Previous work had established that children under 2 1/2 years old tend to interpret novel words in accordance with the Mutual Exclusivity principle, which is the default assumption that words will not have exemplars in common. So, for example, if shown something they can label and something they cannot, and asked, "Which one is a pilson?", even 18-month-olds will tend to select the unfamiliar kind. Selecting the familiar object would violate the Mutual Exclusivity principle. One problem is that this age group has not tended to reinterpret familiar words in line with the Mutual Exclusivity principle. That is, toddlers who believe that a Christmas ornament can be called "ball" have shown no tendency to stop believing this is response to learning to call it "ornament."

This null effect is theoretically important because alternatives to the Mutual Exclusivity principle have been proposed, ant these alternatives only predict the novel word effects. However, because tests of the familiar word effects may have been flawed, we decided to administer a new test. In Experiment 1 (N=32), 24- to 25-month-olds heard stories in which a novel noun was used for an atypical exemplar of a familiar noun. When asked to select exemplars of the familiar noun, they showed a small, but reliable tendency to avoid the object from the story. In Experiment 2 (N=16), the novel nouns in the stories were replaced by pronouns and proper names, and the children did not avoid the story object in the test of the familiar noun. Thus, the aversion to this object that was observed in Experiment 1 was not due to its greater exposure or its being referenced immediately before testing, but to toddlers' Mutual Exclusivity bias. Their bias is hypothesized to be a form of implicit probabilistic knowledge that derives from the competitive nature of category retrieval. We present a mathematical model in the paper that generates all of the empirical effects consistent with the Mutual Exclusivity bias.

We explored one of the novel word effects of the Mutual Exclusivity principle in the verb domain. After presenting a linguistic analysis of what it means to say that two verbs have exemplars in common, four experiments are described in which young 2-year-olds were shown pairs of videotaped actions, one familiar and one novel, and asked to say which were the referents of novel verbs. For actions that did not involve objects, the toddlers tended to favor the novel action in each experiment. For example, they chose the woman who was turning in circles while leaning backwards as "the one who is glarving" more often than the woman who was running. For actions involving objects, novel actions (e.g., shuffling balls) were chosen more often than familiar ones (e.g., kicking balls) in only two of the four experiments. The main objective of the series of experiments became that of isolating the cause of this action type effect. Several hypotheses were disconfirmed by our results. The account most compatible with the data was that when children viewed two actions involving objects, they tended to covertly generate the names for the objects (e.g., "balls"), which interfered with their comparing the test verb with the verbs they were able to retrieve for the two actions. The preference for novel actions was also found to be strengthened by preexposing both actions from a test pair, but to be unaffected by preexposing just the novel actions. Such preexposure effects are evidence that the problem of having to identify the likely referent of an unfamiliar action verb is one that exceeds the working memory capacity of many toddlers.

If young 2-year-olds are read stories in which a certain word-initial phoneme (e.g., l-) and certain word-final rime (e.g., -at) occur repeatedly (e.g., "Lucky Lucy sat on a cat. Look at that..."), they do better on a subsequent novel word mapping test if the test word is composed of these sounds (e.g., lat) than if it is not (e.g., ged). That is, after hearing this story, they are more likely to map lat onto an unfamiliar rather than a familiar kind of object than they are to map ged in this way. When I describe this experiment to other cognitive psychologists, they usually predict the opposite result.

We obtained the result in three experiments, when the name of the familiar choice object sounded like that of the test word (e.g, hat) as well as when it did not (e.g., shoe). I believe that two mechanisms are responsible, part priming and induced distinctiveness. Repeated exposure to a particular word sound reduces the processing resources required to maintain a working memory representation of the novel word that contains the sound; the freed-up resources are then applied to other processes that promote mapping the novel word onto the novel kind (e.g., retrieving the name of the familiar kind, detecting its mismatch with the novel word, etc.) Second, exposure to a set of highly similar stimuli causes one to exaggerate small differences within the set (Krumhansl, 1978, Psych Review). Thus, exposure to several words that end in -at causes one to perceive a greater difference between lat and hat than if one had not been exposed, which promotes mapping lat onto the unfamiliar kind rather than the hat.

A new word learning phenomenon, called the nominal passover effect, was demonstrated in 3-year-olds. In Study 1, the children were shown a pair of objects and heard a novel label used repeatedly for one, but not for the other. In a forced choice test of generalization of the label, the "passed over" object was selected less often by the children than one that had not been present during training. This passover effect was the same whether the speaker had completely ignored the comparison object during training or had referred to it with pronouns. The performance of a no word control group (N=24) indicated that the effect was not due to a preference for the less exposed of the two choice objects. The passover effect is interesting because it shows that children can learn something about a word's meaning from observing how it is not used.

A pragmatic principle, called Exhaustive Reference, was advanced to account for the passover effect. It stipulates that whenever a new generic word is used to name something, expect it to be extended to all entities in a situation that the speaker perceives and believes to be exemplars of the name. Speakers obey this principle because when they name something, they want the addressee to learn the concept that the name conveys; identifying all available exemplars promotes this learning. Note that the principle applies to speech acts of naming (e.g., "This is a sextant."), but not to other speech acts in which a novel name might occur (e.g., "The sextant is broken.") The latter do not necessarily have a pedagogical purpose.

In Study 2, the nominal passover effect was replicated with three new sets of objects and with training language that contained only indefinite forms of reference. The passover experience was often sufficient to counteract children's tendency to generalize a novel label on the basis of perceptual similarity. Although the passover effect was not evident in free choice name generalization tests in either study, it was in a study with 5-year-olds that we reported at the SRCD convention in 1995.

This research was more exploratory than hypothetico-deductive, and as such had a few more loose ends than I'd like. The question of how children label deceptive stimuli (e.g., for an eraser that looks like a pencil; for the color of a blue cup that has a yellow filter in front of it) was judged to be relevant to three trends in preschool cognitive development: the increase with age in the degree to which non-apparent properties influence decisions about what to call things (Keil, 1989); development of the appearance-reality distinction; and the increasing tendency to assume mutual exclusivity in default situations (i.e., ones low in information about the actual relation between two words).

Three-, 4-, and 5-year-olds were asked to select labels for deceptive stimuli. Three types of labeling were investigated -- simple (e.g., 'is an eraser'); appearance-predicated (e.g., 'looks like an eraser'); and reality-predicated (e.g., 'is really and truly an eraser.') An appearance-reality shift in the simple labeling of these stimuli was observed (i.e., when simply asked what something "is," an increase with age in the tendency to accept the name for its reality relative to the tendency to accept the name for its appearance.) The shift was less pronounced with object than with color labels perhaps because of object-specific learning effects. As in Flavell's research, older children mapped different appearance- than reality-predicated labels onto an item more often than younger children did. However, the three age groups were equally reluctant to extend more than one name via a common predicate (i.e., to accept two simple, two appearance-predicated, or two reality-predicated labels.) This one-label-per-predicate pattern was observed more frequently within reality than within appearance predicates, with color than with object names, and with questions blocked by predicate than by name. It is argued that younger children maintained this pattern because of inflexible encoding, but that older ones did so because of better understanding of the appearance-reality distinction and a stronger default mutual exclusivity bias.

The two heuristics were Mutual Exclusivity (of course) and use of a feature that has been highlighted in input as a necessity condition (what I call, "criterial use of a highlighted feature.") The latter is relevant to a concern of recent cognitive developmental researchers about young children's ability to apply rules to a series of related decisions. Thirty-six Serbian-English bilinguals and 42 English monolinguals, who were between 5 and 8 years old, were taught a novel English name in such a way that a physical feature of the training object was highlighted both verbally and functionally. The children were then asked to identify other exemplars of the name from a set. Those who restricted their choices to objects that possessed the feature that had been highlighted were scored as having adopted the "criterial use" heuristic. More older than younger children, and more mono- than bilinguals, used this heuristic. Its use was highly positively correlated with children's receptive vocabulary scores, especially in the bilingual group.

The test of Mutual Exclusivity was how children reacted to being told a second novel name for something that they had selected in the generalization test of the first trained name. The older children were more likely than the younger ones to preserve Mutual Exclusivity by correcting and restricting the first trained name's extension (i.e., so that it did not overlap with that of the second name). Mono- and bilinguals were equally likely to maintain Mutual Exclusivity when the second label was introduced in English (the same language as the first label), but more bilinguals than monolinguals allowed the two labels to overlap when the second label was introduced as a word from French. The results support the view that monolingual and bilingual acquisition foster the development of different word learning skills and expectations.

This was the first published demonstration that children favor a novel kind of action over a familiar kind as the likely referent of a novel verb. In Experiment 1, 4-year-olds selected the novel action about twice as often as the familiar one. When the unfamiliar action had been preexposed, but the familiar one had not, the mapping preference was eliminated; however, only the children with the smaller vocabularies (i.e., below the mean for the sample) were susceptible to this preexposure effect. In Experiment 2, 4-year-olds mapped a novel noun onto an unfamiliar object about five times as often as they mapped it onto a familiar object, and this tendency was not affected by preexposure. Even when action embedding and question complexity were controlled, the disambiguation effect was stronger for object than for action words. These results are quite different that what we recently obtained from children two years younger. An account was presented in which two lexical principles, Mutual Exclusivity and Feeling of Novelty, are hypothesized to apply differently to action than to object words for 4-year-olds. The results are compatible with some theorists' suggestion that default principles for making quick decisions about new words, such as Mutual Exclusivity, have less impact on children's verb learning than on their object name learning.

The relative importance of appearance and function in children's object naming was examined. In Study 1, 3 1/2-, 4 1/2-, and 6-year-olds were taught novel names for unfamiliar objects, then had to decide whether these applied to items that resembled the training objects in either appearance or function. The youngsters were also shown deceptive objects (e.g., an eraser that looked like a pencil) and had to choose between familiar appearance and function names for them (e.g., pencil or eraser). The frequency of function-based responding in both tasks increased with age. In Study 2, the name training procedure was revised so that equal emphasis was given to both apparent and functional features. The main results of the first study were replicated. Neither study obtained evidence of a strong relation between the appearance-function shift and increased understanding of the appearance-reality distinction.

Basically, I replied to Woodward and Markman's critique of my monograph with Bowman and updated the literature review. There was actually more agreement than disagreement between us. The disagreement mostly concerned children 2 years and under. In the three studies in the monograph that included young 2-year-olds, this age group showed none of the Mutual Exclusivity-preserving effects, compelling Bowman and I to conclude the null hypothesis. Woodward and Markman raised all the familiar kinds of arguments against concluding the null. I acknowledged that they might be right and cited Merriman and Schuster's (1991) evidence for a novel word interpretation effect in this age group that is not an artifact of preference for novelty. But I also argued that the jury was still out on this age group because no one had shown that they produced the familiar word reinterpretation effects that the older children do. That is, young 2-year-olds have shown no tendency to correct their tendency to call an object by an old name, or to refrain from generalizing an old name to a new object, in response to the introduction of a new name for the object. Of course, the jury came this past year with Merriman & Stevenson's (in press) demonstration of a weak, but statistically significant tendency of this sort in young 2-year-olds.

In the monograph, I found that the disambiguation effect, which is the tendency to select unfamiliar rather than familiar things as the referents of new names, could be reversed in young 2-year-olds, but not older children, by preexposing the unfamiliar objects. We interpreted this result as indicating that attraction to novelty accounts for toddlers' choices of referents for new names, but that a Mutual Exclusivity principle and/or an expectation that lexical gaps will be filled determines older children's selections. Both the disambiguation effect and its reversal by preexposure in 2-year-olds were replicated in the present study; however, the 2-year-olds' rate of selecting unfamiliar over familiar kinds was less when they were simply asked to choose between the items (i.e., in a no word control condition) than when they were asked to identify referents of unfamiliar names (i.e., in the new word condition). We thus concluded that some toddlers may have both an attraction to novelty and a tendency to honor abstract principles for integrating the meanings of words.

Both 2- and 4-year-olds' disambiguation decisions were also affected by object typicality and word similarity. That is, when the familiar kind was atypical, or when the novel word sounded somewhat similar to the name for the familiar kind (e.g., glower vis a vis flower), the disambiguation effect was reduced. Correlations between the tendency to acknowledge lack of familiarity with a name and to treat it like a phonologically similar familiar name suggested that phonological matching skill may play a central role in 2- and 4-year-olds' interpretation of new names. Also, 4-year-olds' who most often mapped distinctive-sounding new names to unfamiliar kinds also admitted their unfamiliarity with these names most frequently, suggesting that children's increasing awareness of their own knowledge may begin to affect their lexical processing around the age of four.

The goal was to demonstrate experimentally how preexisting nonlinguistic categories could affect young children's interpretation of a novel name. The experiments assumed that children could learn object categories through a process of attribute covariation detection, as both Younger and Mervis had proposed. The best demonstration was the third experiment. Two groups of 3-year-olds were allowed to play with a set of eight toy creatures. The sets shared four creatures in common. The sets were structured so that different attributes covaried. For example, in one set, body color, ear shape, and hair color covaried perfectly (e.g., all the dark blue ones had floppy ears and purple hair), but body length and number of legs did not covary with any other attribute. In the other set, the opposite was true. After play, both groups were taught a name for one of the creatures, then were asked which of a pair of creatures also had the name. One choice matched the training object in body color, ear shape, and hair color and the other one matched it in body length and number of legs. (Note that the children had encountered both choice objects during play.) As predicted, the youngsters favored the object that matched the training object in the attributes that they had observed to covary in the set of creatures. Thus, the 3-year-olds' preexisting nonlinguistically-derived categories influenced their interpretation of a novel word.