Li8 Structure of English

1
Li8 Structure of English

• Evidence for morphological structure

2
Morphology

• Maybe in order to understand mankind, we have to
  look at the word itself. Mankind. Basically, its
  made up of two separate words - mank and ind.
  What do these words mean? Its a mystery, and
  thats why so is mankind.
• Jack Handey, Deep Thoughts

3
Morphemes

• Morphology is the study of morphemes and their
  behavior.
• A morpheme is the minimal unit of meaning.
• dogs contains 2 morphemes
• 1 d?g quadruped of the genus canis
• 2 -z plural
• Note morphemes ? words
• A word is the smallest element that can occur in
  isolation.

4
Evidence for morphological decomposition and
storage

• New singulars
• New morphemes
• Psycholinguistic evidence
• Rule-governed behavior
• Priming studies
• Event Related Potentials
• Speech errors
• Disorders

5
New singulars

• indice, homo sapien

6
New morphemes

• -a/oholic
• -burger
• -gate
• Watergate, Fornigate
• -head
• pothead, metalhead, techhead, gearhead,
  theoryhead, crackhead, hockey head
• Mc-
• McMansion, mcjob, mcnews

It's true, I'm a Rageaholic...I just can't live
without Rageahol!
7
McDonald's not lovin' McJob dictionary definition

• CHICAGO, Illinois (AP, 11-2003) -- McDonald's
  says it deserves a break from the unflattering
  way the latest Merriam-Webster's Collegiate
  Dictionary depicts its job opportunities. Among
  some 10,000 new additions to an updated version
  released in June was the term "McJob," defined as
  "low paying and dead-end work." In an open letter
  to Merriam-Webster, McDonald's CEO Jim Cantalupo
  said the term is "an inaccurate description of
  restaurant employment" and "a slap in the face to
  the 12 million men and women" who work in the
  restaurant industry. The company e-mailed the
  letter to media organizations Friday, and it also
  was published in the Nov. 3 edition of an
  industry trade publication. Cantalupo also wrote
  that "more than 1,000 of the men and women who
  own and operate McDonald's restaurants today got
  their start by serving customers behind the
  counter. McDonald's, the world's largest
  restaurant chain, has more than 30,000
  restaurants and more than 400,000 employees. Walt
  Riker, a spokesman for McDonald's, said the Oak
  Brook, Illinois-based fast-food giant also is
  concerned that "McJob" closely resembles McJOBS,
  the company's training program for mentally and
  physically challenged people.
• "McJOBS is trademarked and we've notified them
  that legally that's an issue for us as well,"
  Riker said.

McJob (m?kdZAb) n. A low-paying job that
requires little skill and provides little
opportunity for advancement. Source
Merriam-Webster Online
8
Psycholinguistic evidence
9
Over-regularization

• went ? goed ? went
• Happens c. late twos
• Suggests acquisition of morphological rules

10
Wug tests

• Berko 1958
• 3 surface manifestations of English regular
  plural /-z/ z, s, ?z
• When/how do children learn these rules?
• Test paradigm
• Children are presented with a pretend creature
  and told, "This is a wug."
• Another wug is revealed, and the researcher says,
  "Now there are two of them. There are two __."
• Results
• Very young children are baffled by the question
  and are unable to answer correctly, responding
  with e.g. two wug."
• Children in grade 1 were almost fully competent
  with both s and z.
• Both preschool and first-grade children dealt
  poorly with ?z, giving the correct answer less
  than half the time, possibly because it occurs in
  the most restrictive context.
• Major finding
• The first experimental proof that young children
  have extracted generalizable morphological rules
  from the language around them.

11
Productive irregulars

• Al Jallad, Flint, and Richardson 2005
• Subjects come up with past tense for unfamiliar
  V.
• 1. Jim BLINGS everyday.
• Yesterday John __________________.
• 2. Alan CHINES everyday.
• Yesterday Alan __________________.
• Subjects in the experiment
• 26 children age 6-7 (oral test)
• 44 adults (written questionnaire)
• Sample Over-Irregularizations
• children
• pake ? pakeded
• bling ? blung, blang
• flink ? flunk
• frim ? frand
• adults
• mang ? mung
• shride ? shrude
• bling ? blank

Overall Rate of Irregular forms supplied
Individual Differences in Irregularization bars
show percentages of subjects making 0-20
irregulars, 20-40 irregulars etc.
12
ERPs

• McKinnon et al. 2003.
• Observation
• Pronounceable non-words (flermuf) elicit
  larger-amplitude N400 components than words
  (muffler).
• Prediction
• If readers treat non-words containing
  non-productive morphemes (in-ceive) as unanalysed
  wholes, then these non-words should elicit larger
  N400s than matched words (receive).
• Results
• Bound-stem non-words elicit a brain response
  highly similar to that elicited by real words.
• Conclusion
• Morphological decomposition and representation
  extend to non-productive morphemes.

13
BrothBrothelBrother

• Rastle et al. 2004
• Participants made visual lexical decisions to
  stem targets preceded by masked primes sharing
• (a) a semantically-transparent morphological
  relationship with the target (e.g.,
  cleaner-CLEAN)
• (b) an apparent morphological relationship, but
  no semantic relationship, with the target (e.g.,
  corner-CORN)
• (c) a non-morphological form relationship with
  the target (e.g., brothel-BROTH).
• Results showed significant and equivalent masked
  priming effects in cases in which primes and
  targets appeared to be morphologically related,
  and priming in these conditions could be
  distinguished from non-morphological form
  priming.
• These findings suggest a level of representation
  at which apparently complex words are decomposed
  based on their morpho-orthographic properties.

14
Morphological speech errors
15
Morphological errors

• Morpheme exchange
• slicely thinned
• Feature shift
• have to went for had to go
• Faulty access
• have teachen for have taught
• concludement for conclusion
• Experimentally controlled morphological errors

16
Janssen and Humphreys 2002

• General logic
• If a string of phonemes with an identifiable
  meaning can move independently during a speech
  error, it is a production unit, with a separate,
  independent representation at some cognitive
  level.
• Inflectional morphemes (e.g. -ed, -ing, -s) are
  much more error prone than derivational morphemes
  (e.g. -er, -ness, -able, -ion) (Garrett 1980,
  Humphreys, 2002).
• Can we also find experimental speech error
  evidence of derivational decomposition?
• Prediction
• If derivational morphemes are stored
  independently from their stems, more
  morphological speech errors should occur when -er
  is a real morpheme than when it is only a
  pseudo-morpheme.

17
Method

• elicited morphological errors from speakers by
  presenting a series of word quads, made up of two
  pairs. Speakers read these words silently, after
  which they disappeared. Speakers were
  occasionally cued to respond aloud, as quickly as
  possible, to the immediately preceding item.
  Speakers had to then either repeat the previous
  phrases exactly, or had to swap words between the
  phrases. The critical pairs each consisted of an
  initial ller word plus a critical word stem or
  stemaffix.

18
Results

• More affix errors occurred on the morphologically
  complex forms than on the pseudo-morphs.
• Conclusion derivational affixes are also stored
  as separate morphemes.

19
Morphological disorders
20
Lexical disorder

• Badecker 2001
• CSS (65 years old) suffered a left
  cerebro-vascular accident in 1990, resulting in
  lexical impairments manifested in reading,
  repetition, and oral and written naming tasks.
• Apart from word-finding problems, CSSs spoken
  output is fluent and he exhibits no apparent
  comprehension impairment.

naming from definition Slender-bodied insect with
broad, often brightly-colored wings. butterfly
picture naming butterfly
monomorphemes 80 correct
doctor fly sun wheel pill
polymorphemes 50 correct!
21
Semantic disorders

• Deep dyslexia semantic errors during reading
• Buchanan et al. 2003
• Deep dyslexic patient (JO, 48, L temporo-parietal
  lobectomy)
• Acquired reading disorder characterized by
  production of
• morphological errors (e.g., SLEEP read as
  SLEEPING)
• semantic errors (e.g. HEART read as BLOOD)

22
Experiment

• Method
• Manipulate transparency in compound word naming
  study
• Results
• 94 stimuli
• 10 read correctly 9 of these were TT
• Transparent components read correctly more often
• Key point if there were no decomposition, all
  compound types should show the same error rate
• Response to PANCAKE
• cake, breakfast, man,. . .cake, man. . .cake,
  mancake, man, man, p-, p-, cake, birthday,
  breakfast, cake, p-a-n. . . cake, syrup...
• Typical semantic error target cake ? output
  birthday (shows decomposition)
• NB 2nd and last attempts are semantic associate
  for the whole compound (BREAKFAST, SYRUP)
• Shows successful access of meaning of the
  compound, even though she never manages to say
  the whole thing
• Conclusion
• JO can successfully parse compound into
  constituents and access meaning of the whole the
  errors occur after this point

23
Deep dyslexia

• Rastle et al. 2005
• Observation
• Deep dyslexics often make morphological errors in
  reading aloud
• e.g. sexist read as sexy
• Two competing explanations
• representations are morphologically structured
• low imageability/frequency word modified to
  visually similar word of higher
  imageability/frequency) (Funnell 1987)
• Prediction
• Theory 1 predicts more stem errors with
  morphologically complex words (killer) than with
  pseudosuffixed words (corner) and monomorphemic
  words (cornea) when one equalizes for
  imageability and frequency.
• Theory 2 predicts no difference between these 3
  categories.
• Result
• Theory 1 (but not Theory 2) supported

24
Regular vs. irregular

• Ullman et al. 1997, Ullman 2001 propose that
• the left inferior frontal cortex (along with the
  basal ganglia) is involved in rule-based
  computations
• Irregular forms dealt with differently
• Prediction
• A lesion to this region should impair only
  regular morphological processes, sparing the
  ability to produce irregular morphology.
• Test
• Four types of patients Parkinsons, Alzheimers,
  anterior aphasics, and posterior aphasics
• Patients were asked to perform a sentence
  completion task requiring the production of a
  verb in the past tense
• E.g. Every day I dig a hole. Just like every
  day, yesterday I ____ a hole
• Results
• anterior aphasics and Parkinson patients
  performed worse with regular than irregular verbs
• the opposite for posterior aphasics and Alzheimer
  patients
• Conclusion
• Results consistent with theory that the left
  frontal cortex and basal ganglia are involved in
  rule-based language processing but not in the
  processing of irregular morphology, and that
  temporal lobe areas are implicated in the storage
  of lexical forms.
• But

25
Regular vs. irregular

• Problem with Ullmans theory
• Agrammatic (Brocas) aphasics who perform worse
  on irregular than regular morphological
  transformations (producing irregular V forms)
• patient RC (Shapiro Caramazza 2003)
• patient MR (Laiacona Caramazza 2004)
• Spanish-Catalan bilinguals (Caramazza et al.
  2004)
• Consistent with the classical view that left
  inferior frontal cortex is involved in
  morphosyntactic processing and not just in
  rule-based transformations.

26
Conclusions

• Humans decompose words into morphemes and store
  them as such.
• Both regular and irregular/restricted morphology
  involve rule-governed (de)composition.
• blang, inceive, Brocas aphasics
• Humans look for morphological structure even when
  there isnt any.
• cheeseburger, corner

27
References

• Andrews, S. 1986. Morphological influences on
  lexical access Lexical or nonlexical effects?
  Journal of Memory and Language, 25, 726-740.
• Andrews, S., B. Miller, K. Rayner. 2004. Eye
  movements and morphological segmentation of
  compound words There is a mouse in mousetrap.
  European Journal of Cognitive Psychology
  16285-311.
• Badecker, William. 2001. Lexical composition and
  the production of compounds evidence from errors
  in naming. Language and Cognitive Processes 16.4.
• Buchanan, Lori, Shannon McEwen, Chris Westbury,
  and Gary Libben. 2003. Semantics and semantic
  errors Implicit access to semantic information
  from words and nonwords in deep dyslexia. Brain
  and Language 846583.
• Caramazza, Alfonso, Alessandro Laudanna, and
  Cristina Romani. 1988. Lexical access and
  inflectional morphology. Cognition 28.
• Fiorentino, Robert. 2006. Masked priming of
  compound constituents Implications for
  morphological decomposition. Manuscript,
  University of Maryland.
• Janssen, Dirk and Karin Humphreys. 2002.
  Morphological speech errors on agentive and
  comparative affixes. Third International
  Conference on the Mental Lexicon, Banff, Canada.
• McKinnon, R., M. Allen, L. Osterhout. 2003. 
  Morphological decomposition involving
  non-productive morphemes ERP Evidence.
  Neuroreport 14883-886.
• Marslen-Wilson et al. 1994. Psychol Rev 1013-33.
• Rastle, Kathleen, Matthew Davis, and Boris New.
  2004. The Broth in my Brothers Brothel
  Morpho-Orthographic Segmentation in Visual Word
  Recognition. Psychonomic Bulletin Review
  11.61090-1098.
• Rastle, Kathleen, et al. 2005. New evidence for
  morphological errors in deep dyslexia. Brain and
  Language 97189-199.
• Shapiro, K., Alfonso Caramazza. 2003. Looming a
  loom Evidence for independent access to
  grammatical and phonological properties in verb
  retrieval. Journal of Neurolinguistics 1685-111.