Learning words and learning sounds: Advances in language development

Phonological development is sometimes seen as a process of learning sounds, or forming phonological categories, and then combining sounds to build words, with the evidence taken largely from studies demonstrating ‘perceptual narrowing’ in infant speech perception over the first year of life. In contrast, studies of early word production have long provided evidence that holistic word learning may precede the formation of phonological categories. In that account, children begin by matching their existing vocal patterns to adult words, with knowledge of the phonological system emerging from the network of related word forms. Here I review evidence from production and then consider how the implicit and explicit learning mechanisms assumed by the complementary memory systems model might be understood as reconciling the two approaches.     

Processing of color words activates color representations

Two experiments were conducted to investigate whether color representations are routinely activated when color words are processed. Congruency effects of colors and color words were observed in both directions. Lexical decisions on color words were faster when preceding colors matched the color named by the word. Color-discrimination responses were slowed down when preceding color words mismatched the test color even if no task had to be performed on these words. These findings are consistent with the experiential view of language comprehension according to which color perception and the comprehension of color words are based on overlapping representational resources.     

Whole words and decoding for short-term learning: comparisons on a "talking-computer" system.

Low-and average-ability readers in first and second grade studied a list of 36 words using a "talking-computer" system. The system highlighted and simultaneously pronounced orthographic units in the words when the children touched the words with a light pen. During two training sessions, the computer presented four groups of 9 words each, one group as whole words, one in syllabic units, one in subsyllabic units, and one as single grapheme-phoneme units. All children learned the least words with single grapheme-phoneme units, having had the greatest difficulty blending the units into words during training. The other presentation units did not differ significantly from each other for most students on post-testing. However, the low first-grade readers learned fewer words segmented and presented by subsyllables than by syllable or word units, but only for multisyllabic words. Monosyllabic words were blended and learned as easily with onset-rime segmentation as with whole word units, for all children.     

Children's mappings of large number words to numerosities

Previous studies have suggested that children's learning of the relation between number words and approximate numerosities depends on their verbal counting ability, and that children exhibit no knowledge of mappings between number words and approximate numerical magnitudes for number words outside their productive verbal counting range. In the present study we used a numerical estimation task to explore children's knowledge of these mappings. We classified children as Level 1 counters (those unable to produce a verbal count list up to 35), Level 2 counters (those who were able to count to 35 but not 60) and Level 3 counters (those who counted to 60 or above) and asked children to estimate the number of items on a card. Although the accuracy of children's estimates depended on counting ability, children at all counting skill levels produced estimates that increased linearly in proportion to the target number, for numerosities both within and beyond their counting range. This result was obtained at the group level (Experiment 1) and at the level of individual children (Experiment 2). These findings provide evidence that even the least skilled counters do exhibit some knowledge of the form of the mapping between large number words and approximate numerosities.     

Why children learn color and size words so differently: evidence from adults' learning of artificial terms.

An adult simulation study examined why children's learning of color and size terms follow different developmental patterns, one in which word comprehension precedes success in nonlinguistic matching tasks versus one in which matching precedes word comprehension. In 4 experiments, adults learned artificial labels for values on novel dimensions. Training mimicked that characteristic for children learning either color words or size words. The results suggest that the learning trajectories arise from the different frames in which different dimensions are trained: Using a comparison (size-like) training regimen helps learners pick out the relevant dimension, and using a categorization (color-like) training regimen helps the learner correctly comprehend and produce dimension terms. The results indicate that the training regimen, not the meanings of the terms or the specific dimensions, determines the pattern of learning.     

Learning to use words: Event-related potentials index single-shot contextual word learning

Humans have the remarkable capacity to learn words from a single instance. The goal of this study was to examine the impact of initial learning context on the understanding of novel word usage using event-related brain potentials. Participants saw known and unknown words in strongly or weakly constraining sentence contexts. After each sentence context, word usage knowledge was assessed via plausibility ratings of these words as the objects of transitive verbs. Plausibility effects were observed in the N400 component to the verb only when the upcoming novel word object had initially appeared in a strongly constraining context. These results demonstrate that rapid word learning is modulated by contextual constraint and reveal a rapid mental process that is sensitive to novel word usage.     

On the naming of color words and color patches

Subjects can name color words faster than they can name color patches. To account for that effect, a generic model of naming is described which assumes that words access the mental lexicon directly, whereas color patches do so only indirectly via an initial imaginal or semantic representation. However, Lund (1927) reported that the naming advantage for words disappeared when all the items to be named on a page were the same (i.e., they were blocked). In the present study, three experiments are reported that were designed to provide a clearer empirical definition of Lund’s blocking effect and to ascertain the extent to which it requires a modification of the generic model. The blocked lists had 50 items arranged into 10 blocks, with each block homogeneous with respect to color. The block lengths were either all a predictable length of 5 items or they varied randomly from 1 to 9 items. The data indicated the following: (1) The blocking effect occurred even when the task required a full identification of each item, and (2) the blocking effect was confined to within-block transitions. Blocking seemed to eliminate the word advantage by allowing the subject to re-use the lexical entry used for the immediately prior item, which is consistent with the generic model.     

EPS Prize Lecture. Learning to read words

The ease with which we process the written word belies its complexities and makes it easy to forget that it is a highly skilled behaviour and one that takes time to master. In this paper, I argue that our ability to read words has its roots in our capacity for language. Good progress has been made towards understanding how children discover the systematic relationship between speech sounds and the letters used to represents those sounds, very early in reading development. However, we understand much less about how beginning readers become skilled readers. To understand this, I argue that it is important to view the visual word recognition system within the context of a broader language system, one that incorporates a rich network of semantic and episodic knowledge.     

Learning about sounds contributes to learning about words: Effects of prosody and phonotactics on infant word learning

This research investigates how early learning about native language sound structure affects how infants associate sounds with meanings during word learning. Infants (19-month-olds) were presented with bisyllabic labels with high or low phonotactic probability (i.e., sequences of frequent or infrequent phonemes in English). The labels were produced with the predominant English trochaic (strong/weak) stress pattern or the less common iambic (weak/strong) pattern. Using the habituation-based Switch Task to test label learning, we found that infants readily learned high probability trochaic labels. However, they failed to learn low probability labels, regardless of stress, and failed to learn iambic labels, regardless of phonotactics. Thus, infants required support from both common phoneme sequences and a common stress pattern to map the labels to objects. These findings demonstrate that early word learning is shaped by prior knowledge of native language phonological regularities and provide support for the role of statistical learning in language acquisition.     

Toddlers' attention to intentions-in-action in learning novel action words.

Toddlers' ability to use cues such as eye gaze and gestures to infer the meaning of novel action words was examined. In Experiment 1, 21- and 27-month-olds were taught labels for pairs of videotaped actions that were either similar or dissimilar in appearance. Similar actions differed mainly in the presence of behavioral cues related to the agents' intentions (e.g., extended arms). Only the older children were able to learn the labels for the similar actions. In Experiment 2, 3 new pairs of labels (2 similar, 1 dissimilar) were taught to children in the same age range. Eye gaze and gestures were the main features distinguishing the similar events. The same developmental effect was observed, with only the older children showing learning of both types of verbs and the younger children being impeded by the appearance of the actions. The results show that by the middle of the 2nd year, children begin to consider intentions-in-action when acquiring the meaning of novel action verbs.     

Associations to unfamiliar words: Learning the meanings of new words

Five experiments were designed to examine whether subjects attend to different aspects of meaning for familiar and unfamiliar words. In Experiments 1–3, subjects gave free associations to high- and low-familiarity words from the same taxonomic category (e.g.,seltzer:sarsparilla; Experiment 1), from the same noun synonym set (e.g.,baby:neonate; Experiment 2), and from the same verb synonym set (e.g.,abscond:escape; Experiment 3). In Experiments 4 and 5, subjects first read a context sentence containing the stimulus word and then gave associations; stimuli were novel words or either high- or low-familiarity nouns. Low-familiarity and novel words elicited more nonsemantically based responses (e.g.,engram:graham) than did high-familiarity words. Of the responses semantically related to the stimulus, low-familiarity and novel words elicited a higher proportion of definitional responses [category (e.g.,sarsparilla:soda), synonym (e.g.,neonate:newborn), and coordinate (e.g.,armoire:dresser)], whereas high-familiarity stimuli elicited a higher proportion of event-based responses [thematic (e.g.,seltzer:glass) and noun:verb (e.g.,baby:cry)]. Unfamiliar words appear to elicit a shift of attentional resources from relations useful in understanding the message to relations useful in understanding the meaning of the unfamiliar word.     

Learning in context: linguistic and attentional constraints on children's color term learning

Three experiments investigated whether linguistic and/or attentional constraints might account for preschoolers' difficulties when learning color terms. Task structure and demands were equated across experiments, and both speed and degree of learning were compared. In Experiment 1, 3-year-olds who were matched on vocabulary score were taught new secondary color terms by corrective, semantic, or referential linguistic contrast. Corrective contrast produced more rapid and more extensive learning than did either semantic or referential contrast, supporting the hypothesis that targeted linguistic feedback facilitates learning. Experiment 2 replicated and extended the first experiment with Italian children and found cross-cultural differences in the amount learned about colors named differently in the two languages. In Experiment 3, some of the children were introduced to the new terms within a context of enhanced perceptual salience. These children learned as fast and performed as accurately as those given corrective linguistic feedback in Experiment 1.     

Retrieval of concrete words involves more contextual information than abstract words: multiple components for the concreteness effect

The current study used the directed forgetting paradigm in implicit and explicit memory to investigate the concreteness effect. Event-related potentials (ERPs) were recorded to explore the neural basis of this phenomenon. The behavioral results showed a clear concreteness effect in both implicit and explicit memory tests; participants responded significantly faster to concrete words than to abstract words. The ERP results revealed a concreteness effect (N400) in both the encoding and retrieval phases. In addition, behavioral and ERP results showed an interaction between word concreteness and memory instruction (to-be-forgotten vs. to-be-remembered) in the late epoch of the explicit retrieval phase, revealing a significant concreteness effect only under the to-be-remembered instruction condition. This concreteness effect was realized as an increased P600-like component in response to concrete words relative to abstract words, likely reflecting retrieval of contextual details. The time course of the concreteness effect suggests advantages of concrete words over abstract words due to greater contextual information.     

Number gestures predict learning of number words

When asked to explain their solutions to a problem, children often gesture and, at times, these gestures convey information that is different from the information conveyed in speech. Children who produce these gesture‐speech “mismatches” on a particular task have been found to profit from instruction on that task. We have recently found that some children produce gesture‐speech mismatches when identifying numbers at the cusp of their knowledge, for example, a child incorrectly labels a set of two objects with the word “three” and simultaneously holds up two fingers. These mismatches differ from previously studied mismatches (where the information conveyed in gesture has the potential to be integrated with the information conveyed in speech) in that the gestured response contradicts the spoken response. Here, we ask whether these contradictory number mismatches predict which learners will profit from number‐word instruction. We used the Give‐a‐Number task to measure number knowledge in 47 children (M_age = 4.1 years, SD = 0.58), and used the What's on this Card task to assess whether children produced gesture‐speech mismatches above their knower level. Children who were early in their number learning trajectories (“one‐knowers” and “two‐knowers”) were then randomly assigned, within knower level, to one of two training conditions: a Counting condition in which children practiced counting objects; or an Enriched Number Talk condition containing counting, labeling set sizes, spatial alignment of neighboring sets, and comparison of these sets. Controlling for counting ability, we found that children were more likely to learn the meaning of new number words in the Enriched Number Talk condition than in the Counting condition, but only if they had produced gesture‐speech mismatches at pretest. The findings suggest that numerical gesture‐speech mismatches are a reliable signal that a child is ready to profit from rich number instruction and provide evidence, for the first time, that cardinal number gestures have a role to play in number‐learning.     

Learning at a distance I. Statistical learning of non-adjacent dependencies

In earlier work we have shown that adults, young children, and infants are capable of computing transitional probabilities among adjacent syllables in rapidly presented streams of speech, and of using these statistics to group adjacent syllables into word-like units. In the present experiments we ask whether adult learners are also capable of such computations when the only available patterns occur in non-adjacent elements. In the first experiment, we present streams of speech in which precisely the same kinds of syllable regularities occur as in our previous studies, except that the patterned relations among syllables occur between non-adjacent syllables (with an intervening syllable that is unrelated). Under these circumstances we do not obtain our previous results: learners are quite poor at acquiring regular relations among non-adjacent syllables, even when the patterns are objectively quite simple. In subsequent experiments we show that learners are, in contrast, quite capable of acquiring patterned relations among non-adjacent segments-both non-adjacent consonants (with an intervening vocalic segment that is unrelated) and non-adjacent vowels (with an intervening consonantal segment that is unrelated). Finally, we discuss why human learners display these strong differences in learning differing types of non-adjacent regularities, and we conclude by suggesting that these contrasts in learnability may account for why human languages display non-adjacent regularities of one type much more widely than non-adjacent regularities of the other type.     

Learning to reproduce a list and memory for the learning.

The ability to reproduce from memory a short list of verbal items immediately following presentation is known to improve over successive trials on that list, even if these trials are embedded in a sequence of trials on other lists of the same sort (Hebb, 1961). Less clear is whether this "Hebb effect" arises without the list repetition being noticed. This question has long been pondered and has recently taken on particular theoretical significance, but the available evidence is scant and inconsistent. Two experiments are described in which, in essence, a sequence of immediate reproduction trials was followed by tests that called for list recognition (Experiments 1 and 2) and/or estimates of list presentation frequency (Experiment 1). These tests provided quantitative measures of repetition awareness. Typical Hebb effects were found, but there was no evidence that the effects occurred without the subjects' being aware of the repetition; effect-size analyses indicated that both the recognition and frequency responses were more sensitive to repetition than were the reproduction responses. Therefore, not only could the recognition and frequency responses not have been made solely on the basis of how readily the test lists were reproduced, but the Hebb effect could have required an awareness of repetition.