Stem Homograph Inhibition and Stem Allomorphy: Stem Homograph Inhibition and Stem Allomorphy: RepresentingRepresenting
and Processing Inflected Forms in a Multilevel Lexical and Processing Inflected Forms in a Multilevel Lexical System, 1999System, 1999
&&Morphological Parsing and the Perception of Lexical Morphological Parsing and the Perception of Lexical
Identity:Identity:A Masked Priming Study of Stem Homographs, 2002A Masked Priming Study of Stem Homographs, 2002
William Badecker and Mark AllenWilliam Badecker and Mark Allen
Presented by Nataliya ChabanyukPresented by Nataliya Chabanyuk
Instructor Nina KazaninaInstructor Nina Kazanina
PsycholinguisticsPsycholinguistics
Research questionResearch question
Two views are opposed:Decompositional, parsing: (novel words as unboyfriendable )
VSWhole word based access(frequent morphological complex
words )
Any preference? Should the both work?
Research questionResearch question
When and where decompositional processes come into play in the comprehension system:
Only for learning?Or, also
For recognition and comprehension?
BackgroundBackground Evidences that morphological structure plays a role in the processing of complex
words are compatible with both
whole-word and decompositionalwhole-word and decompositional models of representation.
• Inhibitory priming between stem homographs in Italian:
• colp-a colp-o ‘guilt’/’blow’• coll-o/pont-e colp-o ‘neck/bridge’
BackgroundBackground
Interpretation:Interpretation: inhibition reflects a competetive realtionship between stem representations rather than whole-word representations in the mental lexicon.
BUTBUTPerhaps the orthographic similarity alone
can account for the inhibition that is observed in the stem homograph condition?
TerminologyTerminology
Processing/representational level:lemma VS lexeme
Lexeme:Lexeme: form based lexical representationLemma:Lemma: modality-neutral lexical
represenation
Lexemes goes & went have one lemme in common,
probably go.
Experiment 1 & 2, 1999Experiment 1 & 2, 1999To take stem homograph inhibition
as evidence for morphological decomposition,it must be shown that the inhibition derives
from stem-level competition and not justfrom word-level orthographic competition.
Ex.2 shows that words that areorthographically related in the same way that
stem homographs are will not inhibiteach other in the same way or to the same
degree that stem homographs do.
Experiment 1 & 2, 1999Experiment 1 & 2, 1999• Stem homograph: MOR-IA• Orthographic relative: MORAL• Stem homograph• allomorph: MUER-E MOR-OS• Orthographic related: MIR-AN• Unrelated control: SILL-A
Recognition of the plural noun mor-os (Moors) should be slowed not just by the verb form mor-ia(die, 1st/3rd person sing. Imperfect), but also by an allomorph of this form, muer-e(die, 3rd person sing, present)
Experiment 1 & 2, 1999Experiment 1 & 2, 1999
Participants: native speakers of Spanish
Procedure: Fixation cross for 400 ms50 ms after:Prime for 250 msImmediately after:Target and remained present until the participant made a response
Experiment 1, 1999
Experiment 2, 1999
DiscussionDiscussion
Competition may arise at another level within the lexical processing system:
specifically, at a level of representation where all the members of an inflectional paradigm share a single (abstract) morphological entry. (M-level).
• Results go AGAINSTAGAINST the hypothesis that the stem homoghraph effect can be reduced to an orthographic phenomena.
Ex.1, 2002Ex.1, 2002
Masked prime with subject awareness control
Partcipants: Spanish nativeProcedure: ££££££££££££ for 500 msMasked prime for 67 ms: 4 screen refresh
ticks at 16.67 msTarget for 500 msLexical desicion taskITI 1500 ms
Ex.1, 1999 & Ex. 1, 2002Ex.1, 1999 & Ex. 1, 2002
To contrast masked masked stem homograph primes with unmaskedunmasked stem homograph primes in the frame of the paersing model of inflectional processing.
Prediction:Prediction: Facilitative effect of masked stem
homographs primes AGAINSTAGAINST
Inhibition effect of unmasked stem homograph prime
Unmasked prime from Ex.1, 1999
Masked prime from Ex.1, 2002
DiscussionDiscussionStem homographs are not not just whole-word
orthographic neibours.The facilitation indicates that ambiguousstems activate multiple affiliated lemmas
prior to recognition. On the decomposition
plus selection model we find facilitation, ratherthan inhibition or no effect, because suppression
occurs only when lexical selection inducesthe conscious perception of a word as a particular
word.
Ex.2 & ex. 3, 2002Ex.2 & ex. 3, 2002To test hard-wired model against
the parsing model:
InhibitionInhibition for masked stem homograph allomorph in former case AGAINSTAGAINST nul-
effect in the latter case. cierr-a cerr-o close/hill To rule outrule out other potential reasons for the
expected null effect: e.g. semantics and to verify the masked prime facilitates other members of their inflectionalm cohort
cierr-a cerr-ar
Experiment 2, 2002
Experiment 3, 2002
Ex. 4, 2002Ex. 4, 2002
• To ensure that a masked stem homograph facilitates an allomorphic memebr of its competitor lemma’s paradigm and not just lexeme level entries that correspond to the exact orthographic form of the prime’s stem.
• cerr-o cierr-a• So, if masked RevSHA priming is
facilitative, then we will have evidence that masked SH facilitaiton derives from facilitaiton derives from lemma-level sourceslemma-level sources.
Experiment 3, 2002
Experiment 4, 2002
Experiment 1, 2002
Experiment 4, 2002
DiscussionDiscussionLexical processing does notdoes not operate exclusivelyin terms of whole-wordterms of whole-word recognition procedures
For familiar words with ambiguous morphological constituents.
Even if the access systemhas supplementary whole-wordsupplementary whole-word procedures
for familiar targets, the facilitation and inhibitionobserved, respectively, in masked and
unmasked stem-homograph priming indicatesthat the decomposition approachdecomposition approach still contributes
to the recognition process for these forms.
ConclusionConclusion Obligatory decompositionObligatory decomposition of complex words may be favored
regardless of whole-word familiarity because it offers a way to extract the
morpho-syntacticand conceptual information that is encoded
separately in affixes and stems.
Timing makes difference!Timing makes difference!