Great effort has been put in recent years on understanding how letter position is represented in the orthographic code (e.g., Kinoshita & Norris, 2009). The same issue also emerges at the morphological level, as the system must obviously be able to distinguish XY from YX, and to understand that PREHEAT is an existing word, while HEATPRE is not. A first crucial issue on this respect is whether or not morpheme representations are position-specific. In Experiment 1, we compared the rejection times obtained by normal readers on (i) nonwords made up of a stem and a suffix in their natural position (e.g., PASSMENT), (ii) nonwords including a stem and a non-morphological ending (PASSMANT), (iii) nonwords in which an existing suffix precedes an existing stem (MENTPASS), and (iv) nonwords in which a non-morphological word ending precedes an existing stem (MANTPASS). While rejection times were slower in condition (i) than in condition (ii) - replicating the morpheme interference effect observed by Caramazza et al. (1988) -, (iii) and (iv) elicited comparably fast responses, thus suggesting that MENT was not recognised as a morpheme at word onset. These results were replicated in Experiment 2, where orthographically implausible words (e.g., HYENA) were used as fillers to make sure that participants could not reject nonwords in conditions (iii) and (iv) just on the basis of their unfamiliar orthography (suffixes are in fact very infrequent word onsets). The claim that suffixes are not recognised when occurring in word-initial position was tested even more severely in Experiment 3, where scrambled-morpheme nonwords (e.g., NESSKIND) were used; in fact, if morphemic representations were even loosely activated by these nonwords, they would pass on this activation to an existing word representation (KINDNESS), thus maximising the interference effect. However, rejection times on scrambled-morpheme nonwords did not differ from those that emerged on orthographic controls (e.g., NELSKIND), confirming that suffixes are not recognised at word onset. Although similar work on prefixes and stems is required before taking firm conclusions, these results strongly suggest that morpheme representations are position-specific.
Crepaldi, D., Rastle, K., Davis, C. (2009). What's the difference between OVERHANG and HANGOVER? Morpheme position representation in the human mind. Intervento presentato a: 6th Morphological Processing Conference (MOPROC), Turku.
What's the difference between OVERHANG and HANGOVER? Morpheme position representation in the human mind
CREPALDI, DAVIDE;
2009
Abstract
Great effort has been put in recent years on understanding how letter position is represented in the orthographic code (e.g., Kinoshita & Norris, 2009). The same issue also emerges at the morphological level, as the system must obviously be able to distinguish XY from YX, and to understand that PREHEAT is an existing word, while HEATPRE is not. A first crucial issue on this respect is whether or not morpheme representations are position-specific. In Experiment 1, we compared the rejection times obtained by normal readers on (i) nonwords made up of a stem and a suffix in their natural position (e.g., PASSMENT), (ii) nonwords including a stem and a non-morphological ending (PASSMANT), (iii) nonwords in which an existing suffix precedes an existing stem (MENTPASS), and (iv) nonwords in which a non-morphological word ending precedes an existing stem (MANTPASS). While rejection times were slower in condition (i) than in condition (ii) - replicating the morpheme interference effect observed by Caramazza et al. (1988) -, (iii) and (iv) elicited comparably fast responses, thus suggesting that MENT was not recognised as a morpheme at word onset. These results were replicated in Experiment 2, where orthographically implausible words (e.g., HYENA) were used as fillers to make sure that participants could not reject nonwords in conditions (iii) and (iv) just on the basis of their unfamiliar orthography (suffixes are in fact very infrequent word onsets). The claim that suffixes are not recognised when occurring in word-initial position was tested even more severely in Experiment 3, where scrambled-morpheme nonwords (e.g., NESSKIND) were used; in fact, if morphemic representations were even loosely activated by these nonwords, they would pass on this activation to an existing word representation (KINDNESS), thus maximising the interference effect. However, rejection times on scrambled-morpheme nonwords did not differ from those that emerged on orthographic controls (e.g., NELSKIND), confirming that suffixes are not recognised at word onset. Although similar work on prefixes and stems is required before taking firm conclusions, these results strongly suggest that morpheme representations are position-specific.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
