Semantic priming in visual word recognition: Activation blocking and domains of processing

The fact that letter search on a prime eliminates the typically robust semantic priming effect in lexical decision is often attributed to the “shallowness” of the prime-processing task. In three experiments we investigated this claim by using two different “shallow” prime-processing tasks: letter search and color identification. Consistent with previous reports, lexical decisions to semantically related targets were not facilitated when subjects searched the prime for a probe letter. In contrast, semantic priming was observed following a color discrimination task on the prime. We suggest that a levels-of-processing interpretation is not an adequate framework for understanding these data. Instead, a domain-specific processing account is offered in which explicit processing at the letter level (as in letter search) makes demands on resources (e.g., activation) that drives processing at the semantic level. This competition is resolved by establishing a temporary activation block at the lexical-semantic interface, which results in the elimination or attenuation of semantic priming. In contrast, global judgment of color is viewed as a domain that does not make demands on the resources that drive the visual word recognition machinery. There is therefore no need for an activation block, and semantic priming is not prevented.     

Associative and phonological priming effects after letter search on the prime

Responses to target words typically are faster and more accurate after associatively related primes (e.g., "orange-juice") than after unrelated primes (e.g., "glue-juice"). This priming effect has been used as an index of semantic activation, and its elimination often is cited as evidence against semantic access. When participants are asked to perform a letter search on the prime, associative priming typically is eliminated, but repetition and morphological priming remain. It is possible that priming survives letter search when it arises from activity in codes that are represented before semantics. This experiment examined associative and phonological priming to determine whether priming from phonologically related rhymes would remain after letter search (e.g., "moose-juice"; rhyming items were orthographically dissimilar). When participants read the primes, equivalent associative and phonological priming effects were obtained; both effects were eliminated after letter search. The impact of letter search on semantic and phonological access and implications for the structural arrangement of lexical and semantic memory are discussed.     

Semantic activation and letter search: Blocking or suppression?

Two experiments investigated associative priming with a letter-search prime task where either the prime and letter probe were presented simultaneously, or the letter probe appeared 200 ms (Experiment 1) or 300 ms (Experiment 2) after the prime. Weak associative priming was observed in both experiments, but unlike Stolz and Besner (1996) we found no evidence that priming was increased when the probe was delayed. However, strong associative priming was observed when a semantic decision had to be made on the prime (Experiment 3). Our results are consistent with an account where semantic activation of the prime occurs but its action on the target is suppressed by the prime task. The persistence of weak priming effects with the letter search task is explained in terms of the low-frequency items used.     

Congruency effects in the letter search task: Semantic activation in the absence of priming

Semantic priming is typically eliminated when participants perform a letter search on the prime, suggesting that semantic activation is conditional upon one's attentional goals. However, in such studies, semantic activation (or the lack thereof) is not measured during the letter search task itself but, instead, is inferred on the basis of the responses given to a later target. In the present study, direct online evidence for semanticactivation was tested using words whose meaning should bias either a positive or a negative response (e.g.,present vs. absent). In Experiment 1, a semantic congruency effect was obtained, with faster responses when the word meaning matched the required response. Experiment 2 replicated the congruency effect while, simultaneously, showing the elimination of semantic priming. It is concluded that letter search does not affect the initiation of semantic activation. Possible accounts for the elimination of priming following letter search include activation-based suppression and transfer-inappropriate processing.     

On the interaction between linguistic and pictorial systems in the absence of semantic mediation: Evidence from a priming paradigm

Use of the letter search task in the context of the priming paradigm has proved to be an invaluable tool for the investigation of the strategic control of processes involved in word recognition. In particular, previous findings that letter search on a prime word interferes with the priming of semantically/associatively related targets, but not with the priming of either identical or morphologically related targets, suggests that letter search may selectively interfere with semantic processing, leaving other levels of processing intact. In the present experiments, this investigation was extended by exploring the priming of pictures following letter search of either a same-concept word (repetition priming) or a semantically/associatively related word (semantic priming). There was significant repetition priming of picture categorization following both silent reading and letter search of the prime word (Experiments 1 and 2). In contrast, semantic priming of pictures was found only following silent reading of the prime; there was no semantic priming following letter search of the prime (Experiment 2). This pattern of results suggests that focusing attention at the letter level during prime processing selectively attenuates activation of the semantic system by the prime. It does not prevent the spread of activation between the lexical and pictogen levels of representation of a given concept.     

Automatic semantic activation is no myth: semantic context effects on the N400 in the letter-search task in the absence of response time effects

The claim that semantic activation is an automatic process was recently called a myth, on the basis of the finding that if letter search is performed on a prime word, semantic priming effects on response time are eliminated, whereas repetition effects are preserved. The absence of semantic activation, however, cannot be validly inferred from the lack of response time effects, and converging evidence is needed. To this end, we examined the event-related potential correlate of priming, the N400 amplitude modulation, in a letter-search priming paradigm. Our experiment replicated the response time effects and demonstrated that the N400 amplitude successfully differentiates cross-case repetition priming, semantic priming, and neutral conditions. The results clearly indicate that the meaning of the prime word was processed and that semantic activation indeed was present in the letter-search task. The notion that semantic activation is an automatic process should not be abandoned prematurely.     

Word-based grouping affects the prime-task effect on semantic priming

Semantic priming between words is reduced or eliminated if a low-level task such as letter search is performed on the prime word (the prime task effect), a finding used to question the automaticity of semantic processing of words. This idea is critically examined in 3 experiments with a new design that allows the search target to occur both inside and outside the prime word. The new design produces the prime task effect (Experiment 1) but shows semantic negative priming when the target letter occurs outside the prime word (Experiments 2 and 3). It is proposed that semantic activation and priming are dissociable and that inhibition and word-based grouping are responsible for reduction of semantic priming in the prime task effect.     

Modulating semantic feedback in visual word recognition

According to the interactive activation framework proposed by McClelland and Rumelhart (1981), activation spreads both forward and backward between some levels of representation during visual word recognition. An important boundary condition, however, is that the spread of activation from lower to higher levels can be prevented (e.g., explicit letter processing during prime processing eliminates the well-documented semantic priming effect). Can the spread of activation from higher to lower levels also be prevented? This question was addressed with a choice task procedure in which subjects read a prime word and then responded to a target, performing either lexical decision or letter search depending on the color of the target. A semantic context effect was observed in lexical decision, providing evidence of semantic-level activation. In contrast, there was no semantic context effect in the letter search task, despite evidence of lexical involvement: Words were searched faster than nonwords. Further evidence of lexical involvement in the letter search task appeared in Experiment 2 in the form of greater identity priming for words than for nonwords. The results of these experiments are consistent with the conclusion that feedback from the semantic level to the lexical level can be blocked. Hence, between-level activation blocks can be instantiated in both bottom-up and top-down directions.     

Basic processes in reading: can functional phonological recoding be blocked?

A widely held view is that phonological processing is always involved in lexical access from print, and is automatic in that it cannot be prevented. This claim was assessed in the context of a priming paradigm. In Experiment 1, repetition priming was observed for both pseudohomophone-word pairs (e.g., brane-brain) and morphologically related word pairs (e.g., marked-mark) in the context of lexical decision. In Experiment 2, subjects searched the prime for the presence of a target letter and then made a lexical decision to a subsequent letter string. Phonological priming from a pseudohomophone was eliminated following letter search of the prime, whereas morphological priming persisted. These results are inconsistent with the claim that a) lexical access from print requires preliminary phonological processing, and b) functional phonological processing cannot be blocked. They are, however, consistent with the conclusion that, for intact skilled readers, lexical access can be accomplished on the basis of orthographic processing alone. These results join a growing body of evidence supporting the claim that there exist numerous points in visual word recognition at which processing can be stopped.     

Levels and speed of processing effects on word analysis

Previous work on semantic priming effects has suggested that priming is contingent upon semantic analysis of the prime stimulus. In the present series of experiments, subjects in Grade 3, Grade 6, and college performed variants of a priming task in which semantic (lexical decision) as well as fast (case judgment) and slow (letter search) nonsemantic levels of processing were required for both prime and target stimuli. Contrary to a levels of processing hypothesis, context effects were not simple functions of the level at which the prime was processed. Priming effects were found with a slow nonsemantic prime task (letter search), but generally not with a fast nonsemantic analysis of the prime (case judgment). The form of the priming effect when the prime was processed semantically by a lexical decision depended on the relationship between prime and target processing: a switch from semantic prime processing to nonsemantic target processing produced Stroop-like interference, with other combinations of prime and target processing producing facilitation. By using a series of discrimination, focusing, and classification tasks in a Garner (1978) paradigm, it was possible to determine how subjects were processing the semantic and nonsemantic dimensions, and these perceptual strategies were compared across educational levels to account for the priming effects. Our results suggest that context effects need to be understood in terms of the speeds of processing of different codes of information inherent in words.     

Scene context influences without scene gist: Eye movements guided by spatial associations in visual search

Although the use of semantic information about the world seems ubiquitous in every task we perform, it is not clear whether we rely on a scene’s semantic information to guide attention when searching for something in a specific scene context (e.g., keys in one’s living room). To address this question, we compared contribution of a scene’s semantic information (i.e., scene gist) versus learned spatial associations between objects and context. Using the flash-preview–moving-window paradigm Castelhano and Henderson (Journal of Experimental Psychology: Human Perception and Performance 33:753–763, 2007), participants searched for target objects that were placed in either consistent or inconsistent locations and were semantically consistent or inconsistent with the scene gist. The results showed that learned spatial associations were used to guide search even in inconsistent contexts, providing evidence that scene context can affect search performance without consistent scene gist information. We discuss the results in terms of hierarchical organization of top-down influences of scene context.     

Attention constraints of semantic activation during visual word recognition

The reduction of semantic priming following letter search of the prime suggests that semantic activation can be blocked if attention is allocated to the letter level during word processing. Is this true even for the very fast-acting component of semantic activation? To test this, the authors explored semantic priming of lexical decision at stimulus onset asynchronies (SOAs) of either 200 or 1,000 ms. Following semantic prime processing, priming occurred at both SOAs. In contrast, no priming occurred at the long SOA following letter-level processing. Of greatest interest, at the short SOA there was priming following the less demanding consonant/vowel task but not following the more attention-demanding letter search task. Hence, semantic activation can occur even when attention is directed to the letter level, provided there are sufficient resources to support this activation. The authors conclude that the default setting during word recognition is for fast-acting activation of the semantic system.     

‘I’

It has traditionally been maintained that every token of ‘I’ refers to its utterer. However, certain uses of indexicals conflict with this claim, and its counterparts with respect to ‘here’ and ‘now’, suggesting that the traditional account of indexical reference should be abandoned. In this paper, I examine some proposed alternatives and the difficulties they face, before offering a new account of indexical reference. I endorse Kaplan’s view that the reference of an indexical is determined on any occasion it is used by applying its character to a particular context, arguing that the problem cases show that this is not always the context of utterance. The task facing the semantic theorist is thus to explain what fixes the reference-determining context. I consider and reject both Predelli’s suggestion that the reference-determining context is the one intended by the utterer, and Corazza et al.’s proposal that the relevant context is fixed by conventions delivered by the utterance setting. The discussion of these two accounts reveals that an adequate theory of indexical reference should allow the speaker to use indexicals in novel ways, whilst holding that what a speaker can refer to with an indexical utterance is constrained by what an audience can understand. I develop an account based around these two requirements.     

Letter search does not affect semantic priming in a probe naming task

The normally robust semantic priming effect observed in lexical decision is usually reduced to the point of being absent, when a letter search has to be performed on the prime. It has been argued that semantic activation is thus not an automatic process but rather cognitively controlled and therefore adaptable to task demands. We examined the effects of letter search priming on pronunciation times and found a reliable semantic priming effect, following letter search that was not affected at all relative to a standard condition, where participants silently read the prime. Thus the nature of the prime task did not seem to affect the processing mode employed, semantic access occurred even though attention was focused on surface properties of the prime.     

Semantic priming in the prime task effect: Evidence of automatic semantic processing of distractors

The automaticity of the semantic processing of words has been questioned because of the reduction of semantic priming when the prime word is processed nonsemantically--for example, in letter search (the prime task effect). In two experiments, prime distractor words produced semantic priming in a subsequent lexical decision task, but with the direction of priming (positive or negative) depending on the prime task. Lexico-semantic tasks produced negative semantic priming, whereas letter search produced positive semantic priming. These results are discussed in terms of task-based inhibition. We argue that, given the results from the distractors, the absence of semantic priming does not indicate an absence of semantic activation but reflects the action of control processes on prepotent responses when less practiced responses are needed.     

Coffa’s Kant and the evolution of accounts of mathematical necessity

According to Alberto Coffa in The Semantic Tradition from Kant to Carnap, Kant’s account of mathematical judgment is built on a ‘semantic swamp’. Kant’s primitive semantics led him to appeal to pure intuition in an attempt to explain mathematical necessity. The appeal to pure intuition was, on Coffa’s line, a blunder from which philosophy was forced to spend the next 150 years trying to recover. This dismal assessment of Kant’s contributions to the evolution of accounts of mathematical necessity is fundamentally backward-looking. Coffa’s account of how semantic theories of the a priori evolved out of Kant’s doctrine of pure intuition rightly emphasizes those developments, both scientific and philosophical, that collectively served to undermine the plausibility of Kant’s account. What is missing from Coffa’s story, apart from any recognition of Kant’s semantic innovations, is an attempt to appreciate Kant’s philosophical context and the distinctive perspective from which Kant viewed issues in the philosophy of mathematics. When Kant’s perspective and context are brought out, he can not only be seen to have made a genuinely progressive contribution to the development of accounts of mathematical necessity, but also to be relevant to contemporary issues in the philosophy of mathematics in underappreciated ways.     

When are morphemic and semantic priming observed in visual word recognition?

J. Stolz and D. Besner (1998) reported a dissociation between morphemic and semantic priming in the context of lexical decision. Morphemic priming was observed following letter search on the prime display, but semantic priming was not. Participants in the present experiment identified the colour of a single letter in the prime display before making a lexical decision to the target. Both morphemic and semantic priming were observed. These results are discussed in relation to the observation that identifying the colour of a single letter of a word in the Stroop task is associated with a reduction in the size of the Stroop effect as compared to when all letters are coloured.     

Controlling lexical contributions to the reading of pseudohomophones

Reynolds and Besner (2005) presented a computational account of six effects that emerge when readers are asked to pronounce pseudohomophones (nonwords—e.g., brane—that sound like words when pronounced). In the dual route cascaded (DRC) model of reading, they varied a parameter controlling the rate of inhibition from letter units to the orthographic lexicon to mimic strategic control over the extent of specific lexical processing. In this article, we provide an account in which the effects are simulated by varying the DRC’s reading-aloud criterion—a parameter that sets the minimal level of phonemic activation required to pronounce a letter string. We show that varying this parameter provides another means of controlling lexical contributions to reading aloud.     

Reading aloud: New evidence for contextual control over the breadth of lexical activation

Computational accounts of reading aloud largely ignore context when stipulating how processing unfolds. One exception to this state of affairs proposes adjusting the breadth of lexical knowledge in such models in response to differing contexts. Three experiments and corresponding simulations, using Coltheart, Rastle, Perry, Langdon, and Ziegler’s (2001) dual-route cascaded model, are reported. This work investigates a determinant of when a pseudohomophone such as brane is affected by the frequency of the word from which it is derived (e.g., the base word frequency of brain) by examining performance under conditions where it is read aloud faster than a nonword control such as frane. Reynolds and Besner’s (2005a) lexical breadth account makes the novel prediction that when a pseudohomophone advantage is seen, there will also be a base word frequency effect, provided exception words are also present. This prediction was confirmed. Five other accounts of this pattern of results are considered and found wanting. It is concluded that the lexical breadth account provides the most parsimonious account to date of these and related findings.     

On Mill,Infallibility, and Freedom of Expression

Philosophers have tended to dismiss John Stuart Mill’s claim that ‘all silencing of discussion is an assumption of infallibility’. I argue that Mill’s ‘infallibility claim’ is indeed open to many objections, but that, contrary to the consensus, those objections fail to defeat the anti-authoritarian thesis which lies at its core. I then argue that Mill’s consequentialist case for the liberty of thought and discussion is likewise capable of withstanding some familiar objections. My purpose is to suggest that Mill’s anti-authoritarianism and his faith in thought and discussion, when taken seriously, supply the basis for a ‘public interest’ account of ‘freedom of expression as the liberty of thought and discussion’ which is faithful to Mill in spirit, if not to the precise letter. I outline such an account, which – as I say in conclusion – can serve as a valuable safeguard against ad hoc, reactive legislation, and the demands of a spurious communitarianism.