ReviewGetting real about Semantic Illusions: Rethinking the functional role of the P600 in language comprehension
Introduction
As a sentence or story unfolds in time, language users incrementally construct an interpretation of the linguistic input. Creating this interpretation draws on various different information sources such as syntactic, semantic, pragmatic, prosodic, and visual information, as well as world knowledge. Exactly when and how these different types of information are combined is a matter of debate. Many theories of language comprehension claim that there is a tight coupling between syntactic and semantic processing (e.g., Frazier, 1987, MacDonald et al., 1994, McRae et al., 1998, van Gompel et al., 2000). According to these theories, it would be impossible to construct an interpretation of language input without immediately taking syntactic information into account. This assumption has been challenged by recent evidence from Event Related brain Potentials (ERPs). Hoeks et al. (2004) for example, studied Dutch sentences in which two plausible verb arguments appeared in a semantically anomalous order, as in ‘De speer heeft de atleten geworpen’ (lit: The javelin has the athletes thrown). Relative to a control sentence ‘De speer werd door de atleten geworpen’ (lit: The javelin was by the athletes thrown), no shift in N400 amplitude (a negative deflection of the ERP signal peaking at about 400 ms after the onset of a critical stimulus) was found. This was surprising because the amplitude of the N400 has been associated with difficulty in semantic integration (see Kutas and Federmeier, 2011, for an overview). Instead, Hoeks et al. (2004) found that the sentence-final verb ‘thrown’ produced a P600-effect (a positive deflection of the ERP signal that reaches maximum around 600 ms post stimulus onset) relative to control. Again, this was unexpected since P600 amplitude has been linked with syntactic revision (see Gouvea et al., 2010, for an overview) but the test items were perfectly grammatical. This phenomenon in which a semantically anomalous, syntactically well-formed sentence elicits a P600-effect, but no N400-effect, has been called a ‘Semantic Illusion’.1 This is because the absence of an N400-effect suggested that participants were temporarily under the illusion that these sentences made sense. The presence of a P600-effect, on the other hand, indicated that participants eventually realized that their interpretations were infelicitous, and that they were trying to resolve this conflict through ‘effortful syntactic processing’ (Hoeks et al., 2004, p. 71).
To account for Semantic Illusions, so-called multi-stream models have been proposed in which a separate semantic analyzer can put forward an interpretation of a sentence that may not be in line with its surface structure (Bornkessel-Schlesewsky and Schlesewsky, 2008, Hagoort et al., 2009, Kim and Osterhout, 2005, Kolk et al., 2003, Kos et al., 2010, Kuperberg, 2007, van Herten et al., 2005, van Herten et al., 2006). In this paper, we critically review five multi-stream architectures and ask whether they can account for the available data on the Semantic Illusion. The review suggests that none of them can explain the full range of relevant results. We argue that this is the case because all of them adopt the view that the amplitude of the N400 indexes the relative difficulty of integrating the meaning of an incoming word into a partial interpretation of a sentence. In contrast, we show that most evidence is consistent with an alternative interpretation of N400-effects in which N400 amplitude reflects the retrieval of the meaning of a word from long-term memory (Federmeier and Laszlo, 2009, Kutas and Federmeier, 2000, Kutas and Federmeier, 2011, Lau et al., 2008, Lau et al., 2009, van Berkum, 2009, van Berkum, 2010). On this view, the absence of an N400-effect in Semantic Illusion sentences follows directly from the workings of the memory system. Crucially, the processes reflected in N400 amplitude are thus not integrative or compositional in nature. We suggest that the process of integration, in which a mental representation of what is being communicated is updated with new information, is reflected in the amplitude of the P600 instead.
Combining this novel view on P600 amplitude with the retrieval perspective on N400 amplitude leads to a parsimonious single-stream account that can explain the full range of Semantic Illusion data. On this account, sentences like ‘De speer heeft de atleten geworpen’ (lit: The javelin has the athletes thrown) do not give rise to a Semantic Illusion. Language users do not perceive such sentences as if they made sense, not even temporarily. There is no N400-effect because participants do not have problems activating the meaning of the anomalous word. A P600-effect is observed instead because the situation or event that is described is difficult to interpret.
Section snippets
Explaining the Semantic Illusion phenomenon
The results of Hoeks et al. (2004) (as well as similar results of Kolk et al., 2003, Kuperberg et al., 2003) raised two critical questions: 1) Why did the thematic violation brought about by the critical verb fail to give rise to an N400-effect effect? and 2) Why did the critical verb produce a P600-effect although these sentences were syntactically well-formed?
According to Hoeks et al. (2004), the absence of an N400-effect suggested that participants had been tricked into some kind of a
Rethinking the functional role of the N400 and the P600
The Semantic Illusion phenomenon suggested that semantic information can to some extent be processed separately and autonomously, causing a shift from single-stream to multi-stream models. What we would like to suggest, however, is that instead of abandoning single-stream models, we should reconsider the functional interpretation of the ERP components that are involved. Following a growing number of studies, we propose to interpret N400 amplitude as reflecting a memory retrieval phase, in which
Semantic Illusions revisited
The ‘Semantic Illusion’ phenomenon has led to a paradigmatic shift from single-stream towards multi-stream models, as it seemed to suggest the existence of an autonomous semantic processing stream. However, from the perspective of the single-stream Retrieval-Integration account outlined above, there is no such thing as a ‘Semantic Illusion’ in any of the SIE studies reported in this paper. Observing a P600-effect instead of an N400-effect in response to sentences such as ‘De speer heeft de
Predictions and future work
On the basis of Retrieval-Integration account, one can make the following predictions:
Conclusion
We have proposed a simple and parsimonious single-stream account of language processing, the Retrieval-Integration account. On this account, the N400 and the P600 component reflect two successive processing stages, in which the output of the retrieval phase (N400 amplitude) serves as input for integration (P600 amplitude). The N400 component thus reflects a retrieval stage in which the syntactic properties and semantic features of a current word are ‘retrieved’ from long-term memory; the N400
Acknowledgments
This research was funded by the Netherlands Organization for Scientific Research (NWO) PGW grant 10-26 (HB, HF, and JH), and VENI grant 275-89-008 (HF). We would like to thank Gina Kuperberg and Miriam Kos for providing us with the materials of their experiments.
References (132)
- et al.
Pitch accents in context: how listeners process accentuation in referential communication
Neuropsychologia
(2011) - et al.
An alternative perspective on semantic P600 effects in language comprehension
Brain Res. Rev.
(2008) Inferential bridging relations reveal distinct neural mechanisms: evidence from event-related brain potentials
Brain Lang.
(2006)- et al.
Accessing world knowledge: evidence from N400 and reaction time priming
Cognitive Brain Res.
(2005) An alternative view of the mental lexicon
Trends Cogn. Sci.
(2004)- et al.
From words to meaning: a semantic illusion
J. Verb. Learn. Verb. Behav.
(1981) - et al.
A rose by any other name: long-term memory structure and sentence processing
J. Mem. Lang.
(1999) - et al.
Time for meaning: electrophysiology provides insights into the dynamics of representation and processing in semantic memory
Psychol. Learn. Motiv.
(2009) - et al.
Storage and integration in the processing of filler-gap dependencies: an ERP study of topicalization and wh-movement in german
Brain Lang.
(2003) - et al.
Separating syntactic memory costs and syntactic integration costs during parsing: the processing of German WH-questions
J. Mem. Lang.
(2002)
The P600 as an indicator of syntactic ambiguity
Cognition
N400-like magnetoencephalography responses modulated by semantic context, word frequency, and lexical class in sentences
NeuroImage
Seeing words in context: the interaction of lexical and sentence level information during reading
Cognitive Brain Res.
The spatial and temporal signatures of word production components
Cognition
Electrophysiological evidence for serial sentence processing: a comparison between non-preferred and ungrammatical continuations
Cognitive Brain Res.
Sentence processing in the visual and auditory modality: do comma and prosodic break have parallel functions?
Brain Res.
The independence of combinatory semantic processing: evidence from event-related potentials
J. Mem. Lang.
Conflict and surrender during sentence processing: an ERP study of syntax–semantics interaction
Brain Lang.
Late positivities in unusual situations
Brain Lang.
Structure and limited capacity in verbal working memory: a study with event-related potentials
Brain Lang.
Event-related potentials, cognition, and behavior: a biological approach
Neurosci. Biobehav. Rev.
Neural mechanisms of language comprehension: challenges to syntax
Brain Res.
Electrophysiological distinctions in processing conceptual relationships within simple sentences
Cognitive Brain Res.
The role of animacy and thematic relationships in processing active English sentences: evidence from event-related potentials
Brain Lang.
Electrophysiology reveals semantic memory use in language comprehension
Trends Cogn. Sci.
Psycholinguistics electrified II: 1994–2005
A lexical basis for N400 context effects: evidence from MEG
Brain Lang.
The pragmatic meanings conveyed by function words in Chinese sentences: an ERP study
J. Neurolinguistics
Meaningful physical changes mediate lexical-semantic integration: top-down and form-based bottom-up information sources interact in the N400
Neuropsychologia
Semantics prevalence over syntax during sentence processing: a brain potential study of noun-adjective agreement in Spanish
Brain Res.
Modeling the influence of thematic fit (and other constraints) in on-line sentence comprehension
J. Mem. Lang.
Brain potentials and syntactic violations revisited: no evidence for specificity of the syntactic positive shift
Neuropsychologia
Testing the limits of the semantic illusion phenomenon: ERPs reveal temporary semantic change deafness in discourse comprehension
Cognitive Brain Res.
P600 related to rule violation in an arithmetic task
Cognitive Brain Res.
Event-related brain potentials elicited by syntactic anomaly
J. Mem. Lang.
ERP effects of the processing of syntactic long-distance dependencies
Cognitive Brain Res.
Brain reflections of words and their meaning
Trends Cogn. Sci.
Morphological analysis in sentence processing: an ERP study
Lang. Cognitive Proc.
The balance between memory and unification in semantics: a dynamic account of the N400
Lang. Cognitive Proc.
A case study of anomaly detection: shallow semantic processing and cohesion establishment
Mem. Cognit.
The interplay between prosody and syntax in sentence processing: the case of subject- and object-control verbs
J. Cogn. Neurosci.
Prosodic breaks in sentence processing investigated by event-related potentials
Lang. Ling. Compass
The role of prosodic breaks and pitch accents in grouping words during on-line sentence processing
J. Cogn. Neurosci.
The extended Argument Dependency Model: a neurocognitive approach to sentence comprehension across languages
Psychol. Rev.
Eliciting thematic reanalysis effects: the role of syntax-independent information during parsing
Lang. Cognitive Proc.
The processing nature of the N400: evidence from masked priming
J. Cogn. Neurosci.
On the electrophysiology of language comprehension: implications for the human language system
The P600 reflects cost of new information in discourse memory
Neuroreport
The N400 as a function of the level of processing
Psychophysiology
Cited by (383)
Anaphoric distance dependencies in visual narrative structure and processing
2024, Cognitive PsychologyThe effects of lexical representation on the dynamic process of phrase comprehension
2023, Journal of Neurolinguistics