Final exam-Review questions 111046001 廖昀萱 Ruby
1. How do we process morphologically complex words? (Please justify your answers with
experimental findings.)
There were two kinds of arrangements concerning how we process complex words
morphologically. One argument is that for the sake of storage economy, words are stored based on
the root (base) word and morpheme. However, some argued that this kind of processing is
complicated instead of just accessing a single word. Several researchers have conducted
experiments to test such arrangements.
Mackay (1978), Taft and Forster (1975) and Snodgrass and Jarvella (1972) had same point of
view. In Mackay’s (1978) test, participants were presented verbs auditorily, then they were asked to
name out a related noun as quickly as they can. The results showed that derivational complexity
influenced the response time (i.e. -ment > -ence > -ion). Taft and Forster (1975) assumed that if a
word is a single-morpheme word, then it would be accessed directly. However, if a word is a
prefixed word, then it would go through an initial prefix-stripping stage. The results showed that the
response times were longer for affixed words than for words without affixes. After the prefix has
been stripped, a search for the base word is undertaken. If successful, the final stage compares the
prefix and base word to see whether they are compatible.
Nevertheless, Taft and Forster (1975) found that when it comes to prefixed words versus
pseudo-prefixed words, the previous one were shorter in lexical decision times. To be more specific,
in Lima’s (1987) task, it clearly showed that Pseudoprefixed words received longer fixation times
than prefixed words.
2. Briefly describe the three lexical access models: Search Model, Logogen Model, and Cohort
Model. p.118-119
Search Model is proposed by Forster (1976). He assumed that the lexicon is autonomous or
independent of other systems involved in language processing. In this model, Forster built it as a
library where there are different types of catalogs, when one has the need to search certain book, he/
she then uses the access files, then goes to the shelf to search for the book that is being looked for.
Word recognition system is divided into the two parts: orthographic and phonetic. When the input is
matched to one of these two items, a pointer to an entry in the master lexicon is retrieved. When this
entry is retrieved, other parts of the word such as its syntactic function are then retrieved as well.
Final exam-Review questions 111046001 廖昀萱 Ruby
Logogen Model is proposed by Morton in 1969. It is a a model of speech recognition that uses
units called "logogen" to explain how humans comprehend spoken or written words. The logogen is
activated in either of two ways: by sensory input (orthographic or phonological features) or by
contextual information. Whenever a logogen has collected enough evidence as an activation, the
threshold of the logogen then reaches to a certain level, and the word is recognized.
Cohort Model is a model of auditory word recognition. In this model, to activate spoken word
recognition, three stages are brought up according to Marslen-Wilson (1987). When individual gets
the input, an acoustic-phonetic analysis based on such input gets involved, following by an
activation a set of lexical candidates (word-initial cohort). Then, by strictly abiding bottom-up
processing, one of the cohort is selected for further analysis. Finally, the selected lexical item is
integrated into the ongoing semantic and syntactic context.
3. How do we process lexically ambiguous words? (i.e., all meanings are activated, or only the
most dominant/frequent one is? Please explain with at least one experiment to justify your
answer.) p.124
When encountering ambiguous words in a sentence or a context, there is a process of activating
more than one meaning of an ambiguous word. This statement was brought up by Foss in 1970, he
applied the phoneme-monitoring technique to the study of lexical ambiguity. Cairns and Kamerman
(1975) later on expanded Foss’ statement, they varied the time between the ambiguous word and the
phoneme that was to be monitored. The result suggest that although multiple meanings of an
ambiguous word are shortly brought up throughout the process, the ambiguity is quickly eliminated.
Also, Swinney (1979) found out that even when an ambiguous word is in a context that has a strong
bias, multiple meanings of ambiguous words are briefly activated.
