Date of Degree


Document Type


Degree Name





Valerie L. Shafer

Committee Members

Sandeep Prasada

William McClure

Subject Categories

Cognition and Perception | Cognitive Neuroscience | Psycholinguistics and Neurolinguistics | Semantics and Pragmatics


N400, parietal occipital slow-wave, prepositions, spatial imagery, lexicon, Functional Geometry


This project used Event-Related Potentials (ERPs) to explore neurophysiological brain responses to prepositional phrases involving concrete and abstract reference nouns (e.g., "plate" and "moment", respectively) after the presentation of objects with varying spatial features. Prepositional phrases were headed by in or on and were either matching (e.g., "in the plate/moment") or mismatching (e.g., "on the plate/moment"). Conjunction phrase matches and fillers were also presented. Before half of the concrete-phrase items, a photographic depiction of the reference noun was presented. In these photographs, objects were displayed in a way that was either more appropriate for in or for on. Similarly, before half of the abstract-phrase trials, photographs of nonce objects with spatial features that were either more appropriate for in or for on were presented. For the remaining trials, either no picture was displayed, or a picture of a random object was displayed.

Results indicated that linguistic and visual context impacted ERPs to words in these phrases. Beginning with linguistic context, all prepositional phrases yielded negative slow-wave activity in parietal and occipital sites, while conjunction phrases did not. Because this negativity is modulated by processes involved in the generation and manipulation of spatial imagery, this finding indicates that a similar spatial-image-formation process is involved in the processing of both concrete and abstract prepositional phrases. There were differences between responses to concrete and abstract phrases as well. Mismatching concrete reference nouns yielded a relatively large centro-parietal N400 response, suggesting that these nouns were semantically unexpected. Mismatching abstract nouns, on the other hand, yielded a late, marginally significant positivity, showing that the presentation of these nouns required phrase reanalysis and/or reconstruction. The latter result casts doubt on accounts of polysemy claiming that abstract uses of prepositions are cognitively and metaphorically linked to their spatial senses.

Visual stimuli also impacted responses to the phrases. The type of object presented in the picture before the phrase impacted N400 responses to prepositions, where pictures of in objects yielded smaller N400 responses to in and vice versa for on, no matter the configuration of the object in the picture. This suggests that an object’s category – rather than its specific visual in a particular context– primes a preposition’s lexical denotation. The impact of object type was also observable downstream from the N400 to prepositions. Parieto-occipital slow-wave negativity increased after the presentation of random objects and after the presentation of no picture as compared to responses to phrases presented after pictures of in or on objects. This result implies increased reliance on internal image-formation processes to scaffold linguistic processing when external visual information does not facilitate phrasal interpretation and/or recollection.

While the configuration of the object presented in the picture before the phrase did not impact responses to prepositions, it did impact responses to concrete reference nouns. Pictures of objects in spatially mismatching configurations elicited frontal N400 effects, which are believed to index the amodal incorporation of image-mediated information into on-line semantic processing. Frontal N400s were also impacted by phrase type (match versus mismatch), where frontal N400s to matching nouns after spatially mismatching pictures dissipated earlier than for mismatching nouns, suggesting that processes involved in integrating visual context are completed more quickly (and perhaps less effortfully) when the noun is primed by semantic context than when it is not. This is similar to the response pattern for prepositions – when prepositions were unprimed by semantic (visual) context, there was increased effort involved in spatial-image formation.

Together, results reveal a multifaceted interaction between phrasal expectations and visual priming during the processing of natural spatial language about real-world objects and abstract concepts. More broadly, findings imply that the processing of all language, even simple phrases containing words that are believed to have limited semantic content, engages a complex neural network involving linguistic and non-linguistic representations.