Lynnika Butler

LING 535

March 25, 2004

Assignment 17

1. Marantz and Embick propose that the irregular/regular difference with respect to frequency effects can be captured even in a system with a rule-based approach to irregulars. What do they propose happens in the case of irregulars to produce this effect?

M & E claim that the past tense forms of irregular verbs are syntactically computed just like regular verbs (in both cases, [+past] suffix is added to a root); however, the VI for irregular [+past] is Æ and is associated with a list of the verbs that require this suffix (-ed doesn’t require such listing since it is the default suffix). Thus, when we encounter a regular past tense form, the frequency tallies of the root and the regular suffix –ed are both upped, but there is no tally for the whole past tense form because no listing is required for it; but when we encounter an irregular past tense form such as gave, the root and suffix tallies are upped, and so is the tally of the combined word (“the instantiation of the relation encoded by having give on the list” of irregular verbs).

2. Longtin et al. make a distinction between orthographically related and pseudo-derived words. In English, which relationship would the following pairs exhibit?
partner / part orthographic (-ner is not a morpheme)
planet / plan orthographic (-et is not a morpheme, at least I don’t think so)
entertain / enter orthographic (-tain is not a morpheme)
teeth / tea pseudo-derived? (since –th is an suffix on some ordinal numbers...?)
window / wind orthographic (-ow is not a morpheme)
daisy / daze pseudo-derived (-y is a common adjectival suffix)

3. What is 'cross-modal' priming?

A priming stimulus is presented to a subject in one mode (usually auditory or visual) for a short time, and then a target stimulus is presented in a different mode (i.e. if the prime was auditory, the target is visual & vice versa). If reaction times are faster when the prime and target are in a certain relationship (could be homophones, share a morpheme, semantic properties etc.) than when they are not, it is assumed that the brief exposure to the priming stimulus has “primed” the target, i.e. has initiated some sort of lexical access or activation process which activates the target somewhat before the target is actually given.

4. The results of experiment 1 suggest that there is agressive morphological parsing very early on in the perception process. Why do the authors think that experiment 2 failed to reveal the effects of this parsing?

They propose that upon exposure to a superficially complex word, an automatic decomposition process breaks the stimulus down into potential morphemes. Only if those morphemes really are semantically transparent, the decomposition facilitates lexical access. Since the masked primes were shown for a very short duration, the effects of the decomposition process could carry over into the response to the target. In the cross-modal priming experiment, however, the prime stimulus was longer and consciously perceived, and by the time subjects received the target stimulus the morphological decomposition process had already run its course and essentially discarded any items whose morphological composition was non-transparent; only in the semantically transparent cases, where the decomposition process had led to a successful lexical lookup, was there a priming effect because the lookup had been successful and an actual word had been found which could activate the target.