Developmental differences in the effects of phonological, lexical and semantic variables on word learning by infants

Abstract

The influence of phonological (i.e., individual sounds), lexical (i.e., whole-word forms), and semantic (i.e., meaning) characteristics on the words known by infants age 1;4 to 2;6 was examined, using an existing database (Dale & Fenson, 1996). For each noun, word frequency, two phonological (i.e., positional segment average, biphone average), two lexical (i.e., neighborhood density, word length), and four semantic variables (i.e., semantic set size, connectivity, probability resonance, resonance strength) were computed. Regression analyses showed that more infants knew (1) words composed of low probability sounds and sound pairs, (2) shorter words with high neighborhood density, (3) words that were semantically related to other words, both in terms of the number and strength of semantic connections. Moreover, the effect of phonological variables was constant across age; whereas the effect of lexical and semantic variables changed across age.Keywords word learning; vocabulary; phonotactic probability; neighborhood density; semantics Three types of representations appear to play a role in word learning: phonological, lexical, and semantic (e.g., Gupta & MacWhinney, 1997). Phonological representations refer to individual sounds (e.g., /k/, /ae/, /t/). Lexical representations refer to whole-word forms (e.g., / kaet/). Semantic representations refer to the meaning or referent of a word (e.g., 'small furry four-legged pet'). The simplest example of word learning is one in which a novel object labeled by a correctly perceived and articulated novel sound sequence is learned. In this case, it is assumed that when the novel word is encountered it activates existing phonological representations. These, in turn, activate lexical representations; however, a novel word will not exactly match an existing lexical representation. Likewise, a novel word will not exactly match an existing semantic representation. Thus, the formation of new lexical and semantic representations presumably is initiated. Word learning consists of creating new lexical and semantic representations, linking these new representations to one another, and integrating these new representations with existing phonological, lexical, and semantic representations.A limitation of current models of word learning is that many aspects of word learning are not fully specified, particularly as related to the phonological, lexical, and semantic characteristics of the ambient language. That is, many models do not specify whether the influence of ambient characteristics on the formation of new representations is similar across phonological, lexical, and semantic representations, and whether this influence might change over developmental time as more words are acquired. In part, these aspects of word learning are underspecified because the necessary data are limited. The goal of this paper is to provide preliminary evidence to address these three issues to promote elaboration of existing models as well as future research. Phonological, Lexical, and Semantic Charac...

Sentence-final word completion norms for European Portuguese children and adolescents

“…We also hypothesized that the number of semantic and structural errors would be higher in children than in adolescents. This would be consistent with previous studies suggesting that there is a refinement of knowledge structures throughout development (e.g., Bjorklund, 1987;Bjorklund & Marchena, 1984;Chi & Ceci, 1987;Gathercole et al, 1992;Munson et al, 2005;Schneider & Pressley, 1997;Storkel, 2002Storkel, , 2009Swingley, 2003;Vicente et al, 2003), thought to affect memory processes and word selection, as in a sentence completion task.…”

“…Together, these findings suggest that, throughout development, the structure of semantic memory undergoes changes not only in terms of the number of items in semantic networks, but also in terms of the richness of representations and items' accessibility (e.g., Bjorklund, 1985Bjorklund, , 1987Chi & Ceci, 1987;Gathercole, Willis, Emslie, & Baddeley, 1992;Munson, Swenson, & Manthei, 2005;Schneider & Pressley, 1997;Storkel, 2002Storkel, , 2009Swingley, 2003;Vicente, Castro, & Walley, 2003). Also, they indicate that these changes are supported by the dynamic interaction of individual factors (e.g., maturation) and environmental variables (e.g., education) (see Thomas & Karmiloff-Smith, 2002, 2003.…”

“…B. Markman & Dietrich, 2000; E. M. Markman, 1990;E. M. Markman & Hutchinson, 1984;Nelson, 1977;Smiley & Brown, 1979), and lexical representations tend to become more segmental and less holistic, as vocabulary increases (Elbro, 1996;Storkel, 2002Storkel, , 2009. There is also evidence pointing to changes in the relative influence of contextual and stimulus factors on word recognition as reading fluency develops, with young children relying more on contextual information to aid word recognition and adults showing increased automatic word processing abilities (Schwantes, Boesl, & Ritz, 1980;West & Stanovich, 1978;West, Stanovich, Feeman, & Cunningham, 1983).…”

The Effects of Phonotactic Probability and Neighborhood Density on Adults' Word Learning in Noisy Conditions

“…Specifically, adults learned more words with low phonotactic probability sound sequences than words with high phonotactic probability sound sequences when learning was measured via partially correct responses, which was assumed to index triggering. Words with low phonotactic probability are presumably more easily recognized as new words, triggering word learning efficiently (Storkel, 2009;Storkel et al, 2006;Storkel & Lee, 2011), because they sound less like other known words, whereas words with high phonotactic probability sound very similar to known words (Frisch, Large, & Pisoni, 2000;Vitevitch et al, 1997). That is, a new word may be mistaken for a known word when phonotactic probability is high, failing to trigger learning of the new word.…”

Contextual vocabulary acquisition: from algorithm to curriculum

“…Learning new concepts and their words is not simply "additional knowledge" or learning a definition. Concept learning requires making ever more refined discriminations of ideas, actions, feelings, and objects; it necessitates "assimilating" (Piaget 1952), "integrating" (Kintsch 1988), "consolidating" (Hansson 1999), or "connecting" (Storkel 2009) the newly learned concept with prior knowledge, which might include inference, belief revision, or reorganizing existing cognitive schemata.…”