Typology of stative/active languages

1
Typology of stative/active languages

• Split intransitives, experiencer objects and
  transimpersonal constructions
• (re-)establishing the connection
• Andrej Malchukov

2
Introduction Sapirs proposal
Sapirs proposal inactive (object
inflecting) intransitive verbs in Amerindian
languages should be better analysed as
transitives Thus, forms like I sleep or I
think could be understood as meaning properly
it sleeps me, It seems to me (Sapir 1917
85). That is an So pattern is analysed as
transimpersonal (indefinite A) construction
with experiencer object
3
Some problems

• Some obvious functional similarities
• Both So constructions and transimpersonal
  experiencer O constructions involve experiential
  predicates
• But also some problems (cf. Merlan 1985)
• Structural
• in the former Experiencer is O, in the latter S
• former intransitive, the latter transitive
• Functional
• it sleeps me ???
• Heterogeneity of split-S languages
• agent/patient vs. active/stative (Mithun 1991)
• accusative based (So is a minor pattern) vs.
  ergative based (Sa is a minor pattern) (Nichols
  1992).
• Experiencer object constructions are transitive
  while split-intransitivity pertains in the first
  place to intransitives

4

• However
• The distinctions between So constructions, on the
  one hand, and transimpersonal constructions (TIC)
  and object experiencer constructions (OEC), on
  the other hand, are not always clear-cut
• Cf. Aikhenvald, Dixon Onishi 2001 (eds.) on
  oblique experiencers as non-canonical subjects.
• In spite of heterogeneity of split S languages
• most split-S languages are agent/patient rather
  than active/stative (Mithun 1991)
• most split-S languages are accusative based in
  the sense that So pattern is a minor class as
  compared to the open Sa class (Nichols 1992).

5
Outline of the talk

• Provide evidence that Sapirs analysis can be
  upheld, if
• Restricted to Split-S language where the
  patientive subject pattern is a minor pattern
• A connection between So pattern and transitive
  patterns (TIC and EOC) is understood in
  diachronic terms
• Present evidence from languages
• where Split-S pattern arose from reanalysis of
  transimpersonal and Object-experiencer
  constructions
• where object experiencers can be analysed as
  non-canonical subjects

6
Slave TIC without Split-S

• A construction with unspecified human subject
  pronoun in Slave
• Slave (Rice 1989 1020)
• tse-jI
• someone is singing
• kínase-tse-reyo
• someone chased him/her s/he is chased
• NB clearly distinct from split-S (note the overt
  AGR/A marker tse- ), but not the quasi-passive
  interpretation of TIC.

7
Extension of TIC Eskimo

• In West Greenlandic transimpersonal construction
  (TIC) restricted to weather verbs
• West Greenlandic (Fortescue 1984 59-61)
• Anurliup-patigut
• storm-3A-gt1pO.IND
• When we were caught by storm (lit. it stormed
  us)

8
Extension of TIC Eskimo

• In (Siberian) Yupik TIC is extended to other verb
  types to indicate lack of control
• Yupik (Emeljanova 1967 cf. Vaxtin 1995)
• Tagnygak axwasag-taa
• child.ABS crawl-3-gt3
• The child crawled
• NB similar to So pattern functionally (indicates
  lack of control), but different structurally (AGR
  clearly transitive). Therefore rather extended
  use of TIC than Split-S.

9
TIC reanalysed as split-S Tunica

• In Tunica (Haas 1941) So intransitives in
  inchoative forms are constructed as
  transimpersonals
• it-sickens-me I become sick
• Haas conclusion involuntary action verbs
  developed from transimpersonals (Haas 1941 59)

10
Split-S originating from EOC Koasati

• Koasati is considered split intransitive on the
  basis of its agreement system (cf. Mithun 1999
  237-8).
• the case system is accusative though
• An So verb
• (Anó-k) ca-libatli-t
• (I-NOM) 1sg.obj-burn-past
• I got burned

11
Split-S originating from EOC Koasati

• Morphologically, however, So verbs look like
  plain transitives (Kimball 1991 251). Cf.
• ca-libatli-t
• 1sg.obj-burn-past
• I got burned
• Nihahci ikba-k ca-libatli-t
• Grease hot-NOM 1sg.obj-burn-past
• The hot grease burned me

12
Conclusion on Koasati

• Kimballs conclusion the So pattern originated
  from reanalysis of impersonal 3 sg forms
• Note what features facilitated reanalysis
• So is marked by AGRo
• 3pA marker is zero
• But the same pattern attested in many other
  Split-S languages

13
Other Split-S languages Ika

• Other Split-S languages with zero 3rd p. zero A
  markers Dakota (Boas Deloria 1941, 76),
  Guarani (Gregores Suárez 1967 131), Ika (Frank
  1985 11)
• Ika (Frank 1985 11)
• So pattern
• Na-tikuma-na
• 1sgO-forget-DIST
• I forgot
• A transitive pattern
• Na-tsua-na
• 1sgO-see-DIST
• He saw me

14
Other Split-S languages Haida

• Haida (Enrico 2003, 93)
• Split-S in free/clitical pronoun marking
• There are no overt inanimate (low potency)
  agentive pronouns

15
Other Split-S languages Kiowa

• Kiowa restricted suppression of A agreement with
  experiential verbs
• ya-tây (Watkins 1980 137)
• (2,3sg.A)1sg.Ppl.O-awake.pf
• I awoke/smth woke me
• Watkins considers them as intransitives (thus,
  Split-S), although clearly modelled on
  transitives
• (or even di-transitives, with a dummy O marker)

16
Conclusions on reanalysis

• Thus reanalysis is facilitated, if
• So is marked by AGRo
• 3pA marker (one of the markers, often inanimate
  if a language has one) is zero.
• NB then a transitive pattern is formally
  indistinguishable from intransitive

17
TICs as semitransitives Navaho

• Even if transitive/intransitive distinction is
  marked otherwise, does not necessary prevent
  reanalysis, as TIC can reveal transitivity
  decrease
• Navaho allows an intransitive marker
  (classifier) in the Indefinite A construction
  (Kibrik 1996 291)
• Né-í-ø-l-zho?
• Md-3/ACC-3/NOM-TRANS-hunt.IT
• He repeatedly hunts it
• Ná-ø-?á-l-zho?
• Md-3/ACC-IND/NOM-DETRANS-hunt.IT
• Someone repeatedly hunts it
• Kibrik (1996) A indefinitensess as another
  transitivity parameter (in the sense of Hopper
  Thompson 1980)

18
From Experiencer Object constructions to Split-S
Papuan languages

• In Papuan languages objects in EOC tend to be
  reanalysed as non-canonical subjects
• Usan (Reesink 1987 139)
• Munon isig toar wA-r-a in-Ab igo
• man old sickness him-shoot-3s.DS lie-SS
  be.3sg.pres
• The old man is sick and lying down
• NB. Experiential verbs similar to ordinary
  transitives, but differ in that Experiencer/Goal
  unlike other objects always in the first topic
  position

19
Reanalysis in Papuan languages Amele

• EOC in Amele similar to Usan
• Amele (Roberts 1987, 315).
• Ija wen øte-na
• 1sg hunger (AUX-)1sg-3sg-PRES
• I am hungry
• But note that V is grammaticalized (phonetically
  zero).
• Apart from (topic) position, the experiencer
  reveals (most) other subject properties
• intraclausal (reflexivization, etc)
• interclausal (control of switch-reference, etc)

20
Conclusion on Papuan languages

• In Papuan languages EOC tend to develop into a
  construction with subject experiencers (cf.
  Roberts 2001 on non-canonical experiencer
  subjects in Amele)
• The Amele pattern where the subject experiencer
  cross-referenced through object AGR is similar to
  an So pattern in a typical split-S language

21
EOC reanalysis beyond split intransitivity I

• Evidence for diachronic instability of the EOC
  constructions
• A-absorption in EOC in Iwadjan. Different degrees
  of grammaticalization/reanalysis (Evans 2004).
• Pattern I. Subcategorized nominal subject
• Nga-ni-ma-ny wunyarru
• 1O-3mA-get-P sickness
• I got sick (lit. sickness got me)
• Here the transitive EOC construction similar to
  the Papuan pattern

22
Experiencer O absorption in Iwadjan II

• Pattern II frozen nominal subject
• Nga-ni-mi-ny ngok
• 1O-3mA-get-P ?
• I am full
• NB the formal subject ngok is not attested
  outside this construction

23
Experiencer O incorporation in Iwadjan III

• Pattern III dummy subject construction
• I-ni-marruku-n
• 3mO-3mA-make.wet-NP
• He is sweating
• This construction is clearly (trans)impersonal
• NB a diachronic instability of the EOC.
  Motivation downgrading/omission of non-prominent
  A.

24
EOC beyond split intransitivity II

• Covert reanalysis of EOC in Germanic
• English please -gt like reanalysis (Jesperson
  1927 Lightfoot 1979, Faarlund 1990)
• ðam cynge licodon peran -gt the king liked pears
• Swedish and German (Seefranz-Montag 1983)
• Det lyckades honom -gt han lyckades I manage
• Mich hungert -gt ich hungere I am hungry
• Motivation for reanalysis upgrading of a
  prominent (animate) O.

25
EOC beyond split intransitivity III

• Reanalysis of EOC/TIC in Himalayan
• Tibetan languages a frequent pattern with
  Goal/Object-experiencers (Cf. Bickel 2003)
• Transimpersonals in Limbu default AGR with
  non-referential A.
• Limbu (van Driem 1987 75)
• Khengha? Moyusi
• They inebriate.3P.3s-gt3ns
• they are drunk (lit. it inebriates them)

26
Reanalysis in Himalayan II Yamphu

• Yamphu (Rutgers 1998 109)
• If experiencer is 3rd p. pattern as EOC
  (experiencer cross-referenced by a transitive
  AGR)
• Wai?m-æ? si-s-w-e?
• thirst-ERG attach-3 -gt3.FCT
• Is he thirsty?
• If experiencer is 1st/2nd p. takes an
  intransitive AGR
• Sag-æ? sis-i?-ma
• hunger-ERG attach-EXPS-1PL
• We were hungry
• NB a split-S system, complicated by a person
  split.
• Motivation for reanalysis upgrading of a
  prominent (1,2 person) experiencers.

27
EOC beyond split-S IV from indefinite A to
(impersonal) passive

• From indefinite A to (impersonal) passive
  (Greenberg 1959 Shibatani 1985 )
• Ainu (Tamura 2000 71 cf. Shibatani 1985)
• Itak-an
• Speak-1pl
• One speaks
• a-e-kóyki na
• In/S2sg/O-scold MOD
• you will be scolded/one will scold you
• NB construction impersonal O is still
  cross-referenced by AGRo.

28
From indefinite A to impersonal passive Ainu

• If an agentive phrase is used, it is clear that
  the indefinite A construction is reanalysed as a
  passive
• Ainu (Tamura 2000 72)
• Unuhu oro wa an-kóyki
• Mother place from Ind/S-scold
• He was scolded by (his) mother
• NB looks like a personal passive, but O has few
  subject properties apart from positional
  (Shibatani 1985 824)

29
Further reanalysis to personal passive Iraqw

• In Iraqw indefinite A construction is used as
  impersonal
• Iraqw (Mous 1992 137, 138)
• ta-na haníis tsati
• IMPS-PAST give.3SM.PAST knives
• They gave knives or Knives were given
• NB also possible with an agent phrase
• Under O topicalization as a personal passive
• ameena ta-n nahhaat
• women(F) IMPS-EXPEC hide-PRES
• Women were hidden/hid themselves

30
From indefinite A to impersonal passive other
languages

• Indefinite (impersonal) passives
• Greenberg (1959) on Maasai,
• Givon (1979) Kimbundu
• Shibatani (1985) on indefinite passives Ainu,
  Trukic, Indonesian
• Motivation for reanalysis downgrading of
  indefinite A (cf. Shibatani on A-defocussing),
  promotes reanalysis to an impersonal structure
  (under O topicalization can develop further to
  personal)

31
Conclusion EOC and TIC in a broader context

• Universal functional pressure for reanalysis of
  EOC and TIC, due to
• syntactic downgrading of non-prominent
  (indefinite, inanimate, cognate) A of TIC
• syntactic upgrading of a prominent (animate) O of
  EOC

32
Functional factours and structural outcome
Split-S

• But these universal functional factors will yield
  a split-S system only under particular structural
  conditions
• AGRo marking
• if AGRo unmarked, more likely covert reanalysis
  (please-gt like).
• AGRs is zero marked
• if AGRs over then rather as extended
  transimpersonal constructions (cf. Eskimo,
  Yamphu), or else reanalysed as a Passive (Ainu,
  Iraqw)

33
Role of the structural factors an illustration

• A consistently ergative language cannot develop
  a split-S structure
• rather experiencer O upgrading will lead to
  formation of (S/O) labile verbs
• (NP/erg) NP/abs V-agr/abs
• Note that this grammatically ambiguous structure,
  allows for covert reanalysis of the ABS-marked
  object-experiencers as subject-experiencers

34
General conclusion

• Unlike the approaches which motivate Split-S
  pattern through role-domination (direct mapping
  from semantic functions to case-marking), I
  regard it as a secondary phenomena which may
  arise through a conspiracy of
• universal functional tendencies
• language particular structural properties

35
A final qualification

• This scenario for the rise of Split-S pattern
  from reanalysis of transitives (transimpersonals,
  experiencer object verbs) applies only for
  languages where
• So is a minor pattern (i.e. Sa-based)
• the split has an agent/patient than
  active/stative basis
• For Split-S languages which are So based (with Sa
  as a deviant pattern) another explanations.
• The latter pattern may also be secondary result
  from reanalysis of a transitive construction with
  a cognate O (cf. Basque, Georgian, etc.)