Li6 Phonology and Morphology

1
Li6 Phonology and Morphology

• Morphological features

2
Todays topics

• compositional vs holistic morphology
• evidence for morphemes being composed of
  (semantic and grammatical) features
• some uses of morphological features

3
Holistic or compositional?

• Semantics
• Holistic (Bock and Levelt 1994)
• Componential (Dell 1986)
• Morphology
• Whole-word representations (Butterworth 92,
  Bybee 95)
• Why?
• Reductionist/null hypothesis
• Morphological composition (Taft Forster 1975,
  Levelt 1989)
• Why?
• general tenet that the brain avoids redundant
  information storage
• Account for productivity

4
Semantic features
Problems

• Fuzzy categories
• we tend to think of concepts expressed by words
  and phrases as clear-cut, but
• what does rich mean? tall? blue?
• core/periphery
• Operators
• some, every, and, or, etc.
• some SOME, where for any sets X, Y
  SOME (X) (Y) iff X ? Y ? Ø

? is a proper subset of ? the
intersection of
5
Gradient CategoryRepresentations
Discrete CategoryRepresentations
6
Semantic features
Problems core vs periphery
ostriches hummingbirds pigeons robins magpies hawk
s storks penguins
its actually easy to model this sort of
structure using features
7
Evidence for semantic features

• Capture generalizations about natural classes
  (humans, etc.), subsets
• What do the following have in common? girl woman
  witch governess dominatrix
• Classifiers
• Thai uses khon to count human items
• khru lâ j khon teacher three person three
  teachers
• ma sí tua dog four body four dogs
• Overt marking needed for morphosyntactic
  agreement (gender, number...)
• Subcategorization
• admire ?my cat admires me my plants admire me
• requires human or animate subject
• assassinate requires human, important,
  (?)political object

8
Inversion effects

• (exact) opposites
• goodbad
• errors (TBD)
• Morphological inversion rules
• Afro-Asiatic gender polarity (Meinhof 1912)
• Arabic 3-10 take opposite gender of their noun
• ?ala?atu banina 3 (f) sons (m)
• ?ala?u banatu 3 (m) daughters (f)
• Somali (non-internal) plurals (Zwicky and Pullum
  83, Lecarme 2002)
• def art masc /-ka/ fem /-ta/
• agr Soomáali-ga the Somali (m) Soomaalí-da
  the Somalis
• Spanish theme vowel polarity (Fitzpatrick,
  Nevins, and Vaux 2004)
• The theme vowel in the present subjunctive is the
  opposite of that found in the present indicative

9
Evidence for morphological features

• speech errors

10
Speech errors

• morphological errors
• have to went for had to go
• have teachen for have taught
• semantic substitution errors
• insertion or blending of related or opposite
  words
• bridge of the neck (nose) a tennis athler
• he has to pay her rent (alimony) I really like
  to hate to get up in the morning
• tend to preserve grammatical category
• a laboratory in our own computer
• tend to preserve grammatical gender if the target
  utterance requires the production of a
  gender-marked element (Vigliocco, Vinson,
  Indefrey, Levelt, Hellwig 2004)
• seems to support gender as a feature, especially
  if forms of different phonological form but
  identical gender pattern together

11
Verb errors and features

• Ashenfelter and Eberhard 200x
• Observation
• words sharing semantic features compete for
  insertion when encoded in the same local context
  (Breedin, Saffran, and Schwartz 1998)
• Hypothesis
• In perseverative and anticipatory speech errors,
  replacement of simpler form (e.g. GO) by more
  complex form (JOG) should be more common
• Results
• Hypothesis supported

12
Evidence for morphological features

• lexical access effects

13
Lexical access and features

• When one wants to name the picture of an object,
  the structural description of the picture is used
  to activate a conceptual representation
  (typically, a bundle of semantic features).
  (Bachoud-Lévi and Dupoux 2003163)

Model from Bachoud-Lévi and Dupoux 2003181,
based on Levelt et al 1999
14
The Internal Lexicon
Conceptual Level
Grammatical Level
Word-form Level
15
Partial access in TOT states and aphasia

• Certain types of grammatical information are
  retrievable independent of lexeme suggests that
  these items are stored as independent features in
  the lexical entry
• Burke et al 1991
• Alternate words retrieved in TOTs are of the same
  grammatical class, number, and verb tense as the
  target word
• Caramazza and Miozzo 1997
• gender retrievable without access of lexeme in
  TOTs
• Similar effect for gender found with some
  aphasics (Caramazza)
• Cf. grammatical gender of distractor affecting
  production of German phrases composes of article
  picture namelonger latency with gender
  mismatch (Schriefers and Teruel 2000)
• Bachoud-Lévi and Dupoux 2003
• Numerals, days, and months spared vs matched
  controls
• Cf also category-specific deficits (animals,
  vegetables, tools)

16
N vs V in aphasia

• Rapp and Caramazza 2002
• double dissociation of grammatical category vs
  modality in single aphasic
• greater difficulty speaking nouns vs verbs
• greater difficulty writing verbs vs nouns
• supports the representation of grammatical
  category distinctions at post-semantic levels of
  representation and processing

17
Evidence for morphological features

• priming effects

18
Semantic priming
19
Meyer Schvaneveldt 1971
Reaction-time pattern suggested semantic
organization of mental lexicon
20
Historical changes

• semantic change
• morphologisation
• syncretism

21
Morphologisation in ASL
Frishberg, Nancy. 1975. Arbitrariness and
iconicity historical change in American Sign
Language. Language 51.3696-719.
22
Modeling semantic change

• Loss/addition of semantic feature(s)
• Parallel to regularization in morphology
• Murder, swarthy
• murder kill murder
• Features added to the lexical entry for murder
• human victim
• unlawful
• intentional
• violent

23
Analogical leveling

• Chaucer, Troilus and Criseyde III.1009-1012My
  goode, myn, not I for-why ne howThat Jalousye,
  allas! that wikked wivere,Thus causeless is
  cropen in-to yowThe harm of which I wolde fayn
  delivere!ne not, wivere wyvern, viper
• Chaucer, The Romaunt of the Rose
  247-251Envye.And by that image, nygh
  y-nough,Was peynt Envye, that never lough,Nor
  never wel in herte ferdeBut-if she outher saugh
  or herdeSom greet mischaunce, or greet
  disese.peynt painted, herte heart, ferde
  seemed

24
Analogical leveling

• plural eyen ? eyes kine ? cows
• past clomb ? climbed crope ? crept lough
  ? laughed yold ? yielded holpen ? helped

25
What is analogy?

• Traditional view
• dog dogs eye X
• Our view of this type of analogya. plural
  ? /-n/ / class 1 /-r?n/ / class
  2 /-Ø/ / class 3 /-z/ / elsewhereb.
  stage I eye, noun, class 1 stage II eye,
  noun
• Here analogy loss of exceptional marking.

26
Syncretism
27
The Latin declensions
from Halle and Vaux 1998
28
The Latin declensions
29
Some uses of morphological features
30
Gender agr with mixed pls

• hierarchy of gender agreement with mixed plurals
  in Sanskrit (e.g. Rama and Sita went to the
  store)
• m n ? masculine
• m f ? masculine
• n f ? neuter

31
Underspecification and competition in
DM(McGinnis 1996)
32
Impoverishment

• Sauerland 1995 on Norwegian
• adjectival endings differ in strong vs weak
  syntactic positions (see (10))
• in the strong set, -e appears to be the default
• by hypothesis, its not accidental that -e is
  default in the strong context, and also appears
  everywhere in the weak context
• DM analysis
• /t/ ? _ , -pl, neut / Adj _
• // ? _ , -pl, -neut / Adj _ Ø?
• /e/ ? elsewhere / Adj _
• Impoverishment rule neuter ? Ø / weak context
• This rule blocks t and - from being inserted,
  since each has a gender specification
• Therefore the default is inserted in all such
  cases

33
Harley and Ritter 2002

• McGinnis 2005
• languages without a dual category conflate the
  dual with the plural. Likewiselanguages without
  an inclusive category conflate the inclusive with
  first person.
• HR 2002 (from McG 2005)
• the morphosyntactic features of a given language
  are subject to MINIMAL CONSTRASTIVE
  UNDERSPECIFICATION only contrastive features
  appear in the underlying representation, while
  non-contrastive features are filled in by default
  rules (HR 498 see also Rice and Avery 1995,
  Brown 1997). HR propose that a language lacking
  a dual category has only Grp in the underlying
  representation, while Min is filled in by a
  default rule when Grp is absent (HR489). Thus
  the underlying representation of the system in
  (6a) is as in (8) the singular category has no
  number features, while the dual/plural category
  has only the feature Grp.

34
References

• Anne-Catherine Bachoud-Lévi A1 and Emmanuel
  Dupoux. 2003. An influence of syntactic and
  semantic variables on word form retrieval.
  Cognitive Neuropsychology 20.2163-188.
• Badecker, W. and Alfonso Caramazza. 1991.
  Morphological composition in the lexical output
  system. Cognitive Neuropsychology, 8(5), 335-367.
  
• Bock, K. and W. Levelt. 1994. Language
  production grammatical encoding. In M.
  Gernsbacher, ed., Handbook of psycholinguistics,
  pp. 945-978. San Diego Academic Press.
• Burke, D., D. MacKay, J. Worthley, and E. Wade.
  1991. On the tip of the tongue what causes word
  finding failures in young and older adults?
  Journal of Memory and Language 30542-579.
• Butterworth, B. 1992. Disorders of phonological
  encoding. Cognition 42261-286.
• Calabrese, Andrea. 1995. Syncretism Phenomena in
  the Clitic Systems of Italian and Sardinian
  Dialects and the Notion of Morphological Change.
  In Jill Beckman, ed., Proceedings of NELS 25, pp.
  151-173.
• Calabrese, Andrea. 1998. Some Remarks on the
  Latin Case system and its development in Romance.
  In J. Lema and E. Trevino, eds., Theoretical
  Advances on Romance Languages, Amsterdam John
  Benjamins, pp. 71-126.
• Calabrese, Andrea. 2002. On Impoverishment and
  fission in the verbal morphology of the dialect
  of Livinallongo. In Christina Tortora (ed.)
  Studies on Italian Dialects. Oxford Oxford
  University Press, pp.3-33.
• Clark 1973
• Dell, Gary. 1986. A spreading activation theory
  of retrieval in sentence production.
  Psychological Review 93283-321.
• Fitzpatrick, Justin, Andrew Nevins, and Bert
  Vaux. 2004. Exchange Rules and Feature-value
  Variables. Presented at The 3rd North American
  Phonology Conference, Concordia University,
  Montréal, Québec.
• Franks, Steven. 1995. Parameters of Slavic
  Morphosyntax. Oxford Oxford University Press.
• Greenberg, Joseph. 1967. The first (and perhaps
  only) non-linguistic distinctive feature
  analysis. Word 23.1214-220.
• Gvozdanovic, Jadranka. 1991.Syncretism and
  Paradigmatic Patterning of Grammatical Meaning.
  In F. Plank, ed., Paradigms. Mouton de Gruyter,
  Berlin, pp. 133160.
• Halle, Morris Alec Marantz. 1993. Distributed
  Morphology and the pieces of inflection. In K.
  Hale S.J. Keyser, eds., The View from Building
  20 Essays in Linguistics in Honor of Sylvain
  Bromberger, 111-176. Cambridge, MA MIT Press.
• Halle, Morris and Bert Vaux. 1998. Theoretical
  Aspects of Indo-European Nominal Morphology The
  Nominal Declensions of Latin and Armenian.  In
  Mír Curad Studies in Honor of Calvert Watkins,
  ed. Jay Jasanoff, H. Craig Melchert and Lisi
  Oliver. Innsbrucker Beitraege zur
  Sprachwissenschaft, Innsbruck, 223-240.
• Halle, Morris. 1997. Distributed morphology
  Impoverishment and fission.'  In MITWPL 30
  Papers at the Interface, ed. Benjamin Bruening,
  Yoonjung Kang and Martha McGinnis.  MITWPL,
  Cambridge, 425-449.
• Halle, Morris. 2003. The Latin noun declension.
  Manuscript, MIT.
• Harley, Heidi Elizabeth Ritter. 2002. Person
  and number in pronouns A feature-geometric
  analysis. Language 78, 482-526.