How are ambiguous words processed? Behavioral

0 downloads 0 Views 1MB Size Report
concreteness, AoA, etc.) Exp. 1 (Dictionary): disadvantage in RTs and errors (all ps < .05). Exp. 2 (Subjective): advantage in RTs (all ps < .05). 618. 617. 593.
How are ambiguous words processed? Behavioral and EEG evidence of the role of number of meanings, relatedness of meanings and stimuli selection criteria Juan Haro and Pilar Ferré. Universitat Rovira i Virgili. [email protected]

Ambiguity advantage? Words with multiple meanings (i.e., ambiguous words; e.g., bat) are usually recognized faster than words with one meaning (i.e, unambiguous words) in LDT (e.g., Azuma & Van Orden, 1997; Jastrzembski, 1981; Hino et al., 2010; Rubenstein et al., 1970). But other studies have observed a disadvantage for words with multiple meanings along with an advantage for words with multiple related senses (e.g., newspaper; Rodd et al.,

2002), suggesting that number-of-meanings (NOM) inhibits recognition, whereas relatedness-of-meanings (ROM) facilitates it.

Aims 1. Examine if the approach for NOM estimation affects LDT results 2. Examine the role of relatedness of ambiguous word meanings (ROM) in word processing 3. Examine the neural correlates of ambiguous word processing 4. Test the assumptions of models of ambiguous word recognition

Study 1 Aim: Examine if the approach for NOM estimation affects LDT results We manipulated the approach for NOM estimation in two LDT experiments: ◦ Experiment 1: Dictionary NOM ◦ Experiment 2: Subjective NOM (homonyms vs unambiguous)

Results Participants: Exp. 1 (39), Exp. 2 (38), Exp. 3 (44)

646

Stimuli: Exp. 1: 36 in both conditions Exp. 3: 27 ambiguous words (NOM > 1.5 and ROM < 3) and 27 unambiguous words (NOM < 1.4).

618

617

593

Conditions were matched in a large number of lexical and semantic variables (e.g., frequency, number of letters, concreteness, AoA, etc.)

Exp.1 (Dictionary NOM) Ambiguous words

Exp.2 (Subjective NOM) Unambiguous words

Exp. 1 (Dictionary): disadvantage in RTs and errors (all ps < .05) Exp. 2 (Subjective): advantage in RTs (all ps < .05)

Conclusions Study 1 The approach for NOM estimation influences LDT results. This may explain the disparity of results observed in previous studies. There is not a full correspondence between dictionary meanings and subjective meanings. NOM facilitates word recognition (but only if it is based on the speakers’ knowledge)

Unsolved questions 1. Why NOM facilitates word recognition? Semantic enhanced activation (e.g., Borowsky & Masson, 1996; Hino & Lupker, 1996). Ambiguous words would elicit a large amount of semantic activation. 2. Does relatedness of meanings (ROM) influence word recognition?

Study 2 We compared ambiguous words (differing in their ROM: polysemes [high ROM] and homonyms [low ROM]) and unambiguous words in LDT. We recorded event related potentials during the task, focusing on the N400 component (semantic processing). According to semantic enhanced accounts: differences between ambiguous and unambiguous words in N400. If ROM influences word recognition: differences between polysemes and homonyms in behavioral measures and N400.

Results Participants: 25 students from the URV

Reaction times (in ms)

Stimuli:

Reaction times (in ms)

597

76 ambiguous words, 38 of them polysemes (ROM>2.5) and 38 homonyms (ROM