = 1.91, p > .05; ηp2 = .06.
Participants’ justifications provided converging support that the dimension fit is signaled more powerfully in the context of tight- than loose-fitting events. In the Tight-first condition, eight of the 16 participants mentioned fit explicitly; in the Loose-first condition, only two did so, χ 2(1, N = 32) = 5.24, p < .05. Interestingly, of the participants mentioning fit, fully 50% commented on the presence or absence of fit (e.g., “The second one was more similar because it fit.”); none mentioned “tight-fit” versus “loose-fit.”
Experiment 1 reveals for the first time that the representation underlying fit is more resilient in English-speakers than previous research has suggested. Despite the fact that specifying tightness-of-fit is optional for English speakers when they describe events like the ones presented here, participants were indeed sensitive to this dimension, and tight-fitting events highlighted attention to fit with greater force than loose-fitting events. In Experiment 2, we turn to consider the representations held by Korean-speakers.
Experiment 2
The goal of Experiment 2 was to consider adult Korean-speakers’ representation of the dimension fit. To the best of our knowledge, there is only a single study (McDonough et al., 2003) that examined attention to fit in Korean-speaking adults. As we have pointed out, this study did not assess whether their attention to fit was signaled more powerfully with tight- than loose-fitting events. Based on the asymmetry documented in speakers of English (Experiment 1), we expect that adult speakers of any language will attend to fit when presented with tight-fitting events. At issue is whether the semantic properties of Korean are sufficiently powerful to direct attention to fit in the context of loose-fitting events as well. Although cross-linguistic analyses have noted that when describing events (as in Figure 1), tightness-of-fit may be more accessible to speakers of Korean than English, what is less clear is whether and how this linguistic fact would effect, in any direct way, the underlying non-linguistic representation of fit. (Gennari, Sloman, Malt, & Fitch, 2002; Munnich et al, 2001; Landau, in press; Papafragou, Massey & Gleitman, 2002).
Method
Participants
Participants were 64 undergraduates (27 females) at Yonsei University in Seoul, South Korea. All were native Korean speakers. None spoke English fluently.
Materials and Procedure
These were identical to Experiment 1, except that (a) the experiment was administered in Korean by a native speaker, (b) participants completed only one trial instead of four, and (c) the practice trials were eliminated.
Results
Korean-speaking adults’ performance mirrored precisely those of their English-speaking counterparts. An ANOVA, using Fit-at-familiarization (Tight versus Loose) and Location-at-familiarization (In versus On) as a between-participants factors and Test Event (Tight-In versus Loose-In) as a within-participants factor revealed the same main effects and interactions as in Experiment 1. See Figure 5. A main effect for Location-at-familiarization, F(1,62) = 9.25, p < .01; ηp2 = .13, revealed that ratings for the two (In) test events were higher following familiarization events involving In (M = 7.13) than On (M = 5.48). This documents that Korean speakers are sensitive to the distinction between support (On) and containment (In) events, despite the fact in describing tight-fitting events (with kkita), Korean speakers do not distinguish support from containment. A main effect for Test Event, F(1,62) = 9.59, p < .01; ηp2 = .14, revealed that similarity ratings were higher for the Tight In (M = 6.56) than the Loose In test event (M = 6.05). This was qualified by a interaction between Test Event and Fit-at-familiarization, F(1,61) = 24.74, p < .0001; ηp2 = .29. A test of simple main effects revealed that, as predicted, when participants were familiarized to tight-fitting events, they attended to fit: they provided higher average similarity ratings for the Tight In than the Loose In test event (M = 6.72 and 5.38, respectively, F(1,61) = 32.57, p < .0001; ηp2 = .35). However, when participants were familiarized to loose-fitting events, they did not: there was no difference in their average ratings for the Tight In and Loose In test events (M = 6.41 and 6.72, respectively, F(1,61) = 1.72, p > .05; ηp2 = .03).
Experiment 2. Similarity judgments as a function of Fit- and Location-at-Familiarization; *p < .05; **p < .01.
Finally, we compared the performance of Korean- and English-speakers, in an ANOVA, using Language (English: Experiment 1 versus Korean: Experiment 2), Fit-at-familiarization (Tight versus Loose), and Location-at-familiarization (In versus On) and as a between-participants factors and Test Event (Tight In versus Loose In) as a within-participants factor. There were no main effects or interactions involving Language (main effect for Language, F(1,120) = .38; Language by Fit-at familiarization interaction, F(1,120) = .10; Language by Location-at-familiarization, F(1,120) = .10; Language by Test Event, F(1,120) = 2.20; Language by Fit-at-Familiarization by Location-at-Familiarization, F(1,120) = 1.22; Language by Fit-at familiarization by Test Event, F(1,120) = .24; Language by Location-at-familiarization by Test Event, F(1,120) = .01; Language by Fit-at-familiarization by Location-at-familiarization by Test Event, F(1,120) = .01).
General Discussion
These experiments reveal for the first time that the conceptual representation underlying fit is asymmetric, with tight-fitting events drawing attention to fit with greater force than loose-fitting events. The evidence from English-speaking adults is important because it reveals the representation underlying fit is more resilient, and more nuanced, than previous research (Hespos & Spelke, 2004; McDonough et al., 2003) has suggested. Despite the fact that they need not mention fit explicitly in their description of events, English-speakers are nonetheless sensitive to this dimension, and particularly in the context of tight-fitting events. The evidence from Korean-speaking adults is important because it reveals that despite the fact that speakers of this language explicitly specify both tight- and loose-fit in their descriptions of events (e.g., Bowerman & Choi, 1991), their non-linguistic representation underlying fit is nonetheless asymmetric. Moreover, despite the fact that they need not specify whether a tight-fit involves support or containment, Korean-speakers are nonetheless sensitive to this distinction. Thus, although the English and Korean languages differ in the ways in which they encode spatial and motion events, speakers of these languages do not appear to differ in their non-linguistic representations of the dimension fit. This is consistent with the view that our sensitivity to fit is not a consequence of the language we have acquired (see Gennari, et al., 2002; Munnich et al, 2001; Papafragou, et al., 2002). In both English (Experiment 1) and Korean (Experiment 2), this underlying representation is asymmetric
Accounting for differences between the current results and previous investigations
How can we account for the finding that English-speakers were sensitive to fit in the current, but not previous, experiments? And why might Korean-speakers, like English-speakers, show a tight-loose asymmetry? We suspect that these differences reflect the question that the experiments were designed to address. In previous work, the question was whether participants would attend to fit in a general sense. Our question differed subtly, but importantly. We asked whether the representation underlying fit was symmetric, and designed our analyses to ask whether participants might be more attentive to fit in the context of tight- than loose-fitting events. Using Hespos and Spelke (2004) as a starting point, we introduced several modifications. For example, although we preserved precisely the spatial dimensions of their materials, we modified their color in an effort to minimize participants’ attention to anything other than fit. Importantly, these design modifications—featured on all trials and in both languages —did not lead participants to attend to fit on all trials. Instead, participants attended reliably to fit only in the context of all tight-fitting events.
Characterizing the asymmetry
To the best of our knowledge, this is the first time that a representational asymmetry underlying fit has been proposed and the first time that this issue has been brought into contact with research on language and thought. However, the tight-loose asymmetry documented here is compatible, at least in spirit, with several other well-established asymmetries, including those involving focal vs. non-focal colors, canonical vs. non-canonical line orientation, known vs. unknown landmarks, and familiar vs. unfamiliar entities (e.g., Bowdle & Gentner, 1997; Holyoak and Gordon, 1983; Huttenlocher, Hedges, & Duncan, 1991; Karylowski, 1990; Medin & Waxman, 2007; Nosofsky, 1991; Rips, 1975; Rosch, 1975; Sadalla, Burroughs, & Staplin, 1980; Treisman & Souther, 1985; Tversky, 1977). Taken together, the evidence suggests that asymmetries arise in tasks that involve comparisons (implicit or explicit) between two elements, one of which enjoys greater psychological privilege or salience than the other.
In future work, it will be important to characterize the tight-loose asymmetry more precisely and to discover whether and how it is related to linguistic, perceptual and conceptual factors. We suspect that it is not created by language. For example, we would be hard-pressed to identify any features of English that could have created it. This asymmetry is more likely to be related to perceptual and conceptual factors. In particular, when two entities fit together tightly, the physical relation between them is quite specific: their surfaces are in close contact. But when two objects are in a loose spatial relation, the physical relation between them is vastly underspecified: their surfaces may or may not be in contact (compare, e.g., an apple placed in a briefcase vs. a shopping bag). Moreover, when two entities fit together tightly, they were often designed to do so. This is less reliably the case for entities that fit together loosely. This raises the intriguing possibility that tight-fitting relations may gain psychological prominence by virtue of the correlation between tightness-of-fit and intended function (Bloom & Markson, 1998; Booth, 2006; Casler & Kelemen, 2005; Diesendruck, Markson, & Bloom, 2003; Keil, 1989; Kemler Nelson, Herron & Holt, 2003; Matan & Carey, 2001).
These observations, coupled with the results reported here, suggest that the continuum of spatial relations underlying fit may be better described as fits versus does not fit than as tight versus loose. Munnich, Landau and Dosher (2001) propose an analogous view of contact versus no contact. Indeed, events involving tight-fit and direct contact may serve as strong reference points, providing clear standards against which other points along their respective continua may be measured (see Needham & Baillargeon, 1993; Leslie, 1984).
Can the evidence of a tight-loose asymmetry in adults’ underlying representation of fit be reconciled with the notion of ‘conceptual tuning’? In our view, the answer is ‘yes’: Interpreting the asymmetry underlying fit within the framework of conceptual tuning leads to precise developmental and cross-linguistic predictions. For example, it predicts that tight-fitting events may highlight the dimension fit robustly, with or without the explicit support of language, but that loose-fitting events (which may be psychologically or perceptually less privileged) may depend more upon the semantics of the ambient language (Casasola, 2005). Although we have shown for adult speakers of Korean, the spatial semantics of their language is not sufficiently powerful to outweigh an underlying tight-loose asymmetry, it remains to be seen whether a language-specific effect may be evident in other non-linguistic tasks.
Avenues for additional research
At a more general level, a comprehensive resolution to questions concerning the relation between linguistic and non-linguistic representations of fit awaits additional evidence from three inter-related areas. First, it will be important to discover whether infants’ and toddlers’ representation of fit is symmetric or asymmetric. Hespos and Spelke’s (2004; Figure 1b) data from 5-month-old infants suggests that the pre-linguistic representation may, in fact, be asymmetric. When infants were habituated to tight-fitting events, they attended reliably and consistently to fit. But when they were habituated to loose-fitting events, performance was inconsistent: they attended to fit after viewing Loose-In events, but not after Loose-On events. This echoes the pattern we observed in Experiment 1.
However, Choi’s (2006) developmental data presents a different picture. She reported that English- and Korean-acquiring toddlers, ranging from 18 to 24 months of age, were equally sensitive to fit in the context of both tight- and loose-fitting events. However, Choi suggested that several months later, toddlers’ sensitivity to fit in the two language communities begins to diverge. At 29 to 36 months of age, Korean-acquiring toddlers maintained their sensitivity to fit in the context of both tight- and loose-fitting events, but within this same developmental period, English-acquiring toddlers’ attention to fit weakened. A closer examination revealed that English-acquiring toddlers were more likely to attend to fit if they had been familiarized to tight-than to loose-fitting events. This finding, although unanticipated by Choi, is consistent with the position that we have advanced here.
Clearly, additional cross-linguistic, developmental research on infants’ and toddlers’ representation of fit will be necessary to resolve whether infants’ pre-linguistic representation is symmetric or asymmetric and how this representation fares as infants acquire the semantics of their native language. Such evidence, important in its own right, is essential if we are to understand whether and how language shapes the pre-linguistic representation.
Second, additional research from adults is warranted. Although both English- and Korean-speakers revealed a tight-loose asymmetry in the similarity judgment task presented here, it remains to be seen whether this asymmetry holds up in both languages across a broader range of tasks. One possibility is that for speakers of English, the asymmetry will be present in a broad range of tasks, but that for Koreans, whose language lexicalizes the distinction between tight- and loose-fit, the asymmetry will be less pronounced in some tasks than in others. It will also be important to examine adult speakers of languages other than English and Korean to ascertain whether across languages, attention to fit is signaled more powerfully with tight- than loose-fitting events.
Third, it is essential that we develop a more comprehensive analysis of the ways in which spatial relations are lexicalized across languages. Bowerman and Choi’s (1991) insight that languages ‘package’ spatial relations differently was instrumental, but several issues remain unresolved. For example, the relevant spatial terms (kkita; nehta; nohta) are lexicalized as verbs in Korean, but as prepositions (in; on), adjectives (tight; loose), or verbs (fits; does not fit) in English. It remains unclear whether speakers’ construals of the relevant terms are affected differentially as a function of their grammatical form. It is also not entirely clear how to most accurately characterize the semantics of the Korean spatial terms. The assertion that the Korean lexical items include specific reference to fit (as suggested by Bowerman & Choi, 1991) has recently been called into question (Kawachi, 2007).
Finally, while considerable attention has been devoted to the question of whether and how the language that we speak structures our representations of space, the reciprocal question is compelling as well. We are now in a position to ask whether and how the asymmetric representation of fit influences language. This underlying asymmetry may be reflected in speakers’ descriptions of fit across languages (Norbury & Waxman, in prep). The asymmetry that we have reported here may also give rise to developmental differences in lexical acquisition: Infants may learn words referencing tight-fit earlier than those referencing loose-fit.
In closing, as we continue to investigate the relation between the language that we speak and our non-linguistic representations of fit, it is essential that we capture both the linguistic and conceptual phenomena with precision.
Experiment 1 (all trials). Similarity judgments as a function of Fit- and Location-at-Familiarization; **p < .01; ****p < .0001.
Acknowledgments
This research was supported by an IES training grant to Norbury and NIH R01 HD30410 to Waxman. Portions of this research were presented at the meetings of the Cognitive Development Society (2005) and the Cognitive Science Society (2006). We are indebted to D. Medin and E. Leddon for extensive discussions, and to S. Hespos for insights into all phases.
Footnotes
1Seven participants in Experiment 1 and two in the follow-up had acquired another language concurrently with English, but had spoken predominantly English for more than 15 years. None spoke Korean or were from a Korean cultural background. We report on analyses that include data from all participants. However, when only monolingual participants were included, the same main effects and interactions emerged.
2In designing the objects, we sought to minimize the possibility that participants would systematically use the width or color of any single object, rather than the spatial relations between two objects, as the basis for their similarity judgments. To minimize their use of width, each familiarization event featured a 3-inch wide base object (container, pedestal, or post); for the tight-in test event, the corresponding base object was either 1.5 inches wide (a reduction of fifty percent) ; for the loose in test event, the corresponding base object was 4.5 inches wide (an increase of fifty percent). To minimize their use of color, the color of the base object (container, pedestal or post) varied randomly for every event within each trial (six familiarization events; two test events).
The relative widths of the stimulus objects present a potential confounding factor: When the familiarization objects fit tightly, the ratio of their widths (approximately 1:1) was identical to that of the tight-fitting test pair. However, when the familiarization objects fit loosely, the ratio of their widths (approximately 1:2) was different from that of the loose-fitting test pair (approximately 1:3). These ratios might bias participants toward the asymmetric responses we describe. However, in an unrelated set of experiments, Hespos (personal communication, 2007) had participants complete a task identical to ours with one exception: The familiarization objects were not placed into tight or loose spatial relations; rather one object occluded the other. If size and not fit were driving the asymmetry, Hespos would have found asymmetric responses between trials in which the stimulus objects were similarly sized and those in which they were not. In fact, she did not find an asymmetry. This suggests that our results stem from participants’ attending to the relations between objects and not to their relative size.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Heather M. Norbury, Department of Psychology, Northwestern University, 2029 Sheridan Road, Evanston, IL 60208 USA, ude.nretsewhtron@yrubron, Tel: +1 847 467 8932, Fax: +1 847 491 7859.
Sandra R. Waxman, Department of Psychology, Northwestern University, 2029 Sheridan Road, Evanston, IL 60208 USA, ude.nretsewhtron@namxaw-s, Tel: +1 847 467 2293, Fax: +1 847 491 7859.
Hyun-Joo Song, Department of Psychology, Yonsei University, 134 Shinchon-dong, Seodaemon-gu Seoul, South Korea 120-749, rk.ca.iesnoy@gnosh, Tel: +82 2123 2449, Fax: +82 365 4354.
References
- Best CT. Learning to perceive the sound patterns of English. In: Rovee-Collier C, Lipsitt LP, editors. Advances in Infancy Research. Norwood, NJ: Ablex; 1995. pp. 217–304. [Google Scholar]
- Bloom P, Markson L. Intention and analogy in children’s naming of pictorial representations. Psychological Science. 1998;9(3):200–204. [Google Scholar]
- Booth AE. Object Function and Categorization in Infancy: Two Mechanisms of Facilitation. Infancy. 2006;10(2):145–169. [Google Scholar]
- Boroditsky L. Does language shape thought? Mandarin and English speakers’ conceptions of time. Cognitive Psychology. 2001;43(1):1–22. [Abstract] [Google Scholar]
- Bowdle BF, Gentner D. Informativity and asymmetry in comparisons. Cognitive Psychology. 1997;34(3):244–286. [Abstract] [Google Scholar]
- Bowerman M, Choi S. Shaping meanings for language: universal and language specific in the acquisition of spatial semantic categories. In: Bowerman M, Levinson SC, editors. Language Acquisition and Cognitive Development. New York: Cambridge University Press; 2001. pp. 475–511. [Google Scholar]
- Casasola M. Can Language Do the Driving? The Effect of Linguistic Input on Infants’ Categorization of Support Spatial Relations. Developmental Psychology. 2005;41(1):183–192. [Europe PMC free article] [Abstract] [Google Scholar]
- Casasola M, Cohen LB. Infant categorization of containment, support and tight-fit spatial relationships. Developmental Science. 2002;5(2):247–264. [Google Scholar]
- Casasola M, Cohen LB, Chiarello E. Six-month-old infants’ categorization of containment spatial relations. Child Development. 2003;74(3):679–693. [Europe PMC free article] [Abstract] [Google Scholar]
- Casler K, Kelemen D. Young children’s rapid learning about artifacts. Developmental Science. 2005;8(6):472–480. [Abstract] [Google Scholar]
- Choi S. Influence of language-specific input on spatial cognition: Categories of containment. First Language. 2006;26(2):207–232. [Google Scholar]
- Choi S, Bowerman M. Learning to express motion events in English and Korean: The influence of language-specific lexicalization patterns. Cognition. 1991;41(1–3):83–121. [Abstract] [Google Scholar]
- Choi S, McDonough L, Bowerman M, Mandler JM. Early sensitivity to language-specific spatial categories in English and Korean. Cognitive Development. 1999;14(2):241–268. [Google Scholar]
- Chomsky N. Knowledge of language: Its nature, origin and use. New York, NY: Praeger; 1986. [Google Scholar]
- Davidoff J, Davies I, Roberson D. Colour categories in a stone-age tribe. Nature. 1999;398:203–204. [Abstract] [Google Scholar]
- Diesendruck G, Markson L, Bloom P. Children’s reliance on creator’s intent in extending names for artifacts. Psychological Science. 2003;14(2):164–168. [Abstract] [Google Scholar]
- Gennari SP, Sloman SA, Malt BC, Fitch WT. Motion events in language and cognition. Cognition. 2002;83(1):49–79. [Abstract] [Google Scholar]
- Hespos SJ, Spelke ES. Conceptual precursors to language. Nature. 2004;430(6998):453–456. [Europe PMC free article] [Abstract] [Google Scholar]
- Holyoak KJ, Gordon PC. Social reference points. Journal of Personality and Social Psychology. 1983;44(5):881–887. [Google Scholar]
- Huttenlocher J, Hedges LV, Duncan S. Categories and particulars: Prototype effects in estimating spatial location. Psychological Review. 1991;98(3):352–376. [Abstract] [Google Scholar]
- Imai M, Gentner D. A cross-linguistic study of early word meaning: Universal ontology and linguistic influence. 1997. Cognition. 1997;62(2):169–200. [Abstract] [Google Scholar]
- Karylowski JJ. Social reference points and accessibility of trait-related information in self-other similarity judgments. Journal of Personality and Social Psychology. 1990;58(6):975–983. [Google Scholar]
- Kawachi K. Korean putting verbs do not contrast space contrastively in terms of “tightness of fit. Lingua. 2007;117:1801–1820. [Google Scholar]
- Keil FC. Concepts, kinds, and cognitive development. Cambridge, MA: MIT Press; 1989. [Google Scholar]
- Kemler Nelson DG, Herron L, Holt MB. The sources of young children’s name innovations for novel artifacts. Journal of Child Language. 2003;30(4):823–843. [Abstract] [Google Scholar]
- Kuhl PK. Effects of language experience on speech perception. Journal of the Acoustical Society of America. 1998;103(5):2931. [Google Scholar]
- Landau B. Space in semantics and cognition: Three themes. In: Maienborn C, von Heusinger K, Portner P, editors. Handbook of semantics. Berlin: Mouton de Gruyter; in press. [Google Scholar]
- Leslie AM. Infant perception of a manual pick-up event. British Journal of Developmental Psychology. 1984;2(1):19–32. [Google Scholar]
- Levinson SC. Relativity in spatial conception and description. In: Gumperz JJ, Levinson SC, editors. Rethinking linguistic relativity. New York, NY: Cambridge University Press; 1996. pp. 177–202. [Google Scholar]
- Li P, Gleitman L. Turning the tables: Language and spatial reasoning. Cognition. 2002;83(3):265–294. [Abstract] [Google Scholar]
- Lucy JA. Grammatical categories and cognition: A case study of the linguistic relativity hypothesis. Cambridge: Cambridge University Press; 1992. [Google Scholar]
- Malt BC, Sloman SA, Gennari S, Shi M, Wang Y. Knowing versus naming: Similarity and the linguistic categorization of artifacts. Journal of Memory and Language. 1999;40(2):230–262. [Google Scholar]
- Matan A, Carey S. Developmental changes within the core of artifact concepts. Cognition. 2001;78(1):1–26. [Abstract] [Google Scholar]
- McDonough L, Choi S, Mandler JM. Understanding spatial relations: Flexible infants, lexical adults. Cognitive Psychology. 2003;46(3):229–259. [Abstract] [Google Scholar]
- Medin DL, Waxman SR. Interpreting asymmetries of projection in children’s inductive reasoning. In: Feeney A, Heit E, editors. Inductive Reasoning. New York, NY: Cambridge University Press; 2007. [Google Scholar]
- Munnich E, Landau B, Dosher BA. Spatial language and spatial representation: a cross-linguistic comparison. Cognition. 2001;81(3):171–207. [Abstract] [Google Scholar]
- Needham A, Baillargeon R. Intuitions about support in 4.5-month-old infants. 1993. Cognition. 1993;47(2):121–148. [Abstract] [Google Scholar]
- Norbury H, Waxman SR. English- and Korean-speakers’ verbal descriptions of fit in preparation. [Google Scholar]
- Nosofsky RM. Stimulus bias, asymmetric similarity, and classification. Cognitive Psychology. 1991;23(1):94–140. [Google Scholar]
- Papafragou A, Massey C, Gleitman L. Shake, rattle, ‘n’ roll: the representation of motion in language and cognition. Cognition. 2002;84(2):189–219. [Abstract] [Google Scholar]
- Rips LJ. Inductive judgments about natural categories. Journal of Verbal Learning & Verbal Behavior. 1975;14(6):665–681. [Google Scholar]
- Rosch E. Cognitive reference points. Cognitive Psychology. 1975;7(4):532–547. [Google Scholar]
- Sadalla EK, Burroughs WJ, Staplin LJ. Reference points in spatial cognition. Journal of Experimental Psychology: Human Learning & Memory. 1980;6(5):516–528. [Abstract] [Google Scholar]
- Slobin DI. From “thought and language” to “thinking for speaking” In: Gumperz J, Levinson SC, editors. Rethinking Linguistic Relativity. New York: Cambridge University Press; 1996. pp. 70–96. [Google Scholar]
- Treisman A, Souther J. Search asymmetry: A diagnostic for preattentive processing of separable features. Journal of Experimental Psychology: General. 1985;114(3):285–310. [Abstract] [Google Scholar]
- Tversky A. Features of similarity. Psychological Review. 1977;84(4):327–352. [Google Scholar]
- Werker JF, Tees RC. Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior & Development. 1984;7(1):49–63. [Google Scholar]
- Whorf BL. In: Language, thought, and reality: selected writings of Benjamin Lee Whorf. Carroll JB, editor. Oxford, England: Technology Press of MIT; 1956. [Google Scholar]
Full text links
Read article at publisher's site: https://doi.org/10.1016/j.cognition.2008.07.019
Read article for free, from open access legal sources, via Unpaywall:
https://europepmc.org/articles/pmc2617708?pdf=render
Citations & impact
Impact metrics
Citations of article over time
Smart citations by scite.ai
Explore citation contexts and check if this article has been
supported or disputed.
https://scite.ai/reports/10.1016/j.cognition.2008.07.019
Article citations
Spatial Semantics, Cognition, and Their Interaction: A Comparative Study of Spatial Categorization in English and Korean.
Cogn Sci, 22 May 2018
Cited by: 1 article | PMID: 29790181
What Develops in Infants' Spatial Categorization? Korean Infants' Categorization of Containment and Tight-Fit Relations.
Child Dev, 89(4):e382-e396, 03 Aug 2017
Cited by: 2 articles | PMID: 28771703
The importance of lexical verbs in the acquisition of spatial prepositions: The case of in and on.
Cognition, 157:174-189, 16 Sep 2016
Cited by: 3 articles | PMID: 27643981
Containment and Support: Core and Complexity in Spatial Language Learning.
Cogn Sci, 41 Suppl 4:748-779, 20 Jun 2016
Cited by: 1 article | PMID: 27323000
Relative contribution of perception/cognition and language on spatial categorization.
Cogn Sci, 36(1):102-129, 04 Oct 2011
Cited by: 9 articles | PMID: 21972797 | PMCID: PMC3257354
Go to all (6) article citations
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
Relative contribution of perception/cognition and language on spatial categorization.
Cogn Sci, 36(1):102-129, 04 Oct 2011
Cited by: 9 articles | PMID: 21972797 | PMCID: PMC3257354
Spatial Semantics, Cognition, and Their Interaction: A Comparative Study of Spatial Categorization in English and Korean.
Cogn Sci, 22 May 2018
Cited by: 1 article | PMID: 29790181
Whereof one cannot speak: How language and capture of visual attention interact.
Cognition, 194:104023, 21 Aug 2019
Cited by: 4 articles | PMID: 31445296
The semantics of space: integrating linguistic typology and cognitive neuroscience.
Neuropsychologia, 44(9):1607-1621, 03 Mar 2006
Cited by: 49 articles | PMID: 16516934
Review
Using experiential optimization to build lexical representations.
Psychon Bull Rev, 26(1):103-126, 01 Feb 2019
Cited by: 4 articles | PMID: 29968206
Review
Funding
Funders who supported this work.
NICHD NIH HHS (5)
Grant ID: R01 HD030410-15
Grant ID: R01 HD030410-14
Grant ID: R01 HD30410
Grant ID: R01 HD030410
Grant ID: R01 HD030410-13