How do we process novel conceptual combinations in context?

Abstract

Most theories of how people interpret novel noun–noun combinations (e.g., ghost forest) do not sufficiently address the role of context, which is surprising given that novel combinations are generally encountered in discourse contexts. We evaluated three hypotheses for the role of context in interpretation. The generation hypothesis states that interpretation is based on sense generation both in and out of context. The anaphor resolution hypothesis states that interpretation initially involves identifying a referent from the prior discourse context, and meaning generation occurs only if anaphoric processing fails. The dual-process hypothesis states that sense generation and anaphor resolution operate in parallel, and both may influence initial interpretation. In two experiments, we examined reading times on novel combinations when the dominant meaning (most likely interpretation) or subordinate meaning (infrequent interpretation) preceded or followed the combination in text. Results supported the dual-process hypothesis.

Keywords:

• Conceptual combination
• Concept
• Discourse
• Comprehension
• Anaphor resolution

Acknowledgments

Erica Middleton is now at Moss Rehabilitation Research Institute, Elkins Park, PA, USA

Notes

¹ One issue regards the kinds of novel combinations used in the current work. Whereas the sense generation process underlying the interpretation of some of our combinations probably involves a relation-linking process (e.g., flu liquid, liquid to cure the flu), the meanings generated for others could be described as resulting from a property mapping process (e.g., skunk mud, mud as stinky as a skunk; Costello & Keane, 2001; Wisniewski, 1997, 1998). Whether property mapping is distinct from relation linking is a topic of debate (Estes, 2003; Gagné, 2000; Tagalakis & Keane, 2006; Wisniewski & Love, 1998), and resolution of this issue is beyond the scope of this paper. For our purposes, sense generation need only involve an algorithmic process, which is able to provide an interpretation in the absence of relevant context. Whether this algorithmic process involves property mapping or relation linking, our conclusions are the same.

² Although technically the design is a 2 [dominant meaning vs. subordinate meaning] by 2 [context before vs. context after] design, we draw the predictions for reading times in reference to three functional conditions: Subordinate context comes before the combination (i.e., subordinate-context condition); dominant context comes before the combination (i.e., dominant-context condition); and relevant context appears after the combination (i.e., context-after condition). The context-after condition collapses across both situations where dominant context or subordinate context appear after the combination. Given that reading time in the region containing the novel combination is of interest, in these two situations, the preceding text is identical until after the novel combination is presented. Thus the reading times on the combination region for the dominant-context and subordinate-context conditions are collapsed in the analyses in both experiments.

^aExperiment 1 only.

^aExperiment 1 only.

^aExperiment 1 only.

^aExperiment 1 only.

^aExperiment 1 only.

^bExperiment 2 only.

^bExperiment 2 only.

^bExperiment 2 only.

^bExperiment 2 only.

^bExperiment 2 only.