Pseudohomophones and word recognition

Pseudohomophones play an important role in visual word recognition research, but they are not often themselves the object of experimental inquiry. In Experiment 1, we explored whether the status of body rime relations in pseudohomophones-whether their body rime relations exist in actual words-predicts the likelihood of word pronunciations to pseudohomophone spellings. In Experiment 2, we tested whether extant body rime relations modulate performance to pseudohomophones, and their context effect on word trials, in a lexical decision task. Extant body rime relations increase the likelihood that a pseudohomophone will be given a word pronunciation, and they produce slower and more error prone performance to pseudohomophones and words in lexical decision.

When SOFA primes TOUCH: interdependence of spelling, sound, and meaning in "semantically mediated" phonological priming

Three experiments test for semantically mediated priming of a word's phonology (e.g., sofa, an associate of couch, primes naming performance to touch). In the first two experiments, words that were body-rime-inconsistent (compare touch to couch) were used as naming targets. In the third experiment, words that were body-rime-consistent were also used (i.e., sofa primed pouch). Low-frequency inconsistent words yield a high rate of pronunciation errors when they were primed by indirectly related words, such as sofa, in both a standard naming task and a speeded naming task. High-frequency inconsistent words yielded slower naming times when they were primed by indirectly related words in a speeded naming task, but consistent words showed no significant effects of the primes. The results suggest that the relationship between semantics and phonology plays an important, early role in word perception.

Reading homographs: orthographic, phonologic, and semantic dynamics

Reading processes were compared across 3 word types: homographs (separate pronunciations and meanings, such as lead), homonyms (singular pronunciations but separate meanings, such as spring), and control words (e.g., clock). In Experiment 1, naming reaction times were significantly slower to homographs than to all other words. Experiments 2 and 3 used an association judgment task, with referent words related to the dominant or subordinate meanings of homonyms and homographs. In Experiment 2, homonyms and homographs were presented 1st, followed by disambiguating associates. In Experiment 3, presentation order was reversed. For homographs, performance costs always occurred for subordinate meanings. For homonyms, these costs vanished when context was provided by the preceding associates. The data underscore the priority of phonologic information in word meaning access and suggest that low- and high-level constraints combine to shape word perception.

Reading homographs: orthographic, phonologic, and semantic dynamics

Reading processes were compared across 3 word types: homographs (separate pronunciations and meanings, such as lead), homonyms (singular pronunciations but separate meanings, such as spring), and control words (e.g., clock). In Experiment 1, naming reaction times were significantly slower to homographs than to all other words. Experiments 2 and 3 used an association judgment task, with referent words related to the dominant or subordinate meanings of homonyms and homographs. In Experiment 2, homonyms and homographs were presented 1st, followed by disambiguating associates. In Experiment 3, presentation order was reversed. For homographs, performance costs always occurred for subordinate meanings. For homonyms, these costs vanished when context was provided by the preceding associates. The data underscore the priority of phonologic information in word meaning access and suggest that low- and high-level constraints combine to shape word perception.

Pathway selection's utility for control of word recognition

Regularity effects signify assembled phonology: the nonlexical pathway in word recognition. Assembled phonology may be strategically controlled. When task conditions do not favor its use, such as when a lexical decision task includes pseudohomophone (durt) foils, an alternative pathway is selected. Consequently, regularity effects will be reduced or absent. This prediction was tested in 3 lexical decision experiments using several definitions of word regularity. Word regularity was crossed in factorial designs with nonword lexicality, the degree to which nonword foils mimic the properties of actual words. Regularity effects to words were large and most reliable in pseudohomophone conditions, opposite to the tested prediction. Instead, the observed pattern corroborated a resonance account with parametric control.

Phonology can help or hurt the perception of print

Phonological manipulations affect performance in a letter search task that requires only a shallow level of processing. In Experiment 1, phonology reduced accuracy in the letter search task when a pseudohomophone (GAIM) contained a target letter ("i") that was missing in the spelling of its (nonpresented) sound-alike base word (GAME). In Experiment 2, phonology increased accuracy in the letter search task when the target letter was present in both the spelling of the pseudohomophone and the spelling of its sound-alike base word ("m" in GAIM and GAME). In Experiment 3, we showed that the phonology-hurts effect of Experiment 1 is not peculiar to nonword letter strings but generalizes to familiar words. In Experiment 4, we obtained a phonology-hurts effect on correct response times when stimuli were visible until participants responded (stimuli were not masked).

Phonology can help or hurt the perception of print

Phonological manipulations affect performance in a letter search task that requires only a shallow level of processing. In Experiment 1, phonology reduced accuracy in the letter search task when a pseudohomophone (GAIM) contained a target letter ("i") that was missing in the spelling of its (nonpresented) sound-alike base word (GAME). In Experiment 2, phonology increased accuracy in the letter search task when the target letter was present in both the spelling of the pseudohomophone and the spelling of its sound-alike base word ("m" in GAIM and GAME). In Experiment 3, we showed that the phonology-hurts effect of Experiment 1 is not peculiar to nonword letter strings but generalizes to familiar words. In Experiment 4, we obtained a phonology-hurts effect on correct response times when stimuli were visible until participants responded (stimuli were not masked).

Building a resonance framework for word recognition using design and system principles

Basic elements of a principle-based approach to model development are presented. The approach is needed to understand the operation of models capable of complex behavior. The use of principles facilitates both assignment of explanatory credit and blame when testing models and guides refinement of models when they fail. Two types of principles are distinguished: design and system. Design principles relate model behavior to observable human behavior. System principles relate model behavior to assumptions about a model's formal structure (architectural axioms). The use of, and relationship between, such principles is illustrated by building the theoretical framework of resonance (S. Grossberg & G. O. Stone, 1986) through the successive addition of principles.

Interdependence of form and function in cognitive systems explains perception of printed words

Perception is described within a complex systems framework that includes several constructs: resonance, attractors, subsymbols, and design principles. This framework was anticipated in J. J. Gibson's ecological approach (M. T. Turvey & C. Carello, 1981), but it is extended to cognitive phenomena by assuming experiential realism instead of ecological realism. The framework is applied in this article to explain phonologic mediation in reading and a complex array of published naming and lexical decision data. The full account requires only two design principles: covariant learning and self-consistency. Nonetheless, it organizes and explains a vast empirical literature on printed word perception.

Building a resonance framework for word recognition using design and system principles

Basic elements of a principle-based approach to model development are presented. The approach is needed to understand the operation of models capable of complex behavior. The use of principles facilitates both assignment of explanatory credit and blame when testing models and guides refinement of models when they fail. Two types of principles are distinguished: design and system. Design principles relate model behavior to observable human behavior. System principles relate model behavior to assumptions about a model's formal structure (architectural axioms). The use of, and relationship between, such principles is illustrated by building the theoretical framework of resonance (S. Grossberg & G. O. Stone, 1986) through the successive addition of principles.

Strategic control of processing in word recognition

Strategic control of word recognition in a lexical decision task was examined by manipulating the similarity of nonword foils to real words (nonword lexicality). Overall correct reaction times to words and the advantage of high- over low-frequency words were greater when nonword foils were more wordlike. This was true for both illegal (BTESE) versus legal (DEEST) nonword foils and legal nonword versus pseudohomophone (BEEST) foils. The same pattern of results was replicated in a 2nd experiment in which the word targets were always irregular (e.g., HAVE). A 3rd experiment demonstrated a large frequency blocking effect for low-frequency words, given pseudohomophone foils. The results are applied to pathway selection and random-walk frame-works. For both framework, canonical models are developed, which characterize qualitative predictions of broad classes of models within that framework. We argue for a pluralistic approach to theory development that moves from lower to higher order isomorphisms between data and theory.

Strategic control of processing in word recognition

Strategic control of word recognition in a lexical decision task was examined by manipulating the similarity of nonword foils to real words (nonword lexicality). Overall correct reaction times to words and the advantage of high- over low-frequency words were greater when nonword foils were more wordlike. This was true for both illegal (BTESE) versus legal (DEEST) nonword foils and legal nonword versus pseudohomophone (BEEST) foils. The same pattern of results was replicated in a 2nd experiment in which the word targets were always irregular (e.g., HAVE). A 3rd experiment demonstrated a large frequency blocking effect for low-frequency words, given pseudohomophone foils. The results are applied to pathway selection and random-walk frame-works. For both framework, canonical models are developed, which characterize qualitative predictions of broad classes of models within that framework. We argue for a pluralistic approach to theory development that moves from lower to higher order isomorphisms between data and theory.

Phonological processing skills and deficits in adult dyslexics

This article presents 4 experiments aimed at defining the primary underlying phonological processing deficit(s) in adult dyslexia. 5 phonological processes, all involving spoken language, were studied: phoneme perception, phoneme awareness, lexical retrieval of phonology, articulatory speed, and phonetic coding in verbal short-term memory. 2 differently ascertained adult dyslexic groups, familial dyslexics (n = 15) and clinic dyslexics (n = 15), were the subjects in each experiment. These dyslexic groups were chosen because deficits that persist until adulthood and that are found in differently ascertained dyslexic groups are more likely to be primary. Each dyslexic group was compared to 2 control groups, chronological age (CA) controls who were similar in age and sex, and younger reading age (RA) controls who were similar in reading age and sex. The main finding was a clear deficit in phoneme awareness in both dyslexic groups, with each dyslexic group performing significantly worse than both CA and RA controls. Moreover, performance on the 2 phoneme awareness tasks together uniquely accounted for substantial variance in nonword reading. The clinic but not the familial dyslexics appeared to have an additional deficit in verbal short-term memory. No clear deficits were found in either dyslexic group in phoneme perception, lexical retrieval, or articulatory speed.

Word identification in reading and the promise of subsymbolic psycholinguistics

The vast literature concerning printed word identification either contradicts or provides ambiguous support for each of the central hypotheses of dual-process theory, the most widely accepted theory of printed word identification. In contrast, clear, positive support exists for an alternative subsymbolic approach that includes a central role for the process of phonologic coding. This subsymbolic account is developed around a covariant learning hypothesis, derived from a design principle common to current learning algorithms within the subsymbolic paradigm. Where this hypothesis applies, and it may apply broadly, it predicts a common empirical profile of development.

Are words represented by nodes?

Semantic priming in a lexical decision task was investigated with concurrent pretarget and posttarget primes. The posttarget prime also served as a pattern mask of the lexical decision target. Forward priming is defined as the presence of a semantically related pretarget prime and an unrelated posttarget prime. Backward priming is defined as the presence of a semantically related posttarget prime and an unrelated pretarget prime. Forward and backward priming were compared both when the nonword foils were "legal" and when they were "illegal" with respect to English orthography. Predictions were derived for two general approaches to word recognition: spreading-activation and expectancy-set theories. Both approaches assume that word representations occupy distinct, nonoverlapping locations in memory. Backward-prime facilitation was equivalent to forward-prime facilitation when nonword foils were illegal; however, backward-prime facilitation was not significant when nonword foils were legal. These results challenge both approaches. The proposed solution uses semantic-space (distributed) representations that are functionally unitized by a resonant matching (verification) process.

Word identification in reading proceeds from spelling to sound to meaning

Van Orden (1987) reported that false positive errors in a categorization task are elevated for homophonic foils (e.g., HARE for A PART OF THE HUMAN BODY). Two new experiments replicate this finding and extend it to nonword homophone foils (e.g., SUTE FOR AN ARTICLE OF CLOTHING). False positive errors to nonword homophone foils substantially exceed false positive errors to nonhomophonic nonword spelling controls, showing that the phonological characteristics of the nonword foils are critical. Because nonwords are not represented in the lexicon, this new result implicates computed phonological codes as a source of the categorization errors. Additionally, in each of two experiments, matched word and nonword homophones produced virtually identical error rates. If stimulus nonword homophones are viewed as extremely unfamiliar words, compared with the relatively familiar stimulus word homophones, then our failure to observe an effect of stimulus familiarity strengthens the case that phonological coding plays a role in the identification of all printed words. The fact that the results are obtained in a categorization task that requires reading for meaning (rather than a lexical decision task) makes it difficult to avoid the conclusion that phonological mediation plays a role in normal reading of text for meaning.

Is phonology bypassed in normal or dyslexic development?

A pervasive assumption in most accounts of normal reading and spelling development is that phonological coding is important early in development but is subsequently superseded by faster, orthographic coding which bypasses phonology. We call this assumption, which derives from dual process theory, the developmental bypass hypothesis. The present study tests four specific predictions of the developmental bypass hypothesis by comparing dyslexics and nondyslexics from the same families in a cross-sectional design. The four predictions are: 1) That phonological coding skill develops early in normal readers and soon reaches asymptote, whereas orthographic coding skill has a protracted course of development; 2) that the correlation of adult reading or spelling performance with phonological coding skill is considerably less than the correlation with orthographic coding skill; 3) that dyslexics who are mainly deficient in phonological coding skill should be able to bypass this deficit and eventually close the gap in reading and spelling performance; and 4) that the greatest differences between dyslexics and developmental controls on measures of phonological coding skill should be observed early rather than late in development.None of the four predictions of the developmental bypass hypothesis were upheld. Phonological coding skill continued to develop in nondyslexics until adulthood. It accounted for a substantial (32-53 percent) portion of the variance in reading and spelling performance in adult nondyslexics, whereas orthographic coding skill did not account for a statistically reliable portion of this variance. The dyslexics differed little across age in phonological coding skill, but made linear progress in orthographic coding skill, surpassing spelling-age (SA) controls by adulthood. Nonetheless, they didnot close the gap in reading and spelling performance. Finally, dyslexics were significantly worse than SA (and Reading Age [RA]) controls in phonological coding skill only in adulthood.