A new approach to gender and classifier systems: Evidence from Austronesian and Papuan languages
Austronesian and Papuan Languages and Linguis<cs, SOAS, London 14 May 2016
Sebas<an Fedden & Greville G. CorbeL University of Sydney & Surrey Morphology Group
With thanks to the AHRC (UK)
Seman<cs
S P D
Form
S 1 1 2
P 1 ? 2
D 1 2 2
Tim Feist
Outline
• DEFINING GENDER AND CLASSIFIERS: THE USEFULNESS OF CANONICAL TYPOLOGY
• MORPHOSYNTACTIC FEATURES
• ESTABLISHING GENDER VALUES
• MIAN: A TEST CASE
• CONCURRENT SYSTEMS: OUR TYPOLOGY • CONCLUSIONS
2
1. The challenge of defini<on
• Defini<on and descrip<on of nominal categoriza<on systems is s<ll difficult
• Progress needed to make possible meaningful comparison across languages
3
... and specifically
• Some familiar systems show peculiar combina<ons of elements
• These combina<ons are not necessary nor should they be defini<onal
(Whether our favourite system is found in Archi, Cree, La<n, Tidore or Zulu, we need a broad view.)
4
Canonical Typology
• Examine clusterings of proper<es (especially the problema<c ones): meaningful or coincidental?
• Draw out the theore<cal space to tease apart the clusters
• Use the canonical ideal as a baseline from which we measure the actual examples we find
CorbeL (2007, 2015); Brown, Chumakina & CorbeL (2013). And see the bibliography at: hLp://www.surrey.ac.uk/englishandlanguages/research/smg/canonicaltypology/bibliography/index.htm. Recent developments: phonology (Hyman 2012), phonaesthemes (Kwon & Round 2015)
5
Canonical Typology
• The canonical ideal is defined – from a specific set of criteria
– that can be used to project the theore<cal space – which is then populated by real instances
• No<onal point where all criteria converge
• Different from “typical”, “frequent” or “prototypical”
6
Relevance to gender and classifiers
• Construing the two as extremes of an opposi<on has not proved fruikul
• So we aLempt a defini<on of one extreme, namely canonical gender (CorbeL & Fedden 2017)
• and we try it out on an intriguing language, Mian
8
2. Canonical morphosyntac<c features
• The canonicity criteria are largely the same for: – gender – number – person – case
CorbeL (2007, 2012: 156-‐199)
9
Principles for canonical features and their values
I: Canonical features and their values are clearly dis<nguished by formal means.
II: The use of canonical morphosyntac<c features and
their values is determined by simple syntac<c rules. III: Canonical morphosyntac<c features and their values
are expressed by canonical inflec<onal morphology.
(CorbeL 2012: 155-‐199) 10
Canonical morphosyntac<c features
• Logical problem: If morphosyntac<c features were fully canonical, they would be iden<cal, no way of telling them apart
• Solu<on: examine the interac<on of morphosyntac<c features with canonical parts of speech
• In canonical parts of speech (CorbeL 2013: 52; Spencer 2005: 102) the seman<cs, syntax and morphology align
• For example, a canonical noun would – denote an en<ty – head a nominal phrase – take the appropriate inflec<onal morphology
11
Canonical parts of speech and canonical morphosyntac<c features
Feature: F1 F2
Value: v1 v2 v11 v12
Lexicon: l1 l2 l11 l12
Part of Speech: PoS1 PoS2
12 CorbeL (2013)
Four criteria: C-‐1. exclusiveness: • a lexical item belongs to just one part of speech • a value belongs to just one feature CorbeL (2013)
Canonical parts of speech and canonical morphosyntac<c features
13
Criterion exhaus6veness: • every lexical item of every part of speech has available all values of all features
• (alterna<vely: every feature value applies to all lexical items)
(Thus if there is a number feature with the values singular and plural, then in the canonical situa<on every lexeme would have singular and plural available to it.)
14
Canonical parts of speech and canonical morphosyntac<c features
C-‐3. open and closed: • All classes are closed, except the class of lexical items.
(Canonical parts of speech have open membership.)
Canonical parts of speech and canonical morphosyntac<c features (3)
15
C-‐4. composi6onality: • given the lexical seman<cs of a lexical item and a specifica<on of its feature values, the meaning of the whole is fully predictable.
16
Canonical parts of speech and canonical morphosyntac<c features (4)
Canonical parts of speech and canonical morphosyntac<c features: weakening
Feature 1: F1 F2
Value: v1 v2 v11 v12
Lexicon: l1 l2 l11 l12
Part of Speech: PoS1 PoS2 Part of Speech: PoS1 (e.g. noun) PoS2 (e.g. adjective)
17
Canonical parts of speech and canonical morphosyntac<c features
• Weakening of canonical exhaus<veness: nouns select one of the values (gender)
• Gender is not a fully canonical morphosyntac<c feature, since controllers do not have all values available
• but it is the canonical agreement feature, for this same reason, since it is a lexical feature
18
3. Agreement classes
A set of nouns is an agreement class if and only if all members of that set have the property that whenever (i) they have the same morphosyntac<c specifica<on and (ii) they occur in the same agreement domain and (iii) they have the same lexical item as agreement target
then their agreement targets have the same morphological realiza<on.
19
Relevance of Agreement Classes
• Sources: Zaliznjak (1964), CorbeL (2012: 80-‐85) • The basic idea: nouns are in the same agreement class iff – given the same condi<ons – they control the same agreement
Canonical Gender – Criterion 1
Canonical gender values match agreement classes (CorbeL and Fedden 2017[2015])
20
Non-‐canonical phenomena
• Phenomena which linguists have named (common nouns, hybrid nouns, sub-‐genders, etc.) can be understood as devia<ons from the canonical ideal
• For Mian, we will discuss in turn: – non-‐canonical controllers – non-‐canonical targets – non-‐canonical domains
21
4. Why Mian?
• Rela<on of gender and classifiers: – an opposi<on? – part of the same space?
• Gender and classifiers are oren treated as mutually exclusive systems of categoriza<on
• Mian has both (Fedden 2011)
22
Mian
• Ok family (ok ‘water, river’) (Healey 1964) • Trans New Guinea (Pawley 2005; Ross 2005; Wurm 1982)
• Telefomin District, Sandaun Province, Papua New Guinea
• Two dialects – Eastern dialect has around 1,400 speakers (Fedden 2011) – Western dialect (aka Suganga) has around 350 speakers
23
4.1 “Gender” in Mian
• M, F, N1, N2 in the 3rd person singular – Targets: free pronouns, ar<cles, demonstra<ves, verb – All finite verbs have subject agreement – Object agreement with these only:
-‐e ‘hit, kill (IPFV)’ -‐fû’ ‘grab (PFV)’ -‐lò ‘hit, kill (PFV)’ -‐nâ’ ‘hit, kill (PFV)’ -‐ntamâ’ ‘bite (PFV)’ -‐têm’ ‘see (PFV)’ -‐temê’ ‘look at (IPFV)’
25
“Gender” in Mian
(1) ē unáng=o wa-‐têm’-‐Ø-‐e=be 3SG.M woman(F)=ART.SG.F 3SG.F.OBJ-‐see.PFV-‐REAL-‐3SG.M.SBJ=DECL ‘He sees the woman.’ (SF field notes) (2) ō naka=e a-‐têm’-‐Ø-‐o=be 3SG.F man(M)=ART.SG.M 3SG.M.OBJ-‐see.PFV-‐REAL-‐3SG.F.SBJ=DECL ‘She sees the man.’ (SF field notes)
26
“Gender” in Mian
(1) ē unáng=o wa-‐têm’-‐Ø-‐e=be 3SG.M woman(F)=ART.SG.F 3SG.F.OBJ-‐see.PFV-‐REAL-‐3SG.M.SBJ=DECL ‘He sees the woman.’ (SF field notes) (2) ō naka=e a-‐têm’-‐Ø-‐o=be 3SG.F man(M)=ART.SG.M 3SG.M.OBJ-‐see.PFV-‐REAL-‐3SG.F.SBJ=DECL ‘She sees the man.’ (SF field notes)
27
“Gender” in Mian
(1) ē unáng=o wa-‐têm’-‐Ø-‐e=be 3SG.M woman(F)=ART.SG.F 3SG.F.OBJ-‐see.PFV-‐REAL-‐3SG.M.SBJ=DECL ‘He sees the woman.’ (SF field notes) (2) ō naka=e a-‐têm’-‐Ø-‐o=be 3SG.F man(M)=ART.SG.M 3SG.M.OBJ-‐see.PFV-‐REAL-‐3SG.F.SBJ=DECL ‘She sees the man.’ (SF field notes)
28
Agreement target: ar<cle
a. naka=e ‘a/the man’ naka=i ‘(the) men’ b. unáng=o ‘a/the woman’ unáng=i ‘(the) women’ c. tóm=e ‘a/the stone’ tóm=o ‘(the) stones’ d. káawa=o ‘a/the steel axe’ káawa=o ‘(the) steel axes’
29
Mian gender values SINGULAR PLURAL assignment
MASCULINE =e =i males FEMININE =o =i females NEUTER 1 =e =o inanimates NEUTER 2 =o =o inanimates: loca<ons, body
decora<on, weather phenomena, illnesses, abstract nouns, some tools and weapons
• All Mian genders are non-‐autonomous values (Zaliznjak 1973[2002]: 69-‐74)
• Assignment is predominantly seman<c
30
4.2 “Verbal classifiers” in Mian
(3) nē memâalo fút=e 1SG now tobacco=SG.N1 tob-‐ò-‐n-‐i=a 3SG.LONG.OBJ-‐take.PFV-‐SS.SEQ-‐1SG.SBJ=MED ‘Now I take the tobacco leaf and then I ...’ (Rolling smokes)
Dried Nico9ana leaves bound together
Source: SF fieldnotes
31
“Verbal classifiers” in Mian
• Prefixed to verbs of object handling and movement, e.g. ‘give’, ‘take’, ‘put’, ‘throw’, ‘lir’, ‘turn’, ‘fall’ (about 50 verbs)
• Absolu<ve basis
32
“Verbal classifiers” in Mian SINGULAR PLURAL assignment
M-‐CLASSIFIER dob-‐ dol-‐ males (M); plate, clothes, mosquito net, some bananas, some pandanus (N1)
F-‐CLASSIFIER om-‐ dol-‐ females (F); inanimates: all nouns of N2 gender
LONG tob-‐ tebel-‐ inanimates: arrow, pen, tobacco leaf, bone, tongs, bush knife, nail, belt (all N1)
COVERING gam-‐ gemel-‐ skin, palm bark, blanket (all N1) BUNDLE gol-‐ gulel-‐ string bag, bundles (all N1) RESIDUE ob-‐ ol-‐ tortoises (F); rest of inanimates (N1)
Strongly based on seman<cs, BUT assignment is less seman<cally transparent than in the gender system 33
Controllers
• In a canonical gender system, each controller has a single gender value
• Source of non-‐canonicity: Common gender nouns, e.g. Mian éil ‘pig’, =l ‘dog’, and about a dozen others – wan=e do-‐fâ! ‘Put down the male bird!’ – wan=o om-‐fâ! ‘Put down the female bird!’
=> This affects both systems in a similar way
35
Assignment
• The canonical gender assignment rule is seman<c
• Gender can be read off the lexical informa<on (i.e. the seman<c specifica<on of the noun)
• Principle of Canonical Typology: In canonical systems everything lines up
=> THE GENDER SYSTEM IS CLOSER TO THE CANONICAL IDEAL
36
Assignment
• This goes against a widely held belief that gender “should” be opaque; but that’s when it’s easier to spot, but not when it’s canonical
37
Recategoriza<on
(4) som=e banana_bunch=SG.N1
dob-‐meki-‐n-‐e=be 3SG.M_CL.OBJ-‐hang_up.PFV-‐REAL-‐3SG.M.SBJ=DECL ‘I've hung up the banana bunch.’
38
(5) som=e banana_bunch=SG.N1
gol-‐meki-‐n-‐e=be 3SG.BDL.OBJ-‐hang_up.PFV-‐REAL-‐3SG.M.SBJ=DECL
‘I've hung up the banana bunch (that is supplied with some string or rope to hang it up on).’
Recategoriza<on
• In a canonical gender system, nouns select one value • Recategoriza<on -‐ source of non-‐canonicity
MIAN • Gender system doesn’t allow this • Classifier system does
=> the gender system is closer to the canonical ideal
39
Targets
• Canonical features and their values are dis<nguished consistently across relevant parts of speech
• Intui<on: The more evidence for gender the more canonical the system (“the more targets the beLer”)
MIAN • Gender system targets: ar<cle, verb, pronoun • Classifier system target: verb (subset only)
=> the gender system is closer to the canonical ideal
40
Syntac<c domains
• Related to the intui<on about targets is the following: “The more syntac<c domains the beLer.”
• Gender system domains: NP (ar<cle), clause (verb), extra-‐clausal (free personal pronoun)
41
Syntac<c domains (6) Futaman mín=e baa-‐n-‐e=o=le
Fu.valley son=ART.SG.M say.PFV-‐SEQ-‐3SG.M.SBJ=N2=TOP ībo wan-‐ībt=e 2PL who-‐2/3PL.AN.EMPH=CQ ge baa-‐s-‐e=ta [intervening material] say.PFV say.PFV-‐DS.SEQ-‐3SG.M.SBJ=and […] ē as=o hà’-‐n-‐e=o=le […] 3SG.M firewood=ART.N2 break.PFV-‐REAL-‐3SG.M.SBJ=N2=TOP […] ‘The man from the Fu river valley said, “Who are you?” and … when he [i.e. the man from the Fu river valley] cut the firewood …’ (Fedden 2011: 530)
ē 3SG.M
42
Syntac<c domains
• Gender system domains: NP (ar<cle), clause (verb), extra-‐clausal (free personal pronoun)
• Classifier system domain: clause (verb), no system of free personal pronouns
=> THE GENDER SYSTEM IS CLOSER TO THE CANONICAL IDEAL
43
Inequality of values
• In any canonical situa<on, mappings are one-‐to-‐one. Specifically for gender, controller genders map to target genders one-‐to-‐one
• Any discrepancy here is a source of non-‐canonicity
44
Mian SG PL
M A C
F B C
N1 A B
N2 B B
Andi SG PL
I A A
II B B
III C C
IV D D
e o
i o
SINGULAR PLURAL
N2
F
M
N1
Target genders in Mian
CorbeL (1991: 198)
SINGULAR PLURAL Assignment MASCULINE =e =i Males FEMININE =o =i Females NEUTER 1 =e =o Inanimates NEUTER 2 =o =o Inanimates: loca<ons, etc.
45
Inequality of values
• Mismatch between controller and target genders
• No such mismatch in the classifier system => THE CLASSIFIER SYSTEM IS CLOSER TO THE CANONICAL IDEAL
46
Inequality of values
G C
G
C
G C
G
C
Common gender
Assignment
Recategoriza<on
Targets
G
C Domains
G C Inequality of values
52
Summary • Specifically on Mian:
– a fairly canonical gender system – a second system further away from the canonical ideal (but not
for all criteria)
• More generally: – noun categoriza<on systems are not discrete – they have components which are more or less close to a
canonical ideal
• Canonical approach – makes possible a typology which can incorporate the
intermediate cases we find in the languages of the world
53
5. Concurrent systems
• Challenge of languages with two or more systems of nominal classifica<on
• Accounts oren simply assume that a given language indeed has two or more classifica<on systems
• To make progress, we need to step back and show that a language really has dis<nct systems
... but this ques<on is typically not even raised some excep<ons: • Dahl (2000) • CorbeL (2012: 174-‐185) • Round & CorbeL (2015)
54
Our sample
• Contains some 20 languages • Larger than some linguists would have imagined • Favouring languages which show at least prima facie evidence for two systems
55
Key ideas in our typology
(i) the degree to which the seman<cs of the two (candidate) systems are orthogonal to each other
(ii) the degree to which their means of realiza<on
are dis<nct
Note: seman<cs = gramma<cal meanings
56
Dialect A1: Canonically one system
• Seman<cs: same
(7) un-‐i sester simpa<c-‐i (8) un-‐e cos simpa<c-‐e one-‐ANIM sister nice-‐ANIM one-‐INAN thing nice-‐INAN ‘one nice sister’ ‘one nice thing’
57
; form: same
i i e e
Dialect C3: Canonically two systems
• Seman<cs: different
(9) un-‐i sester simpa<c-‐u (10) un-‐e cos simpa<c-‐u one-‐ANIM sister nice-‐AUG one-‐INAN thing nice-‐AUG
‘one nice (big) sister’ ‘one nice (big) thing’ (11) un-‐i sester simpa<c-‐o (12) un-‐e cos simpa<c-‐o one-‐ANIM sister nice-‐DIM one-‐INAN thing nice-‐DIM ‘one nice (small) sister’ ‘one nice (small) thing’
58
i
i
e
e
u u
; form: different
o o
Nine types (Fedden and CorbeL (under review))
Seman<cs
S P D
Form
S A1 B1 C1
P A2 B2 C2
D A3 B3 C3
S = same P = par<al overlap D = different
Tim Feist
59
How many systems?
Seman<cs
S P D
Form
S 1
P
D
Same seman<cs + same forms ! canonical single system
60
How many systems?
Seman<cs
S P D
Form
S 1
P
D 2 Different seman<cs + different forms ! canonical concurrent systems
61
How many systems?
Seman<cs
S P D
Form
S 1 2
P 1 2
D 1 2
Regardless of form, different seman<cs ! 2 systems
63
How many systems?
Seman<cs
S P D
Form
S 1 2
P 1 2
D 1 2
Given a par<al overlap in seman<cs, the forms are decisive
64
How many systems?
Seman<cs
S P D
Form
S 1 1 2
P 1 2
D 1 2
Given a par<al overlap in the seman<cs and one set of forms ! 1 system
65
How many systems?
Seman<cs
S P D
Form
S 1 1 2
P 1 2
D 1 2 2
Given a par<al overlap in the seman<cs and 2 sets of forms ! 2 systems
66
How many systems?
Seman<cs
S P D
Form
S 1 1 2
P 1 ? 2
D 1 2 2
But what if we have a par<al overlap in the seman<cs and a par<al overlap in terms of forms? 67
How many systems?
Seman<cs
S P D
Form
S 1 1 2
P 1 ? 2
D 1 2 2
Regardless of form, same seman<cs ! 1 system
Regardless of form, different seman<cs !2 systems
Given a par<al overlap in seman<cs, the forms are decisive
68
KILIVILA Canonically ONE system – Same seman<cs & same forms (our Type A1)
• Austronesian (Papuan Tip Cluster) • Trobriand Islands • 25,000 speakers • Senr (1986, 1996, p.c.)
69
Kilivila
(13) Kilivila (Senr 1986: 69)
mi-‐na-‐si-‐na na-‐yu na-‐manabweta vivila
DEM-‐FEMALE-‐PL-‐DEM FEMALE-‐two FEMALE-‐beau<ful girl
‘these two beau<ful girls’
70
Kilivila: Type A1
• Targets – numerals – adjec<ves (e.g. -‐manabweta ‘beau<ful’) – demonstra<ves (all but one) – one interroga<ve
• Almost always iden<cal forms comparing across targets
• Single system • An almost perfect example of Type A1, a canonically one-‐system language
71
Other Type A1 languages
• Lamnso (Grassfields, Southern Bantu; McGarrity & Botne 2001)
• Ngan'gityemerri (Southern Daly; Reid 1997) • Bora-‐Miraña (Witotoan; Seifart 2005, 2009)
72
MALI
Same seman<cs & par<al overlap in forms (our Type A2)
73
• Baining (Papuan) • New Britain (PNG) • 2,200 speakers • Stebbins (2005, 2011)
Mali (candidate 1)
• Noun classes: MASCULINE, FEMININE, COUNT NEUTRAL, DIMINUTIVE, REDUCED, FLAT, EXCISED, LONG, EXTENDED
• Targets: – anaphoric pronouns – adjec<ves – numerals (1-‐3) – indefinite pronouns – contras<ve pronouns – demonstra<ves
76
Noun class assignment
• Seman<c assignment for humans and animals • Residue distributed across all noun classes
– Assignment of masculine, feminine and count neutral governed by complex principles (Stebbins 2005: 92-‐108)
– Remaining five noun classes based on a single seman<c criterion each (i.e. diminu<ve, flat, cut-‐off, long or extended)
77
Mali (candidate 2)
• Gender: MASCULINE, FEMININE AND NEUTER • Targets:
– possessive pronouns – verb
• Seman<c assignment
78
Example of both candidate systems
(14) Mali (Stebbins 2011: 65) kama pepavēs ma asēgēvēs nge pe ART paper.FLAT.SG REL one.FLAT.SG 3N.PRS there ‘there is a piece of paper there’
79
Mali system matrix
! MASCULINE! FEMININE! NEUTER!MASCULINE! matka!‘older!brother’! ;! ;!FEMININE! ;! likki!‘younger!sister’! ;!COUNT!NEUTRAL! ;! ;! bang!‘!house’!DIMINUTIVE! ;! ;! uratini!‘small!basket’!!REDUCED! ;! ;! srēvēm&‘dwarf’&!FLAT! ;! ;! tēlēngves)‘leaf’!EXCISED! ;! ;! amēngigl(‘board’(!LONG! ;! ;! suchulvet!‘slender!post’!EXTENDED! ;! ;! aupia!‘valley’!!
81
One system collapses onto the other. Mali has nine noun classes, and exactly nine cells are filled
Many-‐to-‐one mapping
• Extended no<on of ‘same seman<cs’
• English´ example of many-‐to-‐one mapping: – pronouns: he for male humans, she for female humans, it for the residue
– only rela<ve pronouns: who for humans only, which for the residue
• On the scale same-‐different, this counts as prac<cally the same
82
Mali: Type A2
• Forms on respec<ve set of targets largely dis<nct, but some overlap – for both sets of targets masculine is marked by /ka/ and the masculine plural (for humans) is /ta/ (Stebbins 2011: 44-‐46, 137)
• Type A2 – Seman<cs: same (noun classes collapse onto gender) – Forms: par<al overlap
83
Other Type A2 languages
• Tatuyo (Tucanoan; Gomez-‐Imbert 2007) • Lao (Tai-‐Kadai; Enfield 2004, 2007; Fedden and CorbeL 2016)
• La<n
82
FRENCH Same seman<cs & different forms (our Type A3)
• Indo-‐European (Romance) • France, Canada, Switzerland, etc. • 75 million speakers
83
PAUMARÍ Canonically TWO systems: Different seman<cs & different forms (our Type C3)
• Arawan • Brazil • around 300 speakers • Chapman & Derbyshire (1991);
Aikhenvald (2010)
84
Paumarí (candidate 1)
• Gender system: MASCULINE vs FEMININE • Targets:
– demonstra<ves (ada ‘this (M)’, ida ‘this (F)’) – some adjec<ves – inalienably possessed nouns – verbs
85
Paumarí (candidate 1)
Possessed noun Adjec<ve Verb gender + dependent intransi<ve
gender + theme
gender + immediate
MASCULINE -‐na~-‐Ø/-‐i -‐na -‐ha -‐ra FEMININE -‐ni -‐hi -‐ja/-‐a
Paumarí gender agreement suffixes (based on Aikhenvald 2010; Chapman & Derbyshire 1991)
(17) Paumarí (Aikhenvald 2010; also cited in Aikhenvald 2000: 71) Voroni-‐’a-‐ha ada kahami. fall-‐ASP-‐TH.M this.M palm.tree(M, NON-‐KA) ‘The palm tree fell down.’
86
Paumarí (candidate 1)
• Seman<c core to gender assignment • Certain types of nouns are clearly associated with a par<cular gender value
• Others are more evenly distributed across the two genders
87
Paumarí (candidate 2)
• ka-‐ vs non-‐ka; prefix ka-‐ (some<mes a-‐, ko-‐ or ki-‐) vs no overt marker
• Targets – inalienably possessed nouns – those adjec<ves that show masculine / feminine agreement – verbs
88
Paumarí (candidate 2)
(18) Paumarí (Aikhenvald 2000: 71) Ka-‐voroni-‐’a-‐hi ida ojoro. KA-‐fall-‐ASP-‐TH.F this.F turtle(F, KA) ‘The turtle fell down.’
Assignment is complex
– no nouns deno<ng humans in the ka-‐ class – non-‐ka-‐class contains humans and non-‐humans
89
Paumarí: Two systems (Type C3)
• Masculine-‐feminine gender system PLUS human-‐non-‐human gender system
• Both seman<cs and forms are different • Example of Type C3, i.e. a canonically two-‐system language
90
Paumarí: Two systems (Type C3)
• No predictability of values from one system to the other
• The two systems are independent in that all four combina<ons are aLested (CorbeL 2007: 257)
MASCULINE FEMININE
ka-‐ kasi’i ‘crocodile’ ojoro ‘turtle’ non-‐ka-‐ kahami ‘palm tree’ arabo ‘land, ground’
Interac<on of systems in Paumarí
91
Other Type C3 languages
• Michif (mixed language of Canada; Bakker & Papen 1997; Bakker 1997)
• Ayoreo (Zamucoan; Ber<neLo 2009; Ber<neLo & Ciucci 2014)
92
RUSSIAN Different seman<cs & same form (our Type C1 )
93
• Indo-‐European (Slavonic) • Russia, etc. • 170 million speakers
NOT YET FOUND Different seman<cs & par<al overlap in forms (our Type C2)
94
• More plausible than Type 7 (different seman<cs & same forms), which exists!
TOBELO Different seman<cs & par<al overlap in forms (our Type C2)
95
• West Papuan • North Halmahera, Indonesia • 30,000 speakers • Holton (2014, p.c.)
Tobelo (candidate 1)
• Gender system: MASCULINE vs FEMININE in the third singular, and HUMAN vs NON-‐HUMAN in the third plural
• Target – verb
96
Tobelo (candidate 2)
• System of 16 numeral classifiers primarily for physical objects (no classifier for humans!)
• Plus an ‘incipient human/non-‐human dis<nc<on’, whereby
‘[n]umeral predicates with human arguments employ the portmanteau pronominal prefix ya-‐, which derives from the third person human plural actor prefix yo-‐ and the third person undergoer prefix a-‐’ (Holton 2014: 94)
97
Tobelo: (almost) Type C2
• Combina<on of yo-‐ with the undergoer prefix means that it is not a clear case of a Type C2
• but it suggests that Type C2 may well occur
98
PNAR Par<al overlap in the seman<cs & different forms (our Type B3)
99
• Austroasia<c • North Eastern India • approx. 400,000 speakers • Ring (2015)
Pnar (candidate 1)
• Gender system : MASCULINE vs FEMININE vs NEUTER • Targets
– free pronouns – demonstra<ves – numerals – rela<ve clauses
• Assignment – animates according to biological sex – inanimates in all three genders, nouns referring to abstract en<<es are consistently neuter
100
Pnar (candidate 1)
(22) Pnar (Ring 2015: 320) εm jap ka=wi ka=knthaj tmmεn have die F=one F=female be.old ‘an old woman died’
101
Pnar (candidate 2)
• Two numeral classifiers: ŋut ‘(living) human’ vs tļli ‘non-‐human’
• Classifiers are free forms and obligatory
102
Pnar: Type B3
Pnar system matrix (3x2, 5 actual combina<ons)
• Systems largely orthogonal
Classifiers HUMAN NON-‐HUMAN
Gender
MASCULINE + + FEMININE + + NEUTER -‐ +
103
Orthogonality score
• Measure of orthogonality
(cells filled – minimum cells filled) (possible cells – minimum cells filled) = (5 – 3) = 2 = .67 (6 – 3) 3
So Pnar is an example of Type B3 (with a higher orthogonality score) 104
NANTI Par<al overlap in the seman<cs & different forms (our Type B3)
105
• Kampan branch of Arawakan
• Peru • 450 speakers • Michael (2008)
Nan< (candidate 1)
• Gender: -‐n ‘animate’ vs. -‐t ‘inanimate’ • Assignment is seman<c • Agreement targets
– subset of adjec<ves – numerals – quan<fiers – existen<al verb
106
Nan< (candidate 2)
• Gender (second system): i-‐ ‘masculine’ vs. o-‐ ‘feminine’
• Agreement targets – verb – possessed nouns – pronouns – demonstra<ves – very small subset of adjec<ves
• A few dimensional adjec<ves (e.g. ‘large’) par<cipate in both systems
107
Nan< (candidate 2)
• Assignment – Humans and animals assigned according to sex – All inanimates are feminine
108
Nan<: Type B3
• Nan< system matrix (2x2, 3 actual combina<ons)
• Seman9cs: mid-‐way between canonically one system (2 cells filled) and canonically two systems (4 cells filled)
• Orthogonality score: (3-‐2)/(4-‐2) = .5 • Forms are different => two systems • Example of Type B3, with a mid orthogonality score
ANIMATE INANIMATE MASCULINE + -‐ FEMININE + +
109
MIAN Par<al overlap in the seman<cs & different forms (our Type B3)
110
• Trans New Guinea, Ok • Papua New Guinea • 1750 speakers • Fedden (2011)
MASCULINE FEMININE NEUTER 1 NEUTER 2
M-‐CLASSIFIER man, boy, boar — sleeping bag, plate, mosquito net
—
F-‐CLASSIFIER — woman, girl, sow — house, steel axe, money (kina note)
LONG — — tobacco, ea<ng implement, bush knife
—
BUNDLE — — string bag (large), tobacco pouch, plas<c bag
—
COVERING — — blanket, band aid —
RESIDUE — tortoise, scorpion cassowary egg, plane, hat
—
Mian: Type B3
• Mian system matrix (6x4, 9 actual combina<ons)
112
Mian: Two systems (Type B3)
• Seman<cs show considerable overlap Orthogonality score for Mian: (9-‐6)/(24-‐6)=.17
• Forms are different
• Example of Type B3, with a low orthogonality score
113
Other Type B3 languages
• Mba (Tucker & Bryan 1966; Pasch 1985, 1986; CorbeL 2012)
• Tariana (Arawakan; Aikhenvald 1994, 2003) • Baniwa of Içana (Arawakan, Aikhenvald 2007) • Yagua (Peba-‐Yaguan; Payne 1986, 2007; Payne & Payne 1990)
• Innu (Algonquian; Drapeau & Lambert-‐Bré<ère 2011) • Khasi (Austroasia<c; Rabel-‐Heymann 1977; Temsen, ms.)
114
BURMESO Par<al overlap in seman<cs & par<al overlap in forms (our Type B2)
115
• East Bird’s Head • West New Guinea • 250 speakers • Donohue (2001), CorbeL
(2012)
Burmeso (candidate 2)
• Gender (second system): six different values (MASCULINE, FEMININE, NEUTER, MASCULINE INANIMATE, FEMININE INANIMATE, NEUTER ANIMATE)
• Target – adjec<ve
117
Burmeso: Type B2
• Burmeso system matrix (6x6, 16 actual combina<ons) agreement on adjec<ve
agreem
ent o
n verb
MASCULINE FEMININE NEUTER MASCULINE INANIMATE
FEMININE INANIMATE NEUTER ANIMATE
I 44 plus all male kin terms
5 (4 birds) 1 (‘neck’) 2 (‘sea’, ‘wound’)
II 7 plus all female kin terms
4 1 (‘small goanna’)
2 (‘sago rinser (lower)’, ‘string.shapes’)
III 3 28, mainly inanimate 10, inanimate 1 (‘goanna’)
IV 9, inanimate
V 2 (‘banana’, ‘sago tree’)
VI 1 (‘arrow’) 1 (‘coconut’)
118
Burmeso: Type B2
• Orthogonality scores: (16-‐6)/(36-‐6) = .33 • Par<al overlap in forms
– Donohue (2001: 105) points out ‘the strong resemblances between the forms of the gender suffixes and the set II verbal class agreement prefixes’
• Example of Type B2 • Right in the middle of our canonical typology!
119
BAGVALAL Par<al overlap in seman<cs & same forms (our Type B1)
120
• Nakh-‐Dagestanian • Dagestan • 1,500 speakers • Kibrik (2001)
ALested types
Seman<cs
S P D
Form
S A1
Kilivila B1
Bagvalal C1
Russian
P A2 Mali
B2 Burmeso
C2 {Tobelo}
D A3 French
B3 Mian
C3 Paumarí
S = same P = par<al overlap D = different
121
6. Conclusions • Specifically on Mian
– a fairly canonical gender system – a second system further away from the canonical ideal (but not
for all criteria)
• More generally – noun categoriza<on systems are not discrete – they have components which are more or less close to a
canonical ideal
• Canonical approach – makes possible a typology which can incorporate the
intermediate cases we find in the languages of the world
122
Conclusions
• Nominal classifica<on systems – there are languages with more than one system – our typology can incorporate phenomena previously called “genders” and “classifiers”, and the intermediate cases we find in the languages of the world
– we have offered a typology of concurrent systems of nominal classifica<on
123
References Aikhenvald, Alexandra Y. 1994. Classifiers in Tariana. Anthropological Linguis9cs 36: 407-‐465. Aikhenvald, Alexandra Y. 2000. Classifiers: A Typology of Noun Categoriza9on Devices. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. 2003. A grammar of Tariana, from Northwest Amazonia. Cambridge Gramma<cal Descrip<ons. Cambridge: Cambridge University Press.
Aikhenvald, Alexandra Y. 2007. Classifiers in mul<ple environments: Baniwa of Içana/Kurripako -‐ A north Arawak perspec<ve. Interna9onal Journal of American Linguis9cs 73(4): 475-‐500.
Aikhenvald, Alexandra Y. 2010. Gender, noun class and language obsolescence: The case of Paumarí. In Eithne B. Carlin & Simon van de Kerke (eds.), Linguis9cs and Archeology in the Americas: The Historiza9on of Language and Society, 235-‐252. Leiden: Brill.
Bakker, Peter. 1997. A Language of Our Own: The Genesis of Michif, the Mixed Cree-‐French Language of the Canadian Mé9s. Oxford: Oxford University Press.
Bakker, Peter & Robert A. Papen. 1997. Michif: A mixed language based on Cree and French. In Sarah G. Thomason (ed.), Contact languages: A wider perspec9ve (Creole Language Library Volume 17), 295-‐363. Amsterdam: John Benjamins.
Ber<neLo, Pier Marco. 2009. Ayoreo (Zamuco). A gramma<cal sketch. Quaderni del laboratorio di Linguis9ca 8 n.s. [Available at hLp://linguis<ca.sns.it/QLL/QLL09.htm]
Ber<neLo, Pier Marco & Luca Ciucci. 2014. Possessive classifiers in gender-‐marking Ayoreo and Chamacoco (Zamucoan), with an overview on the Chaco linguis<c area. Paper presented at the Workshop “Gender and classifiers: Cross-‐linguis<c Perspec<ves” in connec<on with the AHRC project “Combining gender and classifiers in natural language (AH/K003194/1)”, University of Surrey, 17 January 2014.
References Chapman, Shirley & Desmond C. Derbyshire. 1991. Paumari. In: Desmond C. Derbyshire & Geoffrey K. Pullum (eds), Handbook of Amazonian Languages III, 161-‐352. Berlin: Mouton de Gruyter.
CorbeL, Greville G. 1991. Gender. Cambridge: Cambridge University Press CorbeL, Greville G. 2007. Gender and noun classes. In Timothy Shopen (ed.), Language Typology and Syntac9c Descrip9on: III: Gramma9cal categories and the lexicon (2nd edi<on), 241-‐279. Cambridge: Cambridge University Press.
CorbeL, Greville G. 2012. Features. Cambridge: Cambridge University Press CorbeL, Greville G. 2013. Canonical morphosyntac<c features. In: Dunstan Brown, Marina Chumakina & Greville G. CorbeL (eds), Canonical Morphology and Syntax, 48-‐65. Oxford: Oxford University Press.
CorbeL, Greville G. 2015. Morphosyntac<c complexity: a typology of lexical splits. Language 91.145-‐193
CorbeL, Greville G. & Sebas<an Fedden. 2017. Canonical gender. To appear in Journal of Linguis9cs. 53.1 (January 2017)
Dahl, Östen. 2000. Elementary gender dis<nc<ons. In: Barbara Unterbeck, Ma� Rissanen, TerLu Nevalainen & Mirja Saari (eds.) Gender in Grammar and Cogni9on: I: Approaches to Gender, II: Manifesta9ons of Gender (Trends in Linguis<cs: Studies and Monographs 124), 577-‐593. Berlin: Mouton de Gruyter.
Donohue, Mark. 2001. Animacy, class and gender in Burmeso. In: Andrew Pawley, Malcolm Ross & Darrell Tryon (eds.), The boy from Bundaberg: Studies in Melanesian linguis9cs in honour of Tom Du^on (Pacific linguis<cs 514), 97-‐115. Canberra: Pacific Linguis<cs.
Drapeau, Lynn & Renée Lambert-‐Bré<ère. 2011. Verbal Classifiers in Innu. Anthropological Linguis9cs 53(4).293-‐322.
References Enfield, N. J. 2004. Nominal classifica<on in Lao: a sketch. Sprachtypologie und Universalienforschung
(STUF) 57(2/3), 117-‐143. Enfield, N. J. 2007. A grammar of Lao. [Mouton Grammar Library 38]. Berlin: Mouton de Gruyter. Fedden, Sebas<an. 2011. A Grammar of Mian. (MGL 55). Berlin: De Gruyter Mouton. Fedden, Sebas<an & Greville G. CorbeL (under review), Gender and classifiers as concurrent
systems: a first typology, submiLed to Language. Fedden, Sebas<an & Greville G. CorbeL. 2016. Understanding intra-‐system dependencies: classifiers in
Lao. In N.J. Enfield. Dependencies in language. Berlin: Language Science Press. Gomez-‐Imbert, Elsa. 2007. Tukanoan nominal classifica<on: The Tatuyo system. In Leo Wetzels (ed.),
Language Endangerment and Endangered Languages: Linguis9c and Anthropological Studies with Special Emphasis on the Languages and Cultures of the Andean-‐Amazonian Border Area, 401-‐428. Leiden: Research School of Asian, African and Amerindian Studies (CNWS), Universiteit Leiden.
Healey, Alan. 1964. A survey of the Ok family of languages, reconstruc<ng Proto-‐Ok. Canberra: Australian Na<onal University PhD disserta<on.
Holton, Gary. 2014. Numeral classifiers and number in two Papuan outliers of East Nusantara. In: Marian Klamer & Fran<šek Kratochvíl (eds) Number and quan9ty in East Nusantara (papers from 12-‐ICAL, Volume 1), 83-‐102. hLp://hdl.handle.net/1885/11917.
Hyman, Larry M. 2012. In defense of Prosodic Typology: A Response to Beckman & Vendi�. Linguis9c Typology 16.341-‐385.
Kibrik, Aleksandr E. (ed.) 2001. Bagvalinskij jazyk: Gramma9ka: Teksty: Slovari. Moscow: Nasledie. [co‑editors K. I. Kazenin, E. A. Lju<kova & S. G. Tatevosov.]
Kwon, Nahyun & Erich R. Round. 2015. Phonaesthemes in morphological theory. Morphology 25.1-‐27. doi: 10.1007/s11525-‐014-‐9250-‐z
References McGarrity, Laura W. & Robert Botne. 2001. Between agreement and case marking in Lamnso. In Robert
Botne & Rose Vondrasek (eds.), IUWPL 3: Explora9ons in African Linguis9cs: From Lamnso’ to Sesotho, 53-‐70. Bloomington, IN: IULC Publica<ons.
Michael, Lev David. 2008. Nan< eviden<al prac<ce: Language, knowledge, and social ac<on in an Amazonian society. The University of Texas at Aus<n Ph.D. disserta<on.
Pasch, Helma. 1985. Possession and possessive classifiers in ’Dongo‑ko. Afrika und Übersee 68.69-‐85.Pasch, Helma. 1986. Die Mba-‐Sprachen: Die Nominalklassensysteme und die gene9sche Gliederung einer Gruppe von Ubangi-‐Sprachen (Sprache und Geschichte in Afrika 6). Hamburg: Buske.
Pawley, Andrew. 2005. The chequered career of the Trans New Guinea hypothesis: Recent research and its implica<ons. In Andrew Pawley, Robert ALenborough, Jack Golson & Robin Hide (eds.), Papuan Pasts: Cultural, Linguis9c and Biological Histories of Papuan-‐speaking Peoples, 67-‐108. Canberra: Pacific Linguis<cs.
Payne, Doris. 1986. Noun classifica<on in Yagua. In ColeLe Craig (ed.) Noun Classes and Categoriza9on: Proceedings of a Symposium on Categoriza9on and Noun Classifica9on, Eugene, Oregon, October 1983, 113-‐131. Amsterdam: John Benjamins.
Payne, Doris. 2007. The source of the Yagua nominal classifica<on system. Interna9onal Journal of American Linguis9cs 73.447-‐474.
Payne, Doris L. & Thomas E. Payne. 1990. Yagua. In Desmond C. Derbyshire & Geoffrey K. Pullum (eds.), Handbook of Amazonian Languages, vol 2, 249-‐474. Berlin: Mouton de Gruyter.
Rabel-‐Heymann, Lili. 1977. Gender in Khasi nouns. Mon-‐Khmer Studies 6. 247-‐272. Reid, Nicholas. 1997. Class and classifier in Ngan’gityemerri. In Mark Harvey & Nicholas Reid (eds.),
Nominal classifica9on in Aboriginal Australia, 165-‐228. Amsterdam: John Benjamins.
References Ring, Hiram R. 2015. A grammar of Pnar. Nanyang Technological University (Singapore) Ph.D.
disserta<on. Ross, Malcolm. 2005. Pronouns as a preliminary diagnos<c for grouping Papuan languages. In
Andrew Pawley, Robert ALenborough, Jack Golson & Robin Hide (eds), Papuan Pasts: Cultural, Linguis9c and Biological Histories of Papuan-‐speaking Peoples, 15–66. Canberra: Pacific Linguis<cs.
Round, Erich R. & Greville G. CorbeL. (accepted). The theory of feature systems: One feature versus two for Kayardild tense-‐.-‐aspect-‐.-‐mood. Morphology.
Seifart, Frank. 2005. The structure and use of shape-‐based noun classes in Miraña (North West Amazon). Radboud University PhD disserta<on.
Seifart, Frank. 2009. Mul<dimensional typology and Miraña class markers. In Pa<ence Ebbs & Alexandre Arkhipov (eds.), New challenges in typology: Transcending the borders and refining the dis9nc9ons, 365-‐385. Berlin: Mouton de Gruyter.
Senr, Gunter. 1986. Kilivila: The Language of the Trobriand Islanders (Mouton Grammar Library 3). Berlin: De Gruyter Mouton.
Senr, Gunter. 1996. Classificatory par9cles in Kilivila (Oxford Studies in Anthropological Linguis<cs). New York: Oxford University Press.
Spencer, Andrew. 2005. Towards a typology of ‘mixed categories’. In: C. Orhan Orgun & Peter Sells (eds) Morphology and the Web of Grammar: Essays in Memory of Steven G. Lapointe (Stanford Studies in Morphology and the Lexicon), 95-‐138. Stanford: CSLI Publica<ons.
Stebbins, Tonya N. 2005. Nominal classifica<on in Mali. Anthropological Linguis9cs 47.77-‐121.
References Stebbins, Tonya N. 2011. Mali (Baining) grammar. (Pacific Linguis<cs 623.) Canberra: Research
School of Pacific and Asian Studies, Australian Na<onal University. Tucker, A. N. & M. A. Bryan. 1966. Linguis9c Analyses: The Non-‐Bantu Languages of North-‐Eastern
Africa (With a supplement on the Ethiopic Languages by Wolf Leslau). London: Oxford University Press.
Wurm, Stephen. 1982. Papuan Languages of Oceania. Tübingen: Narr. Zaliznjak, Andrej A. 1964. K voprosu o gramma<českix kategorijax roda i oduševlennos< v sovremennom russkom jazyke. Voprosy jazykoznanija no. 4.25‑40.
Zaliznjak, Andrej A. 1973. O ponimanii termina ‘padež’ v lingvis<českix opisanijax. In: Andrej A. Zaliznjak (ed.) Problemy gramma<českogo modelirovanija, 53-‐87. Moscow: Nauka. [Reprinted in: Andrej A. Zaliznjak. 2002. Russkoe imennoe slovoizmenenie: s priloženiem izbrannyx rabot po sovremennomu russkomu jazyku i obščemu jazykoznaniju, 613-‐647. Moscow: Jazyki slavjanskoj kul´tury.]
FRENCH Same seman<cs & different forms (our Type A3)
• Indo-‐European (Romance) • France, Canada, Switzerland, etc. • 75 million speakers
132
French
• Gender system: MASCULINE vs FEMININE; forms expressing these values depend on target
(15) le garçon est content l-‐ə gaʁso ɛ kõtɑ DEF-‐M.SG boy(M)[SG] is happy[M.SG] ‘the boy is happy’
(16) la femme est contente
l-‐a fam ɛ kõtɑ-‐t DEF-‐F.SG woman(F)[SG] is happy-‐F.SG ‘the woman is happy’
133
French: Type A3
• Forms are phonologically different, but realize the same gender values (the same gramma<cal meanings)
• Single system • Example of Type A3
134
RUSSIAN Different seman<cs & same form (our Type C1 )
135
• Indo-‐European (Slavonic) • Russia, etc. • 170 million speakers
Russian
• Candidate 1: MASCULINE vs FEMININE vs NEUTER • Candidate 2: ANIMATE vs INANIMATE • 3 X 2 = 6 possibili<es
• Large literature • Evidence in a small part of the paradigm only (the accusa<ve case)
136
Animacy has no unique forms
!!! SINGULAR! PLURAL! !
FEMININE! NEUTER! MASCULINE! !!!NOMINATIVE! novaja!
novoe!novyj! novye!
!
ACCUSATIVE!INANIMATE!
novuju!!
ANIMATE! !novyx!
!
GENITIVE!
novoj!
! novogo! !LOCATIVE! ! novom! !DATIVE! ! novomu! novym! !INSTRUMENTAL! ! novym! novymi! !
!
The Russian adjec<ve novyj ‘new’: paLerns of syncre<sm
137
Animacy in Russian
(19) ja viž-‐u star-‐yj dom I see-‐1SG old-‐M.INAN.SG.ACC house(M.INAN)[SG.ACC] ‘I see an old house’
(20) ja viž-‐u star-‐ogo drug-‐a
I see-‐1SG old-‐M.ANIM.SG.ACC friend(M.ANIM)-‐SG.ACC ‘I see an old friend’
(21) ja viž-‐u star-‐ogo dedušk-‐u
I see-‐1SG old-‐M.ANIM.SG.ACC grandfather(M.ANIM)-‐SG.ACC ‘I see (my) old grandfather’
Similarly for the plural (all three genders) 138
BAGVALAL Par<al overlap in seman<cs & same forms (our Type B1)
140
• Nakh-‐Dagestanian • Dagestan • 1,500 speakers • Kibrik (2001)
Bagvalal
• Gender: MASCULINE vs FEMININE vs NEUTER • Assignment is seman<c
– humans according to biological sex – rest is neuter
141
Bagvalal Bagvalal (Kibrik 2001: 64-‐66):
(23) waša w-‐iRi (24) waša-‐bi b-‐iRi-‐r a. boy M.SG-‐stop a. boy-‐PL HUM.PL-‐stop-‐HUM.PL
‘the boy stopped’ ‘the boys stopped’ b. jaš j-‐iRi b. jaš-‐i b-‐iRi-‐r
girl F.SG-‐stop girl-‐PL HUM.PL-‐stop-‐HUM.PL ‘the girl stopped’ ‘the girls stopped’
c. ʕama b-‐iRi c. ʕama-‐bi r-‐iRi
donkey N.SG-‐stop donkey-‐PL N.PL-‐stop ‘the donkey stopped’ ‘the donkeys stopped’
142
Bagvalal Bagvalal (Kibrik 2001: 64-‐66):
(23) waša w-‐iRi (24) waša-‐bi b-‐iRi-‐r a. boy M.SG-‐stop a. boy-‐PL HUM.PL-‐stop-‐HUM.PL
‘the boy stopped’ ‘the boys stopped’ b. jaš j-‐iRi b. jaš-‐i b-‐iRi-‐r
girl F.SG-‐stop girl-‐PL HUM.PL-‐stop-‐HUM.PL ‘the girl stopped’ ‘the girls stopped’
c. ʕama b-‐iRi c. ʕama-‐bi r-‐iRi
donkey N.SG-‐stop donkey-‐PL N.PL-‐stop ‘the donkey stopped’ ‘the donkeys stopped’
143
Bagvalal
• Sex-‐based dis<nc<on operates only for humans
• Seman<c opposi<ons suggest four possibili<es – only three are reflected in the gender system
144
Bagvalal
Bagvalal system matrix (two feature analysis) HUMAN NON-‐HUMAN
MASCULINE + -‐ FEMININE + -‐ NEUTER -‐ +
145