Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 Effect of Word Type Does Word Type Direct Children’s Attention to Different Properties of Objects and Actions? M. Elaine Heard Trinity University [email protected] Upon hearing a word for the first time, children must decide what aspects of the scene are important to the meaning of a word, and use that knowledge to extend the word to new examples. Two experiments examined how varying the use of nouns and verbs may guide children’s attention to the perceptual and functional characteristics of objects. In both studies, novel objects and actions were introduced at the same time in conjunction with one label whose category of speech varied by condition (Noun, Verb, No Word) and that could be interpreted as referring to either the action or the object being presented. Children’s choice between a new item that was perceptually or functionally similar to the target in a word extension task was used to measure their understanding that different types of words have different referents. Results from Experiment 1 (N = 60) suggested that while the task naturally drew children’s attention to function, children in the Noun condition extended the novel word to similarly shaped objects. However, children in the Verb condition did not extend based on function. In Experiment 2, (N = 30) the targets and the objects functionally similar to the targets were altered to increase their functional salience, such that they produced an interesting, concrete change. This change altered children’s performance such that no significant difference in responding was found between the conditions. Yesterday at snack time I told my two-year-old nephew that we were eating an apple. I didn’t specify that we were eating a Fuji, or that apples come in different colors and sizes. I supposed that if he could simply learn “apple”, we were doing fine for one day. Today in the grocery store, he reached out and pulled a Granny Smith from the bottom of the stack. As several dozen apples came toppling to the floor, the rascal smiled and murmured the word for yesterday’s snack, “apple.” Though pleased that he had retained the lesson, I couldn’t help thinking a lesson in rudimentary physics would have been more worth my time. The story highlights a couple of interesting points about word-learning. First, children are amazingly adept word-learners. Given limited exposure, children are able to understand and remember words; so well, that after slowly learning a few “first words” before 18 months go on to learn their next 30,000 words at a rate of 10-20 new words per day (Jones & Smith, 1991). More relevant to the current study, this anecdote demonstrates the ability of children to appropriately extend a single label to many new examples. Not only was my nephew able to extend the label “apple” to another specific apple that he had never seen before, he also used the word “apple” in conjunction with a class of apple that differed in obvious ways from the I one initially labeled for him. I am confident that he also could have identified McIntoshes, Golden Delicious, and Braeburn as “apples” had we not fled from the produce section as quickly as we did. Essentially, my nephew’s ability to use the word “apple” in a new way illustrates the general ability we all share to categorize new objects we come in contact with and then appropriately extend a label we already know to these objects. Thus, categorization and naming are linked; we use words to impose uniformity on different, but related, patterns of information. One word (such as apple) can refer to several similar objects, or to an entire category of objects. Given the countless members of any one category (picture hundreds of apples rolling along the grocery floor), it is fortunate that we can extend words to new examples even from a young age. It would be an impossible chore for parents to label each and every newly encountered object for their children, and equally impossible for us to function in this complicated world without the use of labels and the categories they represent. UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 Categories are inherently hierarchical, and basic level categories organize information in the most efficient manner, highlighting similarities between objects within the same category while simultaneously maximizing differences between objects inside and outside of a category. Given the efficiency of the basiclevel grouping, and the relative ease with which objects can be labeled (compared with actions or properties), it should not surprise us to know that children’s early vocabularies, especially in English, are dominated by nouns, most of which refer to basic-level categories of objects. Although we most often use categorization as a way to organize and group objects, we must also categorize events and extend new verbs on the basis of some similarity. For example, a verb such as “ride” refers to similar, but different, actions across varying situations—in a car, on piggyback, on a horse. Although it seems fairly obvious that the words we use often refer to categories describing some type of similarity between concepts, it is more difficult to make generalizations about the types of similarities different types of words highlight. Do we use certain categories of speech to refer to objects with specific similarities? Previous studies suggest that both young children and adults extend new nouns to categories of objects that are similarly shaped (e.g., Landau, Smith & Jones, 1988). A more interesting question to us is whether children make the same assumptions about extension for all word types, or whether they realize that adjectives and verbs should be extended in different ways (by property, action/function). Several studies have investigated early noun learning, and the results from these studies, with few exceptions, have indicated that children primarily use shape when extending nouns. An early study conducted by Landau et al. (1988) found that 2-year-olds, 3-yearolds, and adults all rely on shape as the natural criterion for extending names, over other perceptual properties such as size and texture. In this study, children and adults were presented with novel objects that varied in size, shape, and texture. A target object was labeled with a novel noun, and then children were asked in a forced-choice test to extend the label to an object that matched the target. At test, children and adults accepted same-shaped objects as new examples of the label, even when these objects differed greatly on size and texture. They did not readily extend the label to objects similar to the target object in other ways. Numerous other studies have bolstered evidence for this shape bias to such an extent that its existence in noun learning has Effect of Word Type been generally accepted by most language researchers (e.g., Baldwin, 1989; Smith, Jones, & Landau, 1992.) In addition to investigating children’s choice in a naming condition, Landau et al. (1988) also wondered how children would categorize objects in the absence of a label. As studies of early categorization might predict, the children’s reliance on shape did not stem from a general perceptual bias. Rather, children showed a tendency to focus on shape exclusively in the naming condition. In a classification condition in which children were simply asked to identify objects that were similar to the target object, they attended to size and texture as much as they did to shape. The lack of a general perceptual bias not only indicates that the presence of a label directs children’s attention to specific perceptual properties (or, that children learn to attend to shape when hearing nouns by paying attention to regularities in language), but also opens the possibility that children hearing other types of words (i.e. adjectives or verbs) may begin to form biases that direct attention to other relevant information. A few researchers have explored this possibility, studying how different strategies may be used in the extension of adjectives and verbs. Mapping words to properties as opposed to objects seems to be a more difficult task for the young child. Nonetheless, by 21 months children understand that adjectives refer to varying object-properties (Waxman & Markow, 1998). Adjectives may be learned later, and tend to be more difficult, because they initially depend on children’s knowledge of basic-level categories. For example, Waxman and Markow (1998) found that infants could only extend a novel adjective to a test item when the target and test objects were members of the same basic-level object category. Why would this be so? In order for children to master the concept of adjectives and become efficient communicators, they must also learn to extend adjectives to objects outside the basic-level category introduced in the initial learning environment. In a subsequent study, Waxman and Klibanoff (2000) explored what might be special about learning in most basic-level categories, and in doing so offered an explanation for how novel adjectives are eventually extended appropriately to properties of objects from diverse basic-level categories. They suggested that when learning in a basic-level context, an opportunity for simple comparison usually arises. Whereas children assume nouns refer to entire objects, they can reliably use one obvious property (shape) when extending nouns. Adjectives, on the other hand, can refer to any UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 number of properties, including but not limited to color, texture, flatness, and transparency. Hence, in situations in which objects vary in many ways (such as acrossbasic level categories), the child should have difficulty identifying which property is being referenced. Waxman and Klibanoff (2000) successfully demonstrated that children can extend adjectives in any situation, even across basic categories, as long as objects vary primarily along a single dimension. The research with respect to adjective learning implies that additional, unimportant information distracts children from information that is truly significant to the meaning of a word and thereby prevents them from extending words appropriately. The difficulty children show in extending adjectives in the presence of extraneous information predicts that verb learning should also be difficult. With respect to adjectives, researchers (and parents) can easily manipulate the learning environment to minimize the amount of irrelevant information. The context in which actions occur is less easily controlled. Whereas nouns refer to static objects, and adjectives apply to unchanging properties, verbs refer to transient events. Although children by the age of two seem to rely on somewhat consistent strategies for nouns and adjectives, the few verb extension studies suggest that young children cannot successfully pick out the elements within a scene relevant to the meaning of verbs. In a series of experiments, Kersten and Smith (2002) showed that 3 ½-4-year-old children attend as strongly to the appearance of novel objects as to their actions during verb-learning. In two experiments, Kersten and Smith (2002) exposed children and adults to numerous learning events depicting a particularly shaped object and its action. Though the learning events differed in a number of superficial ways, each time a participant heard either the noun or the verb, both the shape of the object and its actions remained constant. Thus, regardless of condition (noun/verb), the label could be interpreted as referring to either the shape of the agent or the action of the agent. At test, children were shown a series of similar events and asked whether the label could be extended to each event. Kersten and Smith (2002) used four different types of test images to assess children’s extension: an “object +match” event in which the shape of the agent body and the path were consistent with the accompanying label, a “no match” event in which neither the word nor the action were consistent with the accompanying label, an “object match” event, and a “motion match” event. Effect of Word Type Congruent with previous studies examining nounextension, both children and adults were willing to accept an event as an example of a noun as long as it involved a similarly shaped object, and ignored the motion of that object when making their decisions. Unlike adults, however, children did not attend only to the relevant information when learning verbs. Whereas adults accepted any test event that matched on action, regardless of the object involved, children were reluctant to accept an event as an example of the verb if either the object or its motion was changed. More interestingly, children were as likely to extend the verb label to an object match as they were to a motion match. In a third study in which Kersten and Smith (2002) included a familiar object condition, children were better at focusing primarily on actions in the verb condition. These experiments collectively suggest that although children by the age of four can extend verbs to actions, they can do so only in specific circumstances. Just as the young child’s focus during verb learning depends on the nature of the learning environment, it is possible children may focus on object form less in some circumstances than in others. Indeed, many researchers have questioned the idea that children always focus on object form when learning and extending nouns. One of the greatest controversies concerns the relative importance of perceptual information with respect to function in situations where function is apparent or important. The debate likely stems, in part, from evidence citing the importance of functional information in adult’s understanding of artifact categories (Rosch, Mervis, Gray, & Boyes-Braem, 1976; Malt and Johnson, 1992). Evidence also exists implying that functional information plays as least some role in young children’s extension of nouns. For example, in a study conducted by Kemler Nelson (1995), one of two possible functions of a target object was modeled for and practiced by 3-6 year old children. Children were asked whether or not the target’s name could be extended to a variety of test objects, some of which had the same general shape as the target and others that were shaped differently. In addition, half of the test objects could be inferred to support the demonstrated function, and the other half could be inferred to support the alternate function of the target object that had not been demonstrated. For all age groups, children’s acceptance of novel objects as new examples of the label depended on which function had been demonstrated. Perceptual similarities alone could not explain the pattern of results. . UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 Other studies have indicated that even when young children understand function, and recognize items that share functional properties as “similar”, they still primarily focus on shape in naming tasks (Merriman, 1993; Landau, Smith, & Jones, 1998). Landau, Smith, and Jones (1998) found that young children generalize names on the basis of shape even when functional information is emphasized. In their study, 3-year-old, 5-year-old, and adults all heard a novel object labeled. The experimenter explained and demonstrated the function of the object to half of the participants, but only named the object for the other half of participants. All participants were asked to extend the label to new objects, some of which could support the function of the target object, and others that were shaped the same could not support the target’s function. All age groups generalized on the basis of name when no functional information was given. However, when given functional information, only adults extended on the basis of function with novel objects. Even those children who demonstrated an understanding of the function were not affected by this knowledge when extending the label. Although most of the available evidence suggests that children focus on perceptual information even when functional information is available when learning nouns, neither of the test trials described above forced children to choose between function and form in word extension. Instead, children were asked about test objects individually, and were able to extend the label to as many objects, matching in form or function, that they deemed appropriate. The present study pits form directly against function in a forced-choice trial. In addition, within one study, we examine how varying the use of nouns and verbs may guide children’s attention to the perceptual or functional characteristics of objects. A No Word condition is included in the design to enhance these comparisons and provide a baseline for comparison between the noun and verb groups. Experiment 1 The goal of Experiment 1 was to examine how varying word type in a word extension task would direct 2-year-olds’ attention to the perceptual or functional properties of objects. The chief purpose of the current study was to investigate young children’s understanding of verbs, because we were also interested in whether children use different strategies for different word types we included both Noun and No Word conditions to allow us to more confidently interpret our findings. In addition, several researchers have found evidence for a “shape-bias” in noun learning, and to the extent that we Effect of Word Type replicate these findings we can confirm the validity of our methods. Including a No Word condition provides a baseline of how children respond without a given label, and enhances the comparison between the noun and verb groups. Few researchers have explored verb learning, and the research completed thus far has yielded complex results. While strategies for extending nouns to similarly shaped objects and adjectives by property appear to be relatively straightforward, children have difficulty with verbs. Given the complexity of actions, this difficulty is easy to understand. Some verbs refer to a certain result (i.e. squish, exit), and others to a specific manner of movement (i.e. press, run). Furthermore, the agent often assumes special significance, such that the meaning of a verb differs (at least superficially) according to whom or what is performing the action (a person skipping differs from a record skipping). These examples demonstrate that the context is important to the meaning of verbs, and offer some explanation for why 3 and 4-year-olds pay attention to both the agent and the action itself when learning new verbs (Kersten & Smith, 2002). The present study aims to investigate whether 2-1/2-year-olds will rely more on the function of the object or its shape in a forced-choice test when extending novel verbs. In addition, even if 2-1/2-yearolds are reluctant to adopt a specific strategy when extending verbs, the present study allows us at least to find out whether they are extending nouns and verbs differently. Previous research has indicated that showing children several objects from a single category can promote attention to “deeper” commonalities or properties. For this reason, additional category objects were included to promote a deeper appreciation of both perceptual and functional object qualities and to allow children to form 2 different categories based on perceptual and functional similarities. Our main goal was to determine whether children understand that different types of words refer to different aspects of a situation. To measure their understanding, we introduced novel objects and actions at the same time, used one label that could refer to either the action or the object being presented, and looked to see if changing the word-type of the label and general sentence structure affected children’s attention to the action or the object. We considered children’s extension of the label to an object that was either perceptually or functionally similar to the target as an adequate measure of their understanding that different word-types have different kinds of referents. We expect UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 children in the Noun condition perceptual match, children in the extend to the functional match, and Word condition to select between functional matches at chance. to extend to the Verb condition to children in the Nothe perceptual and Method Participants. Sixty 2-1/2-year-old children, 25 girls and 35 boys, with a mean age of 2;8 (range 2;42;10) were recruited from the greater San Antonio area to participate. Most participants completed the study in the Children’s Research Lab at Trinity University. Of these children, all spoke English and had begun producing utterances of two or more words at the time of participation, as determined by the MacArthur Communicative Development Inventory filled out by the parent on the day of the study. The remaining participants were recruited through several daycares located near Trinity University. Of these participants, parents signed consent forms to allow participation, but were not asked to complete the MacArthur Inventory. Parents and children received a small gift for participating, and parents received a results letter in the mail explaining the general goals and findings of the study at its completion. An additional 36 participants were recruited and participated in the study but their data were excluded from the analysis. Of these children, 13 were extremely distracted or did not complete all four trials of the study, 3 failed to pass a preliminary practice trial, 1 spoke only Spanish, and 17 were excluded due to experimenter errors. Thirteen children were omitted from the Noun condition, 9 from the Verb condition, and 14 from the No Word Condition. Stimuli. Four object sets designed for a previous study examining related issues were adopted for use in this study. In that earlier study, each set initially contained three objects: a target object designed to perform a specific action, a test object that shared the same shape as the target (perceptual match), and a test object designed to support the same action as the target (functional match). For this study, we used the same target and test objects but created an additional four category objects for each set. Two of the new objects for each set shared the target’s shape but did not support its function, and two of the new objects supported the target’s function but were shaped differently. Design. Participants were randomly assigned to one of three conditions (Noun, Verb, No Word). Children were shown the same objects with the same functions in all three conditions; only the script and Effect of Word Type word type of the label varied according to condition. Each child chose between a perceptual and functional match at test in four trials; choice at test (perceptual or functional match) was the dependent variable. The presentation of trials was random, whereas the presentation of the perceptual or functional category sets was counterbalanced within trials and across subjects. Procedure. Orientation. Families were greeted in the Trinity parking lot and escorted to the Children’s Research Lab. One experimenter spent a few minutes in casual play with the child while the other experimenter reviewed the consent form with the parent. Both experimenters played with the child for several minutes before beginning the study to help the child acclimate both to the experimenters and to the lab environment. The parent was invited to remain in the room during the study, but asked not to assist the child at test or repeat any of the instructions given the child by the experimenter during the procedure. As part of the consent form, parents were asked to allow their children to be videotaped for coding purposes. We also informed parents that we sometimes wished to use selected portions of the video outside of the lab, such as in classes and at conferences. Parents’ consent to allow outside viewing of their children’s video was optional; this consent did not affect whether the child could participate in the study. Practice Trial. Before beginning any of the trials, the experimenter demonstrated four familiar actions with four familiar objects. During demonstration with two of the four objects, the experimenter labeled the objects herself. For example, while simulating flight with a toy helicopter, the experimenter said, “Look, it’s a helicopter.” During demonstration with the other two objects, the experimenter labeled the objects’ actions. For example, while playing with a motorcycle the experimenter said “Vroom, vroom.” After having demonstrated specific actions with each of the toys, the experimenter laid the toys before the child and asked both a noun question and a verb question. For example, a noun question might be “Where’s the pig? Can you hand it to me?” A verb question might be “Which one was laughing? Can you hand it to me?” Following the child’s first response, the experimenter asked the second question. The order in which the questions were asked (noun or verb first) was counterbalanced across children. The purpose of the practice trial was to give children experience in handing various objects to the experimenter. We also used the trial to determine a UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 child’s ability to follow directions in a simple task. The data of children who answered both the noun question and verb question incorrectly in the Warm-up was excluded from the final analysis, (n = 3). Familiarization. The experimenter selected randomly one of the boxes containing the object sets. The experimenter first presented the target with a specific action (e.g., squishing the ball) while producing a noun (e.g., “Look. It’s a gep. See, it’s a gep.”), verb (“Look. I’m gepping it. See, I’m gepping it.”), or general language phrases (“See? Look. Wow!”). The experimenter repeated the appropriate phrase twice before encouraging the child to handle the target, “Can you play with the gep?” The experimenter then demonstrated the target again while producing the noun, verb, or general language phrase. Category Objects. After the presentation of the target object, the experimenter introduced the child to two category objects that shared the same shape and two category objects that shared the same function as the target. Each of the two perceptual category objects and each of the two functional category objects were presented as a set. Set presentation (perceptual/functional sets) was counterbalanced within trials and between subjects. The experimenter brought out both objects of each set simultaneously and both objects remained in view while the experimenter demonstrated the action with the alternate object. For example, the experimenter brought out both perceptual category objects at the same time, demonstrated a specific action with one, set it to the side in full view, and then demonstrated the same action with the other perceptual category object. In allowing the child to view similar items at the same time, we hoped to encourage the child to form categories based on different types of similarity (perceptual or functional similarity.) The experimenter demonstrated an action with each object twice, accompanied by a general language phrase designed to draw the child’s attention to the object’s perceptual quality in the Noun condition (See? Look at this one. Wow.), to the functional property in the Verb condition (See? Look what I can do with this. Wow.), and to no specific quality in the No Word condition (See? Look. Wow). Children were discouraged from exploring the category objects; however, in cases where children persisted, the experimenter ensured that the child explored one item of each set (one perceptual and one functional comparison item). Exploration of Test Objects. After the presentation of the comparison objects, the experimenter brought out Effect of Word Type the test objects, one perceptual and one functional match. At this point, the experimenter did not ask the child any questions, but simply encouraged the child to explore the test objects, “See? Look at these. What can you do with these?” If the child played exclusively with one object, the experimenter further encouraged the child to explore the neglected object. Touching and playing with the test objects was especially important in this experiment, because unlike in previous studies, the experimenter never demonstrated actions with the test objects. This play time was the only opportunity for children to discover what the test items could and could not do. In addition, Landau, Smith and Jones (1998) found that 2-year-olds were unable to judge whether objects could carry out their functions simply by looking at them. By encouraging children to touch and manipulate the objects for themselves, we hoped they would discover the physical qualities important to the objects’ functional capacities. Test. The experimenter removed the test items from view and then brought out the original target object. Once again, the experimenter demonstrated the target action while producing a noun (“Look. It’s a gep. See, it’s a gep.”), verb (“Look. I’m gepping it. See, I’m gepping it.”), or general language phrase (“See? Look. Wow!”). The experimenter put away the target object, and retrieved each of the test objects. In a forced-choice trial, the experimenter asked the child to extend the label to only one of the test objects, asking “Where’s the gep? Can you hand me the gep?” in the Noun condition or “Which one geps? Which one can gep?” in the Verb condition. In the No Word condition, the child was asked to simply hand the experimenter one of the objects, “Give me one. Which one?” Coding. The behavior of each participant was scored at least twice. First, each session was scored live. Experimenters worked in pairs. One recorded the child’s play with the target, exploration of test objects, and choice at test while the other conducted the experiment. Later, a third colleague watched the videotaped experiment and independently coded the child’s behavior a second time. Any discrepancies were resolved by a third-coder. Results Children’s choice at test was our dependent measure. As children were given only two choices (perceptual/functional) at test, the total number of perceptual choices was directly related to the total number of functional choices made by each child. Therefore, running analyses with respect to both choices UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 would have been redundant. We chose to use the number of perceptual choices at test as the dependent measure. The mean number of perceptual choices made in the Noun, Verb, and No-Word conditions were 2.55, 1.80, and 1.55, respectively, and are depicted in Figure 1. 3 * 2.5 2 1.5 1 0.5 0 Noun Verb No Word Condition Figure 1 Depiction of the mean number of perceptual choices by condition, Experiment 1. Significant differences exist between the mean number of perceptual responses in the Noun condition and the Verb condition, and between the Noun condition and the No word condition. A one-way ANOVA computed with Word (Noun, Verb, No Word) as a between subjects factor, and the number of perceptual choices as the dependent measure, revealed a significant main effect of Word, (F,(2,57) = 6.85, p<.01). Post-hoc independent samples t-tests with Bonferroni corrections showed that children in the Noun condition made more perceptual choices than did children in the Verb and No Word conditions (t(38) = 2.67, p<.02; t(38)= 3.57, p <.01, respectively.) One sample t-tests showed that the number of choices was significantly greater than chance in the Noun condition (t(19)=2.77, p <.02) and significantly less than chance in the No Word condition, (t(19) = -2.27, p <.05). These analyses help show how the children as a group responded differently depending on the word type they heard, but we were also wondered whether individual children showed consistency across the four trials. First, we examined the data with respect to general consistency; children were coded simply as being consistent (at least 3 out of 4 consistent responses across trials) or inconsistent (2 perceptual and 2 functional responses). In the Noun condition, 12/20 children were consistent; whereas in the Verb and No Word conditions, 13/20 in each condition were consistent. As would be expected given this breakdown, chi square analyses found no significant Effect of Word Type difference in simple consistency patterns between the three conditions. To more closely examine patterns of consistency, we gave consistent children more specific codes indicating perceptual or functional consistency. Children classified as perceptually consistent chose the perceptual match at test on at least 3 of the 4 trials. Likewise, children classified as functionally consistent were those that chose the functional match at test on at least of the 3 of the 4 trials. The number of consistent responses in each condition is available for closer inspection in Table 3. Chi square analyses confirmed significant differences in the patterning of perceptual consistency across the three conditions, χ2 (2, N = 30) = 6.19, p<.05. Chi square analyses confirmed significant differences in the patterning of functional consistency across the three conditions as well, χ2(2, N = 30) = 7.80, p<.05. Results from the chi square analyses cumulatively confirm significant differences in the patterning of consistent responses across the three conditions, when taking into account specific strategies adopted. Notice that most of the children who were consistent in the Noun condition were perceptually consistent, whereas most of the persistent children in the No Word condition were functionally consistent. Children in the Verb condition appear to be more closely split in terms of strategy. Experiment 2 Experiment 1 found evidence to support the shape bias, and weak evidence to support the idea that children are beginning to pay attention to action in the Verb condition. The design of Experiment 2 was identical to that of Experiment 1, except that we increased the salience of the functional information by ensuring that each of the targets, their functional category objects and functional test match objects, produced a noticeable change. 1 Suggestions we received at an international conference where we presented results from Experiment 1 influenced the decision to alter our stimuli. A lecture attended at this conference also inspired critical thought about the meaning of “function.” If it is said that an object has a function, does it not naturally follow that this object has a natural purpose, or that it can at least be used to accomplish some goal? Although it is true that the target object had to have certain properties to 1 We presented results from Experiment 1 at the International Conference of Infant Studies in May, 2004. Several researchers made this suggestion; we decided to take their advice. UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 Method Participants. Thirty 2-1/2-year-old children, 17 girls and 13 boys, with a mean age of 2;8 (range 2;42;10) participated in this experiment. Children were recruited in the same manner as in Experiment 1, and none of the children in Experiment 2 had participated in Experiment 1. An additional 42 participants were recruited and participated in the study, but their data were excluded from analysis. Of these children, 17 were very distracted or did not complete all four trials of the study, two scored “0 correct” on a preliminary practice trial, two did not speak English, and 22 were omitted due to experimental errors. Thirteen children were omitted from the Noun condition, 18 from the Verb condition, and 11 from the No Word condition. Stimuli. Three of the four stimuli sets used in Experiment 1 were used in Experiment 2. Two of these sets were modified, and one was left the same. The functional items (target, functional category objects, and functional test match) of the two sets were changed to increase their functional salience. We added magnets to the functional objects within the one set so that these objects now picked up metal discs when squished. With respect to the second set, a small wooden clipboard with attached nylon fibers was added so that instead of only brushing the floor, the functional items now could be used to straighten nylon fibers. The third set already had a salient function (making noise), so this set was not changed. A detailed description of how these sets were altered from Experiment 1 is available in Table 4. The items used in the fourth set were completely replaced. The new target was a cylinder consisting of two connected pieces that, when twisted, could be aligned to create a fun image. The corresponding functional category objects and test object were shaped differently than the target, but could also be twisted to create images. The perceptual category objects and test object were cylindrical like the target, but made of a single piece that could not be twisted. For a description of all the stimuli in this set, see Table 5. Design and Procedure. The design and procedure were identical to that of Experiment 1. At test, only the perceptual and functional test matches were placed before the child; any additional items used to complete the function (such as the metal disc in the Gep set and the clipboard in the Dax set) stayed in the box. These items were only available to the child during the Exploration of Test objects. Making these items available at test would have introduced the possibility of extension to the third item instead of to the perceptual or functional match. We did not wish to complicate our results, so we kept the number of choices available at test consistent across trials. Results The results of Experiment 2 are depicted in Figure 2. 2 Mean number of perceptual choices complete the demonstrated target actions in Experiment 1 (ie, an object must be soft if it is to be squished), it is doubtful that these object properties would in normal circumstances be identified as the “functions” of these objects. The demonstrated actions in most trials did not convey intent. They produced no meaningful change; they were, in a sense, inconsequential. We decided to run a study in which the actions we demonstrated actually conveyed the functions of the objects we used. For practical reasons, we wished to keep the same stimuli we had used in Experiment 1. We were able to creatively alter three of the four sets to meet our needs; the last set we completely replaced. We expected that objects capable of producing an interesting, noticeable change might better help children focus on functional object characteristics, and associate the nonsense words with the action associated with completing a concrete function. Effect of Word Type 1.5 1 0.5 0 Noun Verb No Word Condition Figure 2. Mean number of perceptual choices at test by condition, Experiment 2. No significant differences found between means. As in Experiment 1, we used the number of perceptual choices at test as the dependent measure. The mean number of perceptual choices made in the Noun, Verb, and No Word conditions were 1.90, 1.60, and 1.40, respectively (Figure 2). A one-way ANOVA computed with Word (Noun, Verb, No Word) as a between subjects factor, and the number of perceptual choices as UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 the dependent measure showed no significant main effect of word type, F(2,27) < 1. Independent samples t-tests computed for each condition showed that children failed to answer differently from chance in all three conditions; although the difference between the mean and chance level (2) approached significance in the No Word condition, t(9) = -2.25, p = .051. Consistency was coded in the same way as it had been in Experiment 1. First, we examined the data with respect to general consistency. In all three conditions, 7/10 children were simply consistent. We further examined the results in terms of perceptual and functional consistency; chi square analyses found no significant differences in the patterning of consistent responses across the three conditions. Individual responses are depicted in Table 4. It would have been interesting to compute an ANOVA with Experiment as a between subjects factor to examine how altering the functional objects changed children’s patterns of response. However, before beginning Experiment 2, we had completely replaced one of the four stimuli sets. Any change in the children’s patterning of response could have been due to the functional salience of the test items. To solve this problem, we could have excluded the data from the replaced set, and only included in analysis the data collected from the remaining three sets. We chose not to pursue this solution because of the unequal number of participants run in each study; at this time only 30 participants have been run in Experiment 2, whereas 60 were run in Experiment 1. Discussion The results of Experiments 1 indicate that even children as young as 2-1/2- years old are beginning to understand that nouns and verbs have different referents. We hypothesized that children in the Noun condition would choose the perceptual match, that children in the Verb condition would choose the functional match, and that children in the No Word condition would select at chance. We found that children in the Noun condition did indeed choose the perceptual match at test significantly more often than would be expected by chance. However, we did not find the results we expected with respect to the Verb and No Word conditions. Children chose at chance level in the Verb condition. Even more surprisingly, in the No Word condition children chose the functional match more than expected. Effect of Word Type The difference in responses between the Noun and Verb conditions implies that children at this age do have different expectations about the meaning of various word-types. But, although we found a significant difference between the number of perceptual choices that children made in the Noun and Verb conditions, this difference can be explained by the high proportion of perceptual choices children made in the Noun condition as opposed to a high proportion of functional choices made in the Verb condition. In other words, the results indicate that children appeared to understand that nouns refer to similarly shaped objects, but did not have a strong understanding that verbs refer to actions. The scattered pattern of response in terms of consistency in the Verb condition, when compared to the Noun and No Word conditions, further highlights the trouble children seem to be having when learning verbs. In the No Word condition, children chose the functional match at test more than we had expected. The nature of our task, without the cue of specific word type, drew children’s attention to the functional properties of objects. This finding enhances our interpretation of the children’s performance in the Noun condition. Not only did children in the Noun condition choose the perceptual choice at test significantly more often than would be expected from chance; they had to overcome a baseline that was significantly lower than chance. They were able to focus on perceptual qualities although the situation itself naturally directed their attention elsewhere. Children’s behavior in the Noun condition of Experiment 1 compliment previous studies supporting the shape bias (e.g., Baldwin, 1989; Landau, Smith & Jones, 1988; Smith, Jones, & Landau, 1992). We had hoped that increasing the functional salience of our stimuli in Experiment 2 would help children focus on the functional qualities of the test objects in the Verb condition. For reasons aforementioned, we could not directly compare the results from each of the studies using statistical analysis. Gathering from a casual inspection of the means, it appears that changing the functional salience of the objects did help children focus more on function in the Verb condition, as expected. However, at the same time, this shift in attention also seems to have also affected children’s choices across the board, in the Noun and No Word conditions as well as in the Verb condition. The mean number of perceptual choices in the Noun condition in Experiment 1 was 2.55; in Experiment 2 it dropped to 1.90. The mean number of perceptual choices made also dropped in the other two conditions, but not as drastically (1.8 to 1.6 in the Verb UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 condition, and 1.55 to 1.40 in the No Word condition). The increased functional salience eliminated the differences in responding between the conditions that was found in the Experiment 1. Furthermore, children failed to answer differently than would be expected from chance in all three conditions. Thus, contrary to what we expected, children in the Noun condition appear to have been most influenced by our change. This is puzzling given the results from most other research on the subject. Jones, Smith, and Landau (1991) found that 3 and 5-year-old-children extended nouns to similarly shaped objects more often than to objects that supported the same function as a target in a series of three experiments. Even when they defined objects in terms of their function (“Rifs are made by a special company so they can do this (demonstrate function,”) young children persisted in extending names to similarly shaped objects. Although we also found support consistent with the shape bias in Experiment 1, we did not in Experiment 2 when we increased the functional salience of the demonstrated target actions. Given the fact that Jones, Smith, and Landau (1991) emphasized function even more than we did during the introduction of their objects, one might naturally wonder why our results were so different. One possibility is that our stimuli differed in some important way from the stimuli used by Jones, Smith, and Landau. In both experiments conducted by Jones and colleagues (1991) using novel objects, the shapes of the stimuli they chose to use were well-defined. It might be that the more novel or unique the shape, the more likely it is for a child to identify that shape as the key feature important to the meaning of a new word. The objects that Jones Smith and Landau (1991) used in the first of their experiments were extremely novel. Thus, even when Jones, Smith and Landau stressed function, the shape still may have seemed more interesting or more salient to the children participating in the experiments. Though the objects they used in Experiment 2 were less novel than those used in Experiment 1, they still had well-defined, somewhat unusual shapes not normally associated with a variety of everyday objects. The stimuli we used in our studies, on the other hand, have less distinct shapes than ones used in other similar studies. Given this difference, perhaps our divergent results in Experiment 2 become less perplexing. The shapes of the objects we used may have been less interesting to children, such that when the functional salience of the objects was increased, children’s interest was successfully diverted away from Effect of Word Type perceptual properties, even in Noun condition. For example, we used an everyday round Tupperware dish, as our target in the “Koob” set. Not only had children likely encountered very similar items in their own kitchens, the shape itself is characteristic of many objects we see everyday. (Children in the Noun condition did indeed perform contrary to our hypothesis in this set in both Experiments 1 and 2, see Tables 5, 6.) Directly related to the idea that the novelty of the shape of the stimuli may play a role in the interpretation of words, is the idea that people may show reluctance in accepting a new name for objects for which they already have names. This idea, referred to as the Principle of Mutual Exclusivity (Markman and Watchel, 1988), may also help explain why many children failed to extend names based on perceptual similarity in the Noun condition. We noticed that with respect to some of our stimuli, several children showed that they already had words for the target items we introduced. For example, several children called the “Dax” target a “brush”, although we had altered what may be considered the standard brush shape by adding a second handle. These children obviously identified a brush as something with bristles, rather than by the happenstance fact that brushes usually have one handle. If they accepted the new label as a word at all, they may have equated it with the word “brush.” As a result, they may have been more likely to generalize the name based on familiar object parts than the novel target object as a whole, especially when the functional capacity of the bristles (to disentangle nylon thread, as in Experiment 2) was emphasized. As it turns out, children randomized to the Noun condition chose the perceptual match 12/20 times in the Dax trial in Experiment 1, and 5/10 times in Experiment 2. Whereas the use of novel stimuli may have increased the likelihood that children would choose perceptually in the noun condition, the use of novel stimuli in the Verb condition might be detrimental to verb learning. Kersten and Smith (2002) found that children are better able to focus on the motion rather than on the appearance of objects when familiar objects are used. Given this idea, we might expect that our objects sets that children likely identified as familiar and that might have been problematic in the Noun condition, might have facilitated verb-learning and appropriate extension of the verb to the functional match in the verb condition. For example, we believe that the “Gep” target has the most distinct shape when compared to the target stimuli used for the other three trials. For this reason, UWF.EDU/ARGOJOURNAL Argojournals: The Undergraduate Psychology Journal, Vol.1.1, Spring 2006 we might expect children randomized to the Noun condition to more closely behave at test according to our hypothesis. On the other hand, we might expect children randomized to the Verb condition to have more difficulty focusing on the action and function demonstrated with the novel “Gep” target, due to its novelty, especially in Experiment 2. These expectations are confirmed in Experiment 2; see Table 6. As we did not design our study to examine how specific differences in stimuli may be affecting children’s choices at test; our conclusions concerning the impact of these factors are tentative at best. The stimulus sets varied along a number of dimensions, such that it is impossible to determine the meaning of any difference in response between sets. The results from Experiments 1 and 2 dovetail with results from previous research indicating that verb learning is difficult for young children. Even when the target actions were designed to demonstrate concrete functions of objects, children had trouble extending verbs to the functional match at test. The change in responding in Experiment 2 implies that children are vulnerable to any change in stimuli, and that additional research is needed to determine what specific aspects of stimuli primarily influence children’s extension of nouns and verbs. In the future, we plan to more closely design our stimuli with these issues in mind. As our primary area of interest concerns children’s understanding and extension of verbs, we are considering including more familiar objects to better assess whether 2-year-olds can extend verbs to actions in easier circumstances. References Baldwin, D.A. (1989). Priorities in children’s expectations about object label reference: Form over color. Child Development, 60, 12891306. Booth, A.E, & Waxman, S. (2002). Object names and object functions serve as cues to categories for infants. Developmental Psychology, 38(6), 948957. Effect of Word Type impact of function on the naming of artifacts. Cognitive Development, 10, 347-354. Kersten, A.W., & Smith, L.B. (2002). Attention to novel objects during verb learning. Child Development, 73(1), 93-109. Landau, B., Smith, L.., & Jones, S. (1988). The importance of shape in early lexical learning. Cognitive Development, 3, 299-321. Landau, B., Smith, L., & Jones, S. (1998). Object shape, object function, and object name. Journal of Memory and Language, 38, 1-27. Malt, B., & Johnson, E.C. (1992). Do artifact concepts have cores? Journal of Memory and Language, 31, 195-217. Merriman, W., Scott, P., & Marazita, J. (1993). An appearance-function shift in children’s naming. Journal of Child Language, 20, 101-118. Rosch, E., Mervis, C., Gray, W., Johnson, & BoyesBraem, P. (1976). Basic objects in natural categories. Cognitive Psychology, 8, 382439. Smith, L.B., Jones, S.S, & Landau, B. (1992). Count nouns, adjectives, and perceptual properties in children’s novel word interpretations. Developmental Psychology, 28, 273-286. Waxman, S.R., & Klibanoff, R.S. (2000). The role of comparison in the extension of novel adjectives. Developmental Psychology, 36(5), 571-581. Waxman, S.R., & Markow, D.B. (1998). Object properties and object kind: 21-monthold infants’ extension of novel adjectives. Child Development, 69, 1313-1329. Jones, S., & Smith, L., & Landau, B. (1991). Object properties and knowledge in early lexical learning. Child Development, 62, 499-516. Kemler-Nelson, D. (1995). Principle-based inferences in young children’s categorizations: Revisiting the UWF.EDU/ARGOJOURNAL
© Copyright 2026 Paperzz