Morphology and Meaning in the English Mental Lexicon

1
Morphology and Meaning in the English Mental
Lexicon

• By William Marlsen-Wilson, Lorraine Komisarjevsky
  Tyler, Rachelle Waksler, and Lianne Older
• Presented by Robyn Maler

2
Questions

• How are lexical entries represented in the mental
  lexicon?
• Are their representations based on whole phonetic
  words (full listing hypothesis) or morphemes
  (morphemic hypothesis)?
• Are there differences between lexical
  representations at different levels?

3
Background

• Lexical entry is distinct from access
  representation
• Morphological category the basic linguistic
  characteristics of the affixes (derivational vs.
  inflectional, prefix vs. suffix)
• Semantic transparency whether the form is
  synchronically compositional
• Phonological transparency whether the form has
  the same phonetic shape for both its affixed and
  unaffixed versions

4
Experimental Task Design

• Cross-modal immediate repetition priming subject
  hears a multimorphemic spoken word (prime) and
  immediately after sees a visual probe (target)
• Subject must make a lexical-decision response to
  this probe
• Response facilitation (priming) is measured by
  response latency relative to a baseline condition
  (subjects response to same probe following
  unrelated prime)

5
Questions for Experiments 1-3

• Is the lexical entry for derived suffixed words
  in English morphologically structured?
• How does the semantic and phonological
  transparency of stem and affix morphemes affect
  the representation of a derived form?

6
Experiment 1

• Purpose to determine whether there is evidence
  for a level of morphologically structured lexical
  representation that abstracts away from shared
  surface phonetic properties

7
Table 1 Experiment 1
8
Results and Discussion

• Results are consistent with hypothesis that
  derived suffixed forms prime their free stems
  because of lexical entry processes and not just
  surface phonetic overlap

9
Experiment 2

• Purpose to determine whether the priming
  observed in Morph conditions in Experiment 1
  are simply due to semantic relationships between
  morphologically related pairs instead of shared
  morphemes in a morphologically structured mental
  lexicon

10
Table 2 Experiment 2
11
Results and Discussion

• Priming only occurs when there is a
  synchronically semantically transparent
  relationship between derived and stem forms
• Semantic links alone can produce priming, but
  semantic relatedness is not the only factor
  affecting facilitation!

12
Experiment 3

• Purpose 1 to study effects of morphological
  type and semantic transparency more rigorously
• Purpose 2 to investigate a new prime-target
  combination (stem-derived)

13
Table 3 Experiment 3
14
Results and Discussion

• confirm results of Experiment 2
• fit with prediction of shared-morpheme account of
  Sem, Morph priming

15
Experiment 4

• Purpose 1 to investigate semantic transparency
  for prefixing morphology
• Purpose 2 investigate morphological type
  (whether derived-derived and derived-stem
  prefixed pairs exhibit priming effects)

16
Table 4 Experiment 4
17
Results and Discussion

• Like the suffixed pairs, only Sem prefixed
  pairs showed priming
• Prefixed Sem derived-derived pairs show strong
  priming effects, consistent with idea that they
  are not cohort competitors
• Prefixed -Sem, Morph forms (e.g. mistake) are
  represented as monomorphemic items WHEREAS
  prefixed Sem, Morph forms (e.g. refasten) are
  broken down into abstract stems and prefixes at
  the level of lexical entry

18
Experiment 5

• Purpose to investigate stem-derived order in
  prefixed pairs

19
Table 5 Experiment 5
20
Results and Discussion

• Condition 3 results provide more evidence that
  there is no facilitation when there is no
  synchronic semantic basis for representing a word
  form as morphologically complex
• Results consistent with a model of lexical
  representation in which there are inhibitory
  links between suffixes but not prefixes that
  share the same stem

21
Experiment 6

• Purpose to explore relationship between prefixed
  and suffixed forms

22
Table 6 Experiment 6
23
Results and Discussion

• Consistent with a cohort-based model in which
  there are inhibitory links between competing
  suffixed forms but not prefixed and suffixed forms

24
In summary

• Semantically transparent suffixed pairs prime
  each other except in the case of two suffixed
  forms, which demonstrate a cohort-based
  inhibitory effect
• Semantically transparent prefixed pairs always
  prime each other
• Semantically opaque pairs do not prime each other
• Thus, semantically transparent forms are
  decomposed at the level of lexical entry, while
  semantically opaque forms are represented
  monomorphemically

25
contd

• Phonological opacity had no effect on results
• Thus, morphemes are phonologically abstract

26
What does it all mean?

• Results suggest that there is a
  modality-independent and morphologically
  structured lexical level
• The basic unit in terms of which the lexicon is
  organized, at least for derivational forms in
  English, is the developmentally-defined morpheme
• The findings are (kind of) consistent with a
  partial decomposition view of the lexicon