Modeling the complexity of grammatical gendersystems

Three case studies in synchronic and diachronic typology

Francesca Di Garbo

Stockholm University

21.06.2017

Outline

My approach to grammatical complexity

1 – Exploring grammatical complexity crosslinguistically: the caseof gender

2 – The evolutionary dynamics of gender systems’ complexity

3 – Correlates of restructuring in Bantu gender systems

References

My approach to grammatical complexity

How I define complexity: the number of parts in a system or thelength of its descriptionabsolute complexity (Dahl 2004, 2011; Miestamo2008).

How I investigate complexity crosslinguistically: based onfunctional domainslocal complexity (Miestamo 2006, 2008).

How I approach intra- and inter-language complexity variability:language variation and change can be influenced bysocial factors. Languages as complex adaptivesystems (Beckner et al. 2009; Maitz & Nemeth2014).

Dimensions of gender complexity(Audring 2014)

Complexity of values = number of genders.

Complexity of assignment = number and scope of genderassignment rules.

Complexity and amount of formal marking = number of genderagreement targets, frequency of gender marking indiscourse.

The study: aim, method, data(Di Garbo 2016)

⇒ Aim:

I To identify relevant dimensions of gender complexity andimplement them into a complexity metric.

⇒ Method:

I One linguistic area: Africa

I Intra- and intergenealogical typology

⇒ Data:

I 84 languages from 17 different genealogical units from Africa.

I Descriptive resources and consultation of language experts.

4 / 37

−10 0 10 20 30 40 50

−30

−20

−10

010

2030

0 500 1000 1500 km

scale approx 1:78,000,000

The language sample

Legend

BantuBerberChadicCushiticDizoidE. NiloticHadzaKhoe−KwadiKwa

KxaMelN.C. AtlanticSandaweSemiticS. OmoticT.N. OmoticTuu

5 / 37

Sex−Based vs. Non−Sex−Based systems

−10 0 10 20 30 40 50

−3

0−

20

−1

00

10

20

30

0 500 10001500 km

scale approx 1:78,000,000

Sex−based, 48

Non−sex−based,

36

6 / 37

Proposed complexity metric

Dimension Feature Value Score

Va

lues

Number of gender values (gv)

Two genders 0Three 1/3Four 2/3Five or more 1

Ass

ign

men

tru

les

Nature of assignment rules (ar)Semantic 0Semantic and formal 1

Manipulation of gender assignment triggered by number/countability (m1)Absent 0Present 1

Manipulation of gender assignment triggered by size (m2)Absent 0Present 1

For

m.

mar

kin

g

Number of gender indexing domains (ind)

One 0Two 1/3Three 2/3Four or more 1

Cumulative exponence of gender and number (cum)Noncumulative 0Partially cumulative 1/2Cumulative 1

7 / 37

Results

Gender Complexity Scores

Frequency

0.0 0.2 0.4 0.6 0.8 1.0

05

10

15

20

0.0 0.2 0.4 0.6 0.8 1.0

Languages

Scores

Figure 1: Distribution of the GCSs

I The gender systems of thesampled languages are generallyassociated with high complexityscores.

I Genealogical homogeneity.

8 / 37

Genealogical homogeneity and outliersThe Bantu languages

Table 1: Bantu CGS

ISO Language GCSbaz Tunen 0.95bem Bemba 1⇒bip Bila 0.16bvx Dibole 0.78cgg Chiga 1eto Eton 0.83kik Gikuyu 1kki Kagulu 1ksf Bafia 0.78lea Lega 1⇒lin Lingala (Kinshasa) 0.22lol Mongo-Nkundu 1mcp Makaa 1ndg Ndengereko 1nso Sotho,Northern 0.83nya Chichewa 1sna Shona 1ssw Swati 0.83swa Swahili 1toi Tonga 1tsn Tswana 0.83ven Venda 1zul Zulu 0.83

I Bila [Glottocode: bila1255] (Kutsch Lojenga 2003):I Two genders (Animate vs. Inanimate)I Semantic, non-manipulable assignment rulesI NP-internal agreement onlyI Spoken in a linguistic area at the crossroads

between different language families: Bantu,Ubangi, Nilotic.

I Kinshasa Lingala [Glottocode: ling1269] (Bokamba

1977, 2009; Meeuwis 2013)I Two genders (Animate vs. Inanimate) (fossilized

noun class marking on nouns)I Semantic, non-manipulable assignment rulesI Agreement only on third person pronouns and verbsI Kinshasa Lingala is the direct descendant of the

Bangala pidgin.

9 / 37

Summary

I High gender complexity, constant across genealogically relatedlanguages.

I Gender complexity outliers have peculiar sociolinguisticprofiles (multilingualism, contact).

10 / 37

Next steps

I Integrating the diachronic dimension to the crosslinguisticstudy of gender complexity.

I Digging into the sociohistorical factors that may contribute tocomplexification/simplification of gender systems.

11 / 37

The evolution of gender systems’ complexity(project funded by the Wenner-Gren Foundations, Feb 2015 – Dec 2016)

Aims:

I To study the life-cycle of grammatical gender systems: emergence,expansion, reduction, loss.

I To study the sociohistorical correlates of these patterns of change.

Points of departure:

I Gender systems are very stable (Nichols 1992); they tend to“clusterin adjacent or nearby languages” (Nichols 2003: 300-303).

I Contact-induced loss and emergence of grammatical genderpresuppose intensive bilingualism and heavy borrowing (Thomason2001: 71).

Approach:

I Diachronic/dynamic typology (Greenberg 1978; Croft 2003): thestudy of pathways of change between language types/structures.

Domain of analysis

I Gender agreement patternsHow the marking of grammatical gender on modifiers,predicates, pronouns changes over time and under thepressure of language contact.

Why?

I Inflectional morphology = morphological complexity

I Morphological complexity is sensitive to language contactdynamics (Lupyan & Dale 2010; Bentz et al. 2015).

14 / 37

Method

I Convenience sample of 15 sets of closely related languages (36lngs in total), each representing:

I Reduction/loss/expansion/emergence of gender agreementI A diverse range of sociohistorical profiles:

e.g., standard/prestige languages vs. minority languages;high-contact varieties vs. low-contact varieties.

I Data collected through a questionnaire and descriptiveresources.

15 / 37

The language sample

Legend

Balto−SlavicBantuBasqueChamorroCentral GunwinyguanGermanicGhana−Togo−MountainGreek

Insular CelticIranianKhasianLezgicMekMichifThebor

16 / 37

Patterns of change

Legend

Emergence = 5/36Loss = 7/36Expansion = 6/36

Reduction = 8/36Retention = 8/36Lack = 2/36

17 / 37

Two paths of loss/reduction

1. Morphophonological erosion of agreement morphology

2. Redistribution of agreement patterns

attributive (...) pers. pronoun

MORPH. EROSION

REDISTRIBUTION

Reduction/loss by morphophonological erosionStandard Swedish (Indo-European, Germanic)

I Adnominal gender agreement: Common vs. Neuter GenderI en person ‘a person’I ett hus ‘a house’

Table 2: Personal Pronouns

Hum. and Higher Anim. M F Phan ‘he’ hon ‘she’ de ‘they’

Inanim. C N Pden ‘it’ det ‘it’ de ‘they’

I Comparative evidence from other Swedish dialectsI Elfdalian Sw.: triparite gender system maintained throughout

the agreement system (Akerberg 2012)I Karleby Sw.:

complete gender loss except for definite article, personal anddemonstrative pronouns (Hulden 1972; Hultman 1894).

19 / 37

Reduction/Loss by redistribution

(1) Axo Cappadocian (Indo-European, Greek; Karatsareas2014: 79-80)

tdef.sg.gen

spitcuhouse.sg.gn

tadef.pl

ndix(u)swall.pl

xtizmenabuilt.pl

‘The walls of the house (are) built.’

(2) Modern Standard Greek (Indo-European, Greek; Karatsareas2014: 79-80)

idef.m.pl

tıciwall.m.pl

inebe.prs.3pl

xtixmenibuilt.m.pl

‘the walls are built’.

I Comparative evidence from other Asia Minor Greek dialects:I Pontic Greek: the expansion of neuter agreement is

semantically and syntactically constrained(inanimate nouns, agreement targets non-adjacent to nouns).

20 / 37

Emergent gender agreement patterns

The diachrony of many gender systems “can at best bereconstructed, but not directly observed” (Luraghi 2011: 435).

I Focus of the project: young, and grammatically non-pervasivegender systems.

1. Resulting from light nouns, e.g., ‘man’, ‘woman’,grammaticalizing as anaphoric devices (Walchli accepted)

2. Resulting from borrowing of nouns and agreeing adnominalmodifiers.

21 / 37

Emergent gender agreement patterns

The diachrony of many gender systems “can at best bereconstructed, but not directly observed” (Luraghi 2011: 435).

I Focus of the project: young, and grammatically non-pervasivegender systems. Two types:

1. Resulting from light nouns, e.g., ‘man’, ‘woman’,grammaticalizing as anaphoric devices (Walchli accepted)

2. Resulting from borrowing of nouns and agreeingadnominal modifiers.

21 / 37

Borrowed gender agreement

LanguagesI Chamorro (Austronesian)

Contact language:Spanish

I Lekeitio Basque (Basque)

Contact language:Spanish

I Schumcho, Jangshung(Bodic, Thebor)

Contact language:Northern IndiaIndo-Europeanlanguages

Shared characteristics

X Gender agreement patterns passedthrough borrowing of inflected forms.

X Gender agreement targets are a closedclass of property words.

X Gender agreement patterns are alwayssemantic (natural gender distinctions).

22 / 37

Borrowed gender agreementChamorro (Austronesian, Marian Islands; Huber 2011: 67)

(3) Chamorro Feminine Gender (Stolz 2012: 123)

Ma-nobena-na-yepass-novena-red-ref

idef

mi-milagros-aabound-miraculous-f

nalink

Bithen.Virgin

‘A novena is being conducted for the abundantly miraculousVirgin.’

(4) Chamorro Non-Feminine Gender (Stolz 2012: 125)

desdesince

antititesred:before

nalink

tiempotime

estaalready

gofvery

bunit-unice-nf

nalink

siudatown

idef

yaTN

Hagatna.Hagatna

‘A very long time ago, Hagatna was a very pretty town already.’

23 / 37

Discussion

I Attributive modifiers and personal pronouns have a specialstatus in the unfolding of diachronic change in the domain ofgender marking.

I Under gender reduction/loss,I The direction of change differs depending on the process

involved – morphophonological erosion vs. redistributionI Two distinct functional principles can motivate this

directionality:I the syntactic cohesion between agreement targets and

controller nounsI the sensitivity of different agreement targets to the semantic

properties of nouns and discourse referents.

24 / 37

Within Eurasia, patterns of change cluster aroundlanguage-family edges

Languages Family Contact family Observed patternCappadocian Greek Greek Turkic LossTamian Latvian Balto-Slavic Finnic LossAghul, Udi Lezgic Turkic LossKarleby Swedish North Germanic Finnic Near-lossKelasi, Kaftej Northwestern Iranian Turkic Loss and expansionLekeitio Basque Basque Ibero-Romance EmergenceShumcho, Jangshung Thebor Indo-Aryan Emergence

I Outlier languages within a family are neighbor with eachother.

I This is in alignment with Nichols’ (2003) observation wherebygrammatical gender is a cluster phenomenon.

Asymmetries in the relationship between languages incontact (may) explain the direction of change

I Contact-induced loss and emergence of gender agreementmorphology presuppose prolonged contact and extensivebilingualism.

I The direction of change is to some extent predicted by the prestigedynamics between the languages in contact

Languages Change Dominant contact lng GG in thedominant lng

Aghul, Udi (Lezgic) Loss Azerbaijani (Turkic), NOGeorgian (Kartvelian)

Igo (Ghana-Togo-Mountain) Loss Ewe (Gbe) NOTamian Latvian (Balto-Slavic) Loss Livonian, Estonian (Finnic) YESChamorro (Chamorro) Emergence Spanish (Romance) YESLekeitio Basque (Basque) Emergence Spanish (Romance) YES

26 / 37

To sum up

This project has contributed to highlight:

I types of changes in gender agreement systems and possibledirectionalities in the spread of these changes

I a number of sociohistorical variables that are relavant to theunderstanding of the evolution of gender agreement systems.

Limitations:

I Only a limited number of languages per family.

I Too little data for some of the languages in the sample.

27 / 37

Next step

I Using the results of the study as a starting point for furtherhypothesis testing on larger data sets (one family in detail),and with the support of quantitative methods.

28 / 37

Correlates of restructuring in Bantu gender systems (withAnnemarie Verkerk, MPI – Jena)

I Studying the diversity of the gender systems of the Bantulanguages.

I Testing the models of language change that account best forwithin-family variation in this domain of grammar.

I Investigating socio-historical correlates of the distribution ofthis variation.

The Bantu languages and their gender systems

(5) Gender marking in Chichewa (Kiso 2012: 18)

chi-nkhaniracl7-scorpion

cha-chi-kaziass-cl7-female

chi-ku-dzi-kandacl7.sbj-pres-refl-scratch

“The female scorpion is scratching itself”.

(6) Gender marking in Kinshasa Lingala (Meeuwis2013: 30)

a. Mw-anacl1-child

a-ko-kweya3sg.anim-fut-fall

‘The child will fall.’

b. Ndakocl9.book

e-ko-kweya3sg.inan-fut-fall

‘The house will fall.’

30 / 37

The Bantu languages and their gender systems

(7) Gender marking in Chichewa (Kiso 2012: 18)

chi-nkhaniracl7-scorpion

cha-chi-kaziass-cl7-female

chi-ku-dzi-kandacl7.sbj-pres-refl-scratch

“The female scorpion is scratching itself”.

(8) Gender marking in Kinshasa Lingala(Meeuwis2013: 30)

a. Mw-anacl1-child

a-ko-kweya3sg.anim-fut-fall

‘The child will fall.’

b. Ndakocl9.book

e-ko-kweya3sg.inan-fut-fall

‘The house will fall.’

30 / 37

Questions

I How do we go from the Chichewatype to the Kinshasa Lingala type?

I Why does this happen?

The evolution of Bantu gender marking systemsI Questions

1. Which word classes carry gender marking besides nouns (e.g.,pronouns, verbs, adjectives)?

2. Are animacy-based distinctions part of the gender markingsystem?

I Quantitative data analysisI The coding will be mapped on the Bantu phylogenetic tree

(Grollemund et al. 2015) to estimate transition probabilitiesbetween attested systems using Phylogenetic ComparativeMethods.

Hypotheses

1. Animacy-based distinctions encroach the gender marking systemstarting from anaphoric pronouns and gender markers on verbs.

2. Marking on nouns is more stable than marking on other wordclasses.(Wald 1975; Di Garbo & Miestamo accepted)

31 / 37

Sociohistorical correlates

I Variables we plan to work with:I Population data (both L1 and L2)I Presence/absence of gender systems in neighboring languages

Hypotheses

1. Large populations with high proportions of L2 users and intenselanguage contact predict reduction and/or loss of gender marking.

2. Small populations with low proportions of L2 users and intenselanguage contact predict retention of gender marking (groupidentity marking) or its reduction/loss (shift-induced interference).

3. Geographic proximity between related and unrelated languagespredicts convergence in the domain of gender marking.

32 / 37

What we’ve done so farI Defined the coding procedure

I Collected data about the gender system of 130+ Bantu languages.

0 10 20 30 40 50

−30

−20

−10

0

0 500 1000 km

scale approx 1:32,000,000

Languages sampled so far

A15A15A15A15

A31, A221

A51A601

A72

A74

A75, A751

A83

A84A841, A842

A91

A92

A93

B202

B25B251 B252

B304B305B42

B52

B61

B77b, B78 B85B85

C12 C13

C14

C25

C30A

C30bC30b

C33

C36e

C37C43, C44

C441C52

C63, C62

C71C83

D13

D22

D26D27

D301

D33

E55 E56H10?

H14 − H16

JE15

JE42

L31

R20 (R21 − R24, R11−R118 R421−R422)

S407

A101A13 / A141A21A24−26

A33a

A42 A53

A81, A801

B11B201

B21

B22

B22

B23B24B31

B32

B41, B411B51

B62

B71

B73B81

C16

C21 (C23)C22C24

C31C31

C36

C401

C41 C45C54

C61 C611 C36hC74, C75

C81

D11

D201D211, D311, D313

D23

D25 D251

D28

D304

D308

D32

G24

G42 G43

G42 G43

K11

L52

S11 − S15

S20 (S21)

S31S32 S301 − S304

S33S41

S42S43

S53 (S52)

33 / 37

Maho’s (1999) classification of Bantu gender systems

Nouns

1= Tr. 2= Tr. + An. 2i = Tr+Pl 3 = An. +Sg/Pl 4= Sg/Pl 5=NoneElsewhereA = Tr.B = Tr. + An.C = An. + Sg/PlD = Sg/PlE = None

34 / 37

Maho’s (1999) typology of Bantu gender systems

Nouns

1= Tr. 2= Tr. + An. 2i = Tr+Pl 3 = An. + Sg/Pl 4= Sg/Pl 5= NoneElsewhereA = Tr. ZuluB = Tr. + An. Swahili LundaC = An. + Sg/Pl Lingala K. Lingala Amba, Bera Pande HomaD = Sg/Pl YansE = None Kituba Komo

34 / 37

0 10 20 30 40 50

−30

−20

−10

0

0 500 1000 km

scale approx 1:40,000,000

The languages of the sample based on Maho's types

1A1A1A1A

1A

1B1B1A1A

1A

1A

1A1A1A

3D

3C

1A1A1A 1A

1A1A1A1A

1A

1B 1B1D

4C 4C

1A

1A

1A

1C2'C

1B

1B

1A 1B1A1A

1B

1A1A

1A

3C

1A1A

3C

1B

1A1E

1B

1A

1A

1A

1A

1A

1A1A1A1A

1A

1A 1A

1A

1A1B

1A

1A

1A

1A1A1A

1A

1B1A

1A1A

1B1A

1A

1A1A1A

1A1A1A

1A

1B 1B1A

1B1A

1A

1A

1B3C

3E

1A

1A

5C4E

1A

1B

1B

1B

1B

2B

1A

1A

1A 1A

1A1A

1A1A

1A

35 / 37

Some observations

I The gender systems of several Bantu languages show a biastowards the overt expression of animacy distinctions.

I The spread of this feature within the family is NOT a unitaryprocess.

I Multiple developments must be posited in different subareasof the Bantu speaking world.

36 / 37

The northern Bantu borderlands

0 500 1000 km

scale approx 1:26,000,000

Radically restructured gender systems in the Bantu northern borderlands

keb

nra

swc

nda

bsibssbqzmbonkc

bvbewo

bum

fan

mcp

njy

ozm

kwu

sxe

koqsakmhb

picbuwsnq

nzb

mdt

zmx

mdw

bxg

bjabbmsoc

tllbuf

zmq

zmbbnx

kam

lug

guz

lua

bdubwtbridua

yko

abbksf

nmg

mye

syi

zmnwumtsv

kbs

dma

mbm

teg

tii

loq

mdu

akwkoh

mmz

ndw

lse

pae

loo

yel

dez

mdq

lea

hoo

buu

lfabag

ifm

szg

bkt

bww

nxd

nlj

kng

swj

iyx

ngc agh

lol

lik

bou

wmw

swh

cjk

lin

yns

ktu

Nzadi

lun

kkj

kbj

bip brfrwm

pmm

kmw

bkj mdn

boy

hom

Legend

Type 1: N_tr; AG_an

Type 2: N_an; AG_an

Type 3: N_an; AG_sg/pl

Type 4: N_rel; AG_none

No radical restructuring

Atlantic

Ocean

Two possible scenarios:1. Substratum interference from pre-Bantu populations shifting to

Bantu languages, possibly including Pygmies.

2. Continued contact between Bantu and non-Bantu languages in thearea.

37 / 37

Thank you very much!Thanks are also due to:

Anna Ahlstroms och Ellen TerserusStiftelse

38 / 37

References I

Akerberg, Bengt. 2012. Alvdalsk grammatik. Ulum Dalska.

Audring, Jenny. 2014. Gender as a complex feature. Language Sciences 43. 5–17. [Special issue: Exploringgrammatical gender].

Beckner, Clay, Nick C. Ellis, Richard Blythe, John Holland, Joan Bybee, Jinyun Ke, Morten H. Christiansen, DianeLarsen-Freeman, William Croft & Tom Shoenemann. 2009. Language is a complex adaptive system: PositionPaper. Language Learning 59. 1–26.

Bentz, Christian, Annemarie Verkerk, Douwe Kiela, Felix Hill & Paul Buttery. 2015. Adaptive communication:Languages with more non-native speakers tend to have fewer word forms. Plos One 10. 1–23.

Bokamba, E. 2009. The spread of Lingala as a lingua franca in the Congo Basin. In Fiona McLaughlin (ed.), Thelanguages of urban Africa, 50–70. London: Continuum.

Bokamba, Eyamba. 1977. The impact of multilingualism on language structures: the case of Central Africa.Anthropological Linguistics 19. 181–202.

Bostoen, Koen & Hilde Gunnink. in preparation. The impact of autochthonous languages on bantu languagevariation: A comparative view on southern and central africa. In Cambridge handbook of language contact, .

Croft, William. 2003. Typology and universals. Cambridge: Cambridge University Press.

Dahl, Osten. 2004. The growth and maintenance of linguistic complexity. Amsterdam: John Benjamins.

Dahl, Osten. 2011. Grammaticalization and linguistic complexity. In Heiko Narrog & Bernd Heine (eds.), TheOxford handbook of grammaticalization, 153–162. Oxford: Oxford University Press.

Di Garbo, Francesca. 2014. Gender and its interaction with number and evaluative morphology: An intra- andintergenealogical typological survey of Africa. Stockholm: Department of Linguistics, Stockholm Universitydissertation.

Di Garbo, Francesca. 2016. Exploring grammatical complexity crosslinguistically: The case of gender. LinguisticDiscovery 14. 46–85. DOI:10.1349/PS1.1537-0852.A.468.

Di Garbo, Francesca. under review. The complexity of gender and language ecology, .

References IIDi Garbo, Francesca & Matti Miestamo. accepted. The evolving complexity of gender agreement systems. In Di

Garbo, Francesca and Bernhard Walchli (ed.), Grammatical gender and linguistic complexity, To be submittedto: Berlin: Language Science Press.

Greenberg, Joseph. 1978. How does a language acquire gender markers? In Joseph Greenberg, Charles Ferguson &Edith Moravcisk (eds.), Universals of human language, vol. 3: Word structure, 47–92. Standford: StanfordUniversity Press.

Grollemund, Rebecca, Simon Brandford, Koen Bostoen, Andrew Meade, Chris Venditti & Mark Pagel. 2015. Bantuexpansion shows that habitat alters the route and pace of human dispersals. Proceedings of the NationalAcademy of Science of the United States of America 112. 13296–13301.

Huber, Christian. 2011. Some notes on gender and number marking in Shumcho. In Gerda Lechleitner & ChristianLiebl (eds.), Jahrbuch des phonogrammarchivs, vol. 2, 52–90. Gottingen: Cuvillier Verlag.

Hulden, Lars. 1972. Genussystemet i Karleby och Nedervetil. Folkmasstudier 22. 47–82.

Hultman, Oskar Fredrik. 1894. Oskar fredrik hultman. Helsinki: Svenska landsmalsforeningen.

Karatsareas, Petros. 2014. On the diachrony of gender in Asia Minor Greek: the development of semanticagreement in Pontic. Language Sciences 43. 77–101.

Kiso, Andrea. 2012. Tense and aspect in Chichewa, Citumbuka and Cisena. a description and comparison of thetense-aspect systems in three southeastern Bantu languages: Stockholm University dissertation.

Kutsch Lojenga, Constance. 2003. Bila (D 32). In Derek Nurse & Gerard Philippson (eds.), The Bantu languages,450–474. London: Routledge.

Lupyan, Gary & Rick Dale. 2010. Language structure is partly determined by social structure. PLOS one 5(1). 1–10.

Luraghi, Silvia. 2011. The origin of the Proto-Indo-European gender system: Typological considerations. FoliaLinguistica 45. 435–464.

Maho, Jouni. 1999. A comparative study of Bantu noun classes. Goteborg: Orientalia et Africana Gothoburgensiadissertation. Acta universitatis gothoburgensia.

Maitz, Peter & Attila Nemeth. 2014. Language contact and morphosyntactic complexity: Evidence from German.Journal of Germanic Linguistics 26(1). 1–29.

41 / 37

References III

Meeuwis, Michael. 2013. Lingala. In Susanne Michaelis, Philipe Maurer, Martin Haspelmath & Magnus Huber(eds.), The survey of pidgin and creole languages, vol. III, Contact languages based on languages from Africa,Asia, Australia and the Americas, 25–33. Oxford: Oxford University Press.

Miestamo, Matti. 2006. On the feasibility of complexity metrics. In Krista Kerge & Maria-Maren Sepper (eds.),Finest Linguistics. Proceedings of the Annual Finnish and Estonian Conference of Linguistics, Tallin, May 6–7,2004, 11–26. Tallin: TLU.

Miestamo, Matti. 2008. Grammatical complexity in a cross-linguistic perspective. In Matti Miestamo, KaiusSinnemaki & Fred Karlsson (eds.), Language complexity: Typology, contact, change, 23–41. Amsterdam: JohnBenjamins.

Nichols, Johanna. 1992. Linguistic diversity in space and time. Chicago: University of Chicago Press.

Nichols, Johanna. 2003. Diversity and stability in language. In Brian Joseph & Richard Janda (eds.), Thehandbook of historical linguistics, 283–310. Oxford: Blackwell.

Stolz, Thomas. 2012. Survival in a niche. On gender-copy in Chamorro (and sundry languages). In MartineVanhove, Thomas Stolz, , Hitomi Otsuka & Aina Urdtze (eds.), Morphologies in contact, 93–140. Munich:Akademie-Verlag.

Thomason, Sarah. 2001. Language contact: an introduction. Washington D.C.: Georgetown University Press.

Walchli, Bernhard. accepted. The feminine gender gram, incipient gender marking, maturity, and extractinganaphoric gender markers from parallel texts. In Di Garbo, Francesca and Bernhard Walchli (ed.), Grammaticalgender and linguistic complexity, To be submitted to: Berlin: Language Science Press.

Wald, Benji V. 1975. Animate concord in northeast coastal Bantu: Its linguistic and social implications as a case ofgrammatical convergence. Studies in African Linguistics 6. 267–314.

42 / 37