Semantic transparency in free stems: The effect of Orthography-Semantics Consistency on word recognition

Abstract

A largely overlooked side effect in most studies of morphological priming is a consistent main effect of semantic transparency across priming conditions. That is, participants are faster at recognizing stems from transparent sets (e.g., farm) in comparison to stems from opaque sets (e.g., fruit), regardless of the preceding primes. This suggests that semantic transparency may also be consistently associated with some property of the stem word. We propose that this property might be traced back to the consistency, throughout the lexicon, between the orthographic form of a word and its meaning, here named Orthography-Semantics Consistency (OSC), and that an imbalance in OSC scores might explain the “stem transparency” effect. We exploited distributional semantic models to quantitatively characterize OSC, and tested its effect on visual word identification relying on large-scale data taken from the British Lexicon Project (BLP). Results indicated that (a) the “stem transparency” effect is solid and reliable, insofar as it holds in BLP lexical decision times (Experiment 1); (b) an imbalance in terms of OSC can account for it (Experiment 2); and (c) more generally, OSC explains variance in a large item sample from the BLP, proving to be an effective predictor in visual word access (Experiment 3).

Keywords:

• Orthography-Semantics Consistency
• Distributional semantic models
• Megastudies
• Visual word identification

Authors' contributions are as follows: Simona and Marco conceived the study and developed the OSC measure; Marco developed the semantic model and analysed the data, with assistance from Simona; Davide gathered the data from previous experiments; Davide and Marco ran the metanalysis; Marco drafted the paper, which was critically revised by all authors; Marco supervised the project.

We would like to thank Joanna Morris, Joseph Devlin, Kevin Diependaele, Mirjana Bozic, Nina Kazanina, Sally Andrews, and William Marslen-Wilson for having contributed to this work by sharing their data. We also thank Marco Baroni, Petar Milin, and one anonymous reviewer for their insightful comments on previous versions of the paper.

This research was partially supported by the ERC 2011 Starting Independent Research Grant n. 283554 (COMPOSES) and by a FIRB – Futuro in Ricerca Grant n. RBFR085K98 from the Italian Ministry of Education, University and Research.

Notes

¹Since the reference corpora were POS-tagged, in the resulting DSM we obtained separate vectors for homographs with different grammatical class (e.g., a vector for the noun run and a vector for the verb run). When target items were ambiguous in relation to their grammatical class, they were assigned the one most frequently observed in the corpus in order to extract the corresponding vector.

²When all vector components are non-negative (as resulting from the settings of our DSM), the cosine is also non-negative.

³In a sample of similar size (n = 113) extracted from the Semantic Priming Project (Hutchison et al., 2013) no correlation between OSC and priming effect was found (ρ = .01; p = .8898). This additional piece of evidence confirms that OSC is efficiently capturing a dimension encompassing both form and meaning; in fact, it is associated with morphological priming, but becomes irrelevant when purely semantic conditions are under examination.