Subjecthood of the agent argument in Estonian passive constructions

36
1 Subjecthood of the agent argument in Estonian passive constructions TO APPEAR IN: M.-L. HELASVUO & T. HUUMO (EDS.), CANONICAL AND NON-CANONICAL SUBJECTS IN CONSTRUCTIONS. AMSTERDAM/PHILADELPHIA: JOHN BENJAMINS. LIINA LINDSTRÖM University of Tartu This paper tackles diachronic changes in the choice of elative or adessive case for marking the agent in Estonian periphrastic passive constructions in two time periods. In 1800-1850, the main agent-marking device was the elative case, whereas in the 1990s the elative was limited to inanimate actors, and the use of the adessive had increased considerably. However, the adessive can only be used for marking volitional, animate agents. Changes are observed with regard to semantic constraints, subject properties of the adessive and elative agents, and language contacts. Adessive arguments behave like non- canonical subjects in many constructions in Estonian, and the use of the adessive for marking agents in passives is strengthened by the possessive perfect construction in Eastern Circum-Baltic languages. 1. Introduction The passive is one of the most often described and discussed issues in the field of syntax. In Estonian linguistics, the most intriguing and discussed question concerning the category of voice is the distinction between the impersonal compound tenses and periphrastic passive, as both constructions share basically the same morphosyntactic means. While the Estonian impersonal voice typically demotes the human agent, the periphrastic passive, additionally, promotes the patient to subject (cf. Rajandi 1999 [1968], Erelt 1979, Pihlak 1993, Holvoet 2001a, Blevins 2003, Torn 2002, 2006, Torn- Leesik 2009, Lindström & Tragel 2007, 2010). Thus, the impersonal is used mainly to background the subject, whereas the periphrastic passive is used to foreground the patient and background the agent. Less attention has been paid to different agent marking devices and conditions of using these devices in Estonian passive and impersonal clauses (see section 2.2. for references). The Estonian passive is typically agentless,

Transcript of Subjecthood of the agent argument in Estonian passive constructions

1

Subjecthood of the agent argument in Estonian passive constructions

TO APPEAR IN: M.-L. HELASVUO & T. HUUMO (EDS.), CANONICAL AND NON-CANONICAL

SUBJECTS IN CONSTRUCTIONS. AMSTERDAM/PHILADELPHIA: JOHN BENJAMINS.

LIINA LINDSTRÖM

University of Tartu

This paper tackles diachronic changes in the choice of elative or adessive case

for marking the agent in Estonian periphrastic passive constructions in two

time periods. In 1800-1850, the main agent-marking device was the elative

case, whereas in the 1990s the elative was limited to inanimate actors, and the

use of the adessive had increased considerably. However, the adessive can

only be used for marking volitional, animate agents. Changes are observed

with regard to semantic constraints, subject properties of the adessive and

elative agents, and language contacts. Adessive arguments behave like non-

canonical subjects in many constructions in Estonian, and the use of the

adessive for marking agents in passives is strengthened by the possessive

perfect construction in Eastern Circum-Baltic languages.

1. Introduction

The passive is one of the most often described and discussed issues in

the field of syntax. In Estonian linguistics, the most intriguing and discussed

question concerning the category of voice is the distinction between the

impersonal compound tenses and periphrastic passive, as both constructions

share basically the same morphosyntactic means. While the Estonian

impersonal voice typically demotes the human agent, the periphrastic passive,

additionally, promotes the patient to subject (cf. Rajandi 1999 [1968], Erelt

1979, Pihlak 1993, Holvoet 2001a, Blevins 2003, Torn 2002, 2006, Torn-

Leesik 2009, Lindström & Tragel 2007, 2010). Thus, the impersonal is used

mainly to background the subject, whereas the periphrastic passive is used to

foreground the patient and background the agent.

Less attention has been paid to different agent marking devices and

conditions of using these devices in Estonian passive and impersonal clauses

(see section 2.2. for references). The Estonian passive is typically agentless,

2

but it is still possible to express the agent in the passive clause by various

means, e.g. postpositional phrases, the genitive, adessive and elative case (cf.

Rajandi 1999). In this paper, two types of agent phrases are compared: those

marked with the elative and the adessive case. The use of elative and adessive

agents in Estonian has changed considerably; in the early 19th century the main

agent-marking device in the passive construction was the elative case, whereas

by the end of the 20th century the use of the elative was limited to inanimate

actors and the use of the adessive had increased noticeably. This article

discusses this change and the possible reasons for it.

Note that the term agent is used in a wide sense here as it refers not

only to volitional agents, but also to non-volitional and sometimes even

inanimate actors. To avoid further confusion, I use the abbreviation A for the

most agent-like argument in this paper and P for the most patient-like argument

(similarly to Comrie 2008: 4). Thus the term A is used for the most agent-like

argument which is coded as a subject of an active clause and is demoted to an

oblique or omitted in the corresponding passive clause.

The available options for expressing adessive and elative As are

investigated during two time periods: from 1800 to 1850, and during the 1990s.

Between these two time periods, the use of the elative case as an agent-marking

device decreased and the adessive case spread. Possible explanations for these

changes can be summoned from three different areas: 1) the semantics of the

respective cases, 2) the subject properties of the adessive and elative

arguments, and 3) their use in different constructions within the Circum-Baltic

language area.

The properties of the subject have been much discussed since E.

Keenan’s pioneering article in 1976 (cf. Keenan 1976, Andrews 1985, Onishi

2001, Barðdal 2006, among others); these subject properties have been tested

on particular constructions in particular languages (Timberlake 1976, Beck

2000, Barðdal & Eythórsson 2003, Barðdal 2006 among others, regarding

Estonian: Erelt 2004, Koks 2004, Lindström 2012, Lindström, to appear,

Metslang, to appear). This article discusses some subject properties that can be

used to test the A of a passive construction – agentivity, topicality, and control

over reflexivization. Since the passive sentence has a syntactic subject (P), the

competition between adessive or elative As and the P for the position of subject

in these constructions is discussed.

The data for this study was obtained primarily from the Estonian

language corpora. The data from 1800 to 1850s was taken from the Corpus of

Old Written Estonian1 (OWE); contemporary Estonian is represented by the

1990s fiction texts subcorpus of the Corpus of Literary Estonian2, and to a

1 http://www.murre.ut.ee/vakkur/Korpused/korpused.htm

2 http://www.cl.ut.ee/korpused/grammatikakorpus/

3

lesser extent the texts from the Corpus of Spoken Estonian3 (Corpus of Spoken

Estonian, SE).

All passive sentences present in the data were included in the analysis.

Estonian passive constructions contain the verbs ole- ‘be’ (1) or saa- ‘get,

become’ (2) as auxiliaries and the past passive participle (PPP). The promoted

patient phrase shows agreement with the auxiliary (as canonical subjects do).

The A can be oblique or – more typically – is absent.

(1) Tibu-d ol-i-d prae-tud (tun-tud koka poolt).

P A

Chick-PL.NOM be-PST-PL fry-PPP (know-PPP chef.GEN by)

‘The chickens were fried (by a well-known chef)’

(2) Sibula-d sa-i-d (mei-l) lõpuks söö-dud.

P A

onion-PL.NOM get-PST-3PL (we-ADE) finally eat-PPP

‘The onions were finally eaten by us’

The paper is organized as follows: section 2 gives a brief overview of

the category of passive in Estonian. Section 3 compares the dynamics of the

use of agent-marking means in two time periods: 1800-1850 and the 1990s.

Section 4 compares the use of the adessive and elative As in the two time

periods, taking into account their semantic constraints, topicality and control

over the use of the possessive reflexive pronoun. In sections 5, 6 and 7,

possible explanations for the increase of the use of the adessive and decrease of

the use of the elative are considered: section 5 takes into account semantics of

the adessive and elative case, section 6 takes a look at their functions and

subject-like behavior in other constructions, and in section 7, the spread of the

possessive perfect construction in Eastern Circum-Baltic languages as a

possible explanation for these dynamic tendencies in Estonian is taken under

consideration.

2. Estonian impersonal and passive constructions

2.1. Impersonal and passive: an overview

Estonian, like other Baltic-Finnic languages, has historically only

distinguished between personal and impersonal voice (Viitso 2003: 216). The

Estonian impersonal is subjectless; the A of the event is covert. Syntactically,

the P is an object of the impersonal clause and is marked with either the

partitive (partial object, see example 3) or the nominative (total object,

example 4). The impersonal is formed with the verb suffixes -ta(kse)-

3 http://www.cl.ut.ee/suuline/

4

(present), -ti (past), or with compound verb forms: the verb olema and the past

passive participle ending -tud (perfect and pluperfect). The impersonal can be

derived from both transitive and intransitive clauses.

(3) Se-da raamatu-t loe-ti suure huvi-ga.

this-PAR book-PAR read-IPS.PST big.GEN interest-COM

‘(People) read this book with great interest.’

(4) See raamat loe-ti suure huvi-ga läbi.

this book read-IPS.PST big.GEN interest-COM through

‘(People) read this (whole) book with great interest.’

The A of the impersonal construction is typically left uncoded. The referent of

the implicit argument is always human, mostly a general or plural participant

(Rajandi 1999, Pihlak 1993, Torn 2002, Blevins 2003, Erelt 2003, Vihman

2008, Torn-Leesik 2009, Torn-Leesik & Vihman 2010). The Estonian

impersonal is used mainly speaker-exclusively (Torn-Leesik & Vihman 2010,

Pajusalu, this volume).

The Estonian personal passive is also referred to as a resultative or

stative passive. It has an overt subject in the nominative case and expresses a

state into which the referent of the subject (semantically the patient) has

entered as a result of the action. The Estonian passive is more stativizing and

resultative than the English passive (Vihman 2008).

The passive can be formed only from transitive clauses. In the passive

construction, the P is promoted to subject and agrees with the verb olema ‘be’

(5) or saama ‘get’ (2).

(5) a. Raamat ol-i läbi loe-tud.

book.SG.NOM be-PST.3SG through read-PPP

‘The book was read (all the way through).’

b. Raamatu-d ol-i-d läbi loe-tud.

book-PL.NOM be-PST-3PL through read-PPP

‘The books were read (all the way through).’

There has been a great deal of discussion about the relationship between

impersonal and passive categories in Estonian (Wiedemann 1875, Erelt 1979,

Pihlak 1993, Rajandi 1999 [1968], Torn 2002, 2006, Vihman 2007, Torn-

Leesik 2009, Lindström & Tragel 2007, 2010, among others). The source of

disagreement is that the Estonian passive and impersonal paradigms are not

always distinguishable. There is a notable overlap between impersonal

compound tenses (present and past perfect) and passive simple present and

past: both constructions use the auxiliary verb olema ‘be’, sometimes also

5

saama ‘get, become’, the past passive participle (which historically belongs to

the impersonal paradigm) and the NP in the nominative case, which can be

interpreted as a total object (in the impersonal) or subject (in the passive

construction); in both constructions the nominative NP can trigger agreement

with the verb. Some researchers have distinguished impersonal compound

tenses with a nominative object from passive simple present and past mainly on

the basis of semantics: the impersonal is more dynamic and expresses the event

of change, which brings about a new state, while the passive is stative and

expresses being in the state (Rajandi 1999 [1968]: 94). However, in analyzing

corpus data, this distinction is hardly applicable as a criterion, as most of the

clauses are ambiguous between the two.

In this article, I distinguish impersonal compound tenses with NPNOM

from passive clauses by formal criteria; all clauses with olema ‘be’ or saama

‘get, become’ + PPP + NPNOM with verb agreement are categorized as passive.

The passive voice in Estonian typically does not express the A. In cases where

the A is expressed, many different means can be used. As there is more than

one auxiliary verb and more than one possible way to express the agent, we are

dealing with a bundle of constructions rather than one single passive

construction. Hence I prefer to talk about passive constructions in this paper.

My main focus is on describing historical changes in marking A in these

constructions during the past 200 years and finding explanations for these

changes.

2.2. Agent marking in Estonian passive constructions: previous studies

The first text about expressing the A in Estonian is a manuscript, in

which Ney (1928) names the following possibilities for expressing the A in the

passive construction: the genitive, elative, and (in South Estonian) ablative

cases, and the postpositions poolt, poolest ‘by, from’, läbi ‘through’. Ney also

claims that the elative construction was used most in earlier times, but that in

the contemporary language, it was giving ground to the poolt ‘by’-construction.

Uuspõld (1970) only discusses attributive PPP-constructions which

cannot be considered passive. However, the past passive participle (PPP) also

occurs in the passive construction, thus the agent-marking devices are largely

similar in both of these. Uuspõld lists the genitive, the postposition poolt ‘by,

from’, and the elative as the main devices for marking the A.

Rajandi (1999 [1968]) mentions the same agent-marking devices as

Uuspõld does, adding the adessive to the list.

Ross (1997) has mentioned in connection with the use of local cases in

17th

-century texts that the elative is often used to mark the agent. Although her

analysis concerns also the adessive, she has not mentioned it in this function.

Torn (2006) has closely studied the use of genitive + poolt ’by’-

constructions in impersonal and passive sentences based on the corpus of

literary Estonian and has found that the poolt ’by’-construction can be used in

6

impersonal sentences with some restrictions: the noun in the poolt ’by’-

construction usually refers to a group or an institution (e.g. by the police) but

not an individual. However, in the stative passive the choice of the noun is not

as limited and refers more often to an individual.

Lindström & Tragel (2007) have discussed the use of the adessive

argument in impersonal and passive sentences and have found that in an

impersonal sentence the adessive argument expresses non-agentive roles

(place, possessor etc.), whereas in the passive and the compound tenses of the

impersonal it mainly expresses the volitional agent. Some of the sentences with

adessive topical agents form a distinctive construction – the possessive perfect

construction (cf. section 7).

Sahkai (2011) has discussed the functions of the genitive A in

participial and infinitival constructions more closely. The genitive deviates

from other agent marking means by its syntactic dependency – it cannot be an

argument on the clause level but belongs syntactically to the participle.

3. Expression of the agent in the passive construction in the 19th and

the late 20th century

In this section, I take a brief look at the agent-marking devices during

two time periods, from 1800-1850 and during the 1990s, based on the corpora.

3.1. Expression of the A in the passive construction in the 19th century

In May 2010, when this corpus query was performed, the corpus of

19th century texts contained approximately 409,000 words. This material

contained a total of 876 passive sentences. Table 1 shows the agent-marking

devices in the early 19th-century texts and the number of occurrences for each

of them.

A marking

device

No. Examples

Adessive 9

(+4*)

Nattukesse aea-ga oll-i temma-l üks raske kimp koggo-tud ….

short.GEN time-COM be-PST.3SG he/she-ADE one heavy bunch

gather-PPP

‘In a short time, a heavy bunch was gathered by him/her

and he/she was binding it up tight’ (1844-Schwelle3_3)

Elative 70 Et nemmad kül sepärrast teis-te-st sa-wad teo-tud ja naer-tud,

ei panne nemmad sedda naero siiski ennestele raskeks.

that they PART therefore other-PL-ELA get-3PL disgrace-PPP and

laugh-PPP‘Though they will therefore be disgraced and

derided by others, they do not find this laughter hard for

themselves’(1847-Kersten_87)

Genitive 8 Nemmad ep olle mitte innimesse teh-tud, waid kaswawad ja

sünniwad issi+ +suggu konna+ +karpides, mis Perlmusliks

nimmetakse.

they not be not man-GEN make-PPP

7

‘They are not made by a person but grow and are born in

particular mussels which are called pearlmussels’

(1818_Masing_60)

Comitative 14* Kui üks usklik hing nenda allandusse öi-te-ga ehhi-tud on,

siis paistab se ka temmast wälja, sest allandusse-+ +emma on

waimo waesus.

if/when one faithful soul so humility-GEN blossom-PL-COM

decorate-PPP be.3SG

‘If/when a faithful soul is decorated with the flowers of

humility, it can be seen in him/her because the mother of

humiliation is the poverty of the mind’ (847-Kersten_97)

genitive +

läbi

'through,

by'

6

(+10*)

Pärrast juttustas temma keikidele, kuidas ta omma paggana

pimmedusse-st ja waimolikko surma-st lapse läbbi oll-i ülles

ärra-tud …

… how he/she own devil.GEN darkness-ELA and spiritual death-

ELA child.GEN through be-PST.3SG up awake-PPP …

‘Later he told everyone how he was awakened from his

devilish darkness and spiritual death by a child…’ (1839-

Masing_13)

genitive +

käest

'from', lit.

'from the

hand'

2 Lambrine peab wägga murret piddama et ka umbest need järad

lammaste jure sawad, mis wannematte käest temma-le näide-

tud ja käs-tud on.

what parent.PL.GEN from he/she-ALLshow-PPP and order-PPP

be.3SG

‘The shepherd must take care that these rams also get to the

lambs which are shown to him and ordered by the parents’

(1840-Jordan_31)

poolt,

poolest

'from, by',

lit. 'from

the side

of...'

2 Selle peäle sa-i temma-le kohto poolt, temma üllekohto

pärrast, surma nuhtlus moiste-tud ja sedda ka temmale

kulutud.

this after get-PST.3SG he/she-ALL court.GEN from he/she-GEN

injustice.GEN because death.GEN penalty assign-PPP

‘After this, the death penalty was assigned to him by the

court for his injustice, and this was announced to him.’ (1847-

Kersten_117)

Agentless 751 Härjepä ei tohhi mitte söde-tud sa-da, kui saddo pöllu

pehmeks teind…

clover.NOM not must.CONNEG not feed-PPP get-INF…

‘The clover must not be fed when the rain has made the field

soft… ’ (1840-Jordan_9)

Table 1. Agent-marking devices from 1800-1850.

(*interpretation as A is ambiguous)

In the 19th century, passive constructions were used frequently, possibly

because of the German influence which was very strong before the 20th

century. From the point of view of modern Estonian usage, the amount of

8

passive sentences in the 19th-century texts seems much greater than it is

nowadays, and compared to modern Estonian the saama ‘get’-passive is much

more common. Using the verb saama as an auxiliary in passive constructions is

at least partially a German influence (Raun, Saareste 1965: 67).

As can be seen in Table 1, the A in the passive sentences has been left

out altogether in the overwhelming majority of examples – the agentless

passive is characteristic of Estonian. If the A is expressed, it is most typically

coded with the elative case (56% of clauses with explicit A). The occurrence of

all other devices is much less frequent.

The comitative case and the postposition läbi ’through’ are rather

exceptional among A-marking devices, as they are more often used for

instrument-like arguments (alanduse õitega ‘with the blossoms of humility’ in

ex. 6a). However, since they can be expressed as subjects in corresponding

active sentences (alanduse õied in 6b), they behave like other As.

(6) a. Kui üks usklik hing nenda allandusse

if/when one faithful soul so humility.GEN

öi-te-ga ehhi-tud on,

blossom-PL-COM decorate-PPP be:SG3

‘If/when a faithful soul is decorated with the blossoms of

humility’ (847-Kersten_97)

b. kui usklikku hinge ehi-vad alanduse

Obj Subj

if/when faithful.PAR soul-PAR decorate-PL3 humility.GEN

õie-d

blossom-PL:NOM

‘if/when the blossoms of humility decorate the faithful soul’

The genitive is different from other agent-marking devices since it belongs

syntactically together with the PPP and thus its position in the sentence is fixed

– always before the PPP (Sahkai 2011; see the example in the table).

3.2. Agent-marking options in modern Estonian

In the 1990s corpus of fiction texts, a selection of data approximately

equal in size to that of the corpus of 19th-century texts (roughly 2/3 of the total

1990s fiction text corpus) was included in the analysis. A total of 4636

instances of the PPP were found, of which 84 were used in passive sentences

where the A was expressed in some way (see table 2). Compared to the texts

from the early 19th century, both the total number of passive sentences and

9

passive sentences with explicit As had decreased considerably. In table 2,

agentless passive sentences have been omitted.

For comparison, data from the corpus of spoken Estonian, which

contains everyday, face-to-face conversations have also been added to the

table. Since the total amount of spoken data is considerably lower, the

numerical data are not comparable. Nevertheless, some trends can be observed.

A

marking

device

fiction spoken

corpus

Examples

Adessive 21 8 Ja mamma oli see, kes tõi veel oma pisikese

hõbedakohvri välja, mis ta-l ol-i paki-tud.

what she-ADE be-PST.3SG pack-PPP

‘And grandma was the one who brought out her small

trunk of silver which was already packed by her’

(ILU1990\ilu0273)

Elative 40 1 ja kask majaotsas kohises endiselt, nagu poleks ta oksa-

d ja tüvi saa-nud-ki riku-tud raudse-st vihma-st...

like be-COND.CONNEG it.GEN branch-PL and trunk get-

APP-PART ruin-PPP iron-ELA rain-ELA

‘and the birch was still soughing, as if its branches and

trunk had never been ruined by the hard rain’

(ILU1990\ilu0028)

Genitive 6 2 Ta ju saab aru, et need lille-d on sinu oste-tud.

… that these flower-PL be.3PL you.GEN buy-PPP

‘He understands that these flowers have been bought by

you’ (ILU1990\ilu0166)

Comitat-

ive

9 1 Ann oli koju jõudes kohe kirjutanud ning korranud oma

kutset, mis ol-i ka teis-te kodakondse-te arvamus-te-ga

kinnita-tud.

what be-PST.3SG too other-PL.GEN member_of_house-

hold-PL.GEN opinion-PL-COM firm-PPP

‘After coming home, Ann had written at once and

repeated her invitation, which was affirmed also by the

opinions of the other members of the household’

(ILU1990\ilu0262)

post-

position

poolt

8 0 Eilsed tunnistuse-d sa-i-d vahialuse-le ette loe-tud ja

tema poolt kinnita-tud.

yesterday testimony-PL.NOM get-PST-3PL arrestant-ALL

ahead read-PPP and he/she.GEN side_of affirm-PPP

‘Yesterday’s testimonies were read to the arrestant and

affirmed by him/her’ (ILU1990\ilu0224)

10

Table 2. Agent-marking devices in texts from the 1990s.

First, we can see from table 2 that compared to the early 19th century, the

modern time period has considerably fewer passive constructions even though

the texts included in the analysis were comparable. It is possible that this is a

result of the nature of fiction texts, as the passive is not a common construction

in fiction. However, it is likely that the decrease in German influence plays a

more important role. In the early 19th century the number of native Estonian

authors writing in Estonian was not very high, and the whole written culture

was under strong influence from German. Also, many of the authors whose

texts can be found in the corpus were not Estonians but Germans who had

learnt Estonian, thus they do not represent spoken Estonian at that time.

However, by the end of the 20th century the direct influence of German had

been absent for a long time and the Estonian written culture had developed,

with a strong tradition and a diverse body of native authors.

As we can see in the table, by the end of the 20th century the most

frequent agent-marking device in fiction texts had become the elative (49% of

clauses where A is expressed). However, compared to the earlier period, its use

has strong constraints which will be discussed in section 4.1.

The use of the adessive has broadened, as has the use of the

postposition poolt ’by’. Some of the agent-marking devices – the postpositions

läbi ’through’ and käest ’from, lit. from the hand’ – have disappeared

completely, as witnessed by the current material. The comitative case is still

used, but mostly in sentences where the main verb is katma ’to cover’ (ex. 7),

i.e. mainly in instrument-like functions.

(7) /…/ pruss ol-i kae-tud rohelis-te vetika-te-ga,

baulk be-PST.3SG cover-PPP green-PL.GEN seaweed-PL-COM

libe nagu saarmanahk.

slippery as otter_skin

‘ /…/ the baulk was covered with green seaweed, slippery like the skin

of an otter.’

(ILU1990\ilu0167)

In comparison with the language of fiction texts, the spoken language corpus

very rarely shows the elative and the poolt ’by’-construction used, and the

main agent-marking device in spoken language is the adessive case.

11

4. Comparison of the use of elative and adessive As in 1800-1850 and

the 1990s: subject properties

In this section, the subject properties of elative and adessive As are

compared with regard to semantic properties, topicality and control over

reflexivization.

4.1. Semantic constraints on the use of adessive and elative as agent-

marking devices

While the frequency of the adessive and elative as agent-marking

devices has not changed much – at least based on the corpora used for this

study – the semantic constraints show a greater change. In this subsection I will

concentrate on the constraints for the use of elative and adessive As during the

two time periods in question. The main characteristic under observation will be

the animacy of the A, following the animacy hierarchy described first in

Silverstein (1976), later adjusted by others (cf. Croft 2003: 130): 1st & 2

nd

person > 3rd

person > proper names > humans > non-human animates >

inanimates.

Elative

In the early 19th century, the elative was widely used for denoting the

volitional agent as well as inanimate As. Using the elative to refer to animate

As was more frequent (49 instances) than using it for inanimate As (21

instances). Following the animacy hierarchy it can be assumed that in the 19th

century, the elative could be used to refer to nearly all levels of the hierarchy.

While there are no examples of the 1st and 2

nd person pronouns (probably due

to the nature of the texts), there are examples of 3rd person pronouns (see

example 8), proper names, humans (example 9), non-human animates (10), and

inanimates (11) which can also be abstract (12). The elative can also be used to

refer to a collective A (institution) (13). By far the most frequent A referred to

with the elative was God (also Creator, Lord, Father, Holy Ghost). Jumal

’God’ alone was used as an elative A 20 times in the material (14).

(8) Et nemmad kül se++pärrast teis-te-st sa-wad teo-tud

that they PART therefore other-PL-ELA get-3PL disgrace-PPP

ja naer- tud, ei panne nemmad sedda naero siiski ennestele

and laugh-PPP,

raskeks.

‘Though they will be therefore disgraced and derided by others, they do

not find this laughter hard for themselves’ (1847-Kersten_87)

12

(9) Keik nee-d Sanna-d sa-i-d, senni kui ta räki-s,

all these-PL word-PL get-PST-3PL until then he/she talk-PST.3PL

Sikkertari-st ülles kirjo-tud.

secretary-ELA up+ write-PPP

‘All these words were written down by the secretary, while he was

speaking.’ (OWE: 1817-Holtz_9)

(10) Kui naad agga sui ussi-de-st wägga ärra++

If they but summer worm-PL-ELA very

rikku-tud on, neid ei sünni raio-da.

blemish-PPP be:PL3, they.PL.PAR not befit.CONNEG chop_down-

INF

‘But if they are very spoiled by worms, then they are not fit for

chopping down’(1840-Jordan_25)

(11) Noor-t härjapea-d ei pea ka mitte söde-ta-ma,

young-PAR clover-PAR not must.CONNEG PART not feed-IPS-SUP

kui pöld wihma-st on ärraleu-tud …

when field.NOM rain-ELA be.3SG soak-PPP

‘Young clover must not be fed [to animals] when the field is soaked

by rain.’

(1837-Knüpffer_9)

(12) Ei mahtu-nud rööm mis senne-st temma-l oll-i,

not fit-APP joy what it-ELA he/she-ADE be-PST.3SG

temma süddame-sse, mis ligarmu-st hopis täide-tud.

he/she.GEN heart-ILL what overmuch_love-ELA PART fill-PPP

‘The joy he had because of this did not fit into his heart, which was

filled with overflowing love.’

(1816-Masing_25)

(13) Agga keik nisugguse-d kauba-d

but all such-PL.NOM agreement-PL.NOM

pea-wad Kohtu-st sa-ma kinni-tud.

must-3PL court-ELA get-SUP affirm-PPP

‘But all such agreements must be affirmed by the court’ (1817-

Holtz_79)

13

(14) Olgo need kül sure-d, siis õmmetige selle wasto ei

be-JUS:PL3 these PART great-PL then however it:GEN against not

seisa, kes Jummala-st nende ülle on

stand:CONNEG who god-ELA they.PL.GEN over be.3SG

pan-tud wallitse-ma.

put-PPP govern-SUP

‘Though they are great, still [they] can’t stand against the one who is

appointed to govern them by God’ (1818_Masing_108)

Thus, the elative has no considerable semantic constraints as a marker of A in

the early 19th-century texts; it is the most ordinary and frequent way to denote

the A.

In the texts from the second period (the 1990s), however, the situation

is quite different. Only five sentences had an animate A marked with elative

case, while 34 sentences had an inanimate A, thus the elative is used mainly to

denote inanimate As. Moreover, the elative may express animate As only in a

few constructions, mainly in the passive experiencer construction.

Namely, four out of five instances of the animate As marked with

elative occurred in the passive experiencer construction as a stimulus argument

(minust ‘I-ELA’ in ex. 15, sinust ‘you-ELA’ in ex. 16). In its active counterpart

the stimulus is marked with the nominative and agrees with the verb (as ma ’I’

in ex. 15b) whereas the experiencer is marked with the partitive (kedagi in

15b). The construction as in ex. 15b represents the experiencer-object

construction where the experiencer is marked as a typical object (Lindström, to

appear). The experiencer-object construction is non-canonical in the sense that

the participant roles and grammatical relations are inverted compared to typical

or canonical transitive clauses: the most likely A (=experiencer) is marked as

the object and the (typically non-human) stimulus is marked as the subject of

the clause.

(15) a. Viieteistkümnesele on see traagika: keegi ei märka mind,

Keegi pole minu-st huvita-tud,/---/

Experiencer stimulus

anybody not_be.CONNEG I-ELA interest-PPP

‘For a 15-year-old, it is a tragedy: nobody notices me, nobody is

interested in me, nobody cares about me’ (ILU1990\ilu0002)

(15) b. Ma ei huvita kedagi.

Stimulus experiencer

I.NOM not interest.CONNEG anybody.PAR

‘Nobody is interested in me’

14

(16) Ah, kuidas ma sinu-st huvita-tud ole-n, Algi!

oh how I.NOM you-ELA interest-PPP be-1SG NAME

Oh, how I am interested in you, Algi!

Thus, the elative stimulus is used often in the passive counterpart of the

experiencer-object construction. Interestingly, in this passive counterpart the

relations of semantic roles and syntactic relations resemble a canonical

(transitive or intransitive) clause, usually described in Estonian grammars as a

normal clause (EKG II: 14, Erelt 2003: 93) or unmarked basic clause (Erelt &

Metslang 2006) – the experiencer (the more agentive participant in the event) is

the grammatical subject in this passive sentence, as it is case-marked by the

nominative and agrees with the verb (16) and thus resembles the canonical

subject more closely than the partitive experiencer in the corresponding active

sentence does. The stimulus is in the elative case as in other types of

experiencer constructions (sinust ’you-ELA’in ex. 17).

(17) Mu-l on sinu-st kahju.

I-ADE be:SG3 you-ELA sorry

‘I am sorry for you’

However, the use of the elative A (=stimulus) is not restricted to human or

animate stimuli, but also to clearly non-agentive stimuli, as millestki

’something’ in ex. (18).

(18) Henri pid-i enne töö-le tuleku-t ole-ma

Henri must-PST:SG3 before work-ALL coming-PAR be-SUP

millestki häiri-tud.

something.ELA disturb-PPP

‘Henry must have been disturbed by something before coming to work’

(ILU1990\ilu0159)

The number of predicates that allow the use of elative As like these is limited

to those fitting into the experiencer-object construction [NP(EXPERIENCER)-

PAR V NP(Stimulus)-NOM], for example huvitama ’to interest’, häirima ’to

disturb’, rõõmustama ’to gladden’, vaimustama ’to inspire, to enrapture’,

ehmatama ’to startle’, erutama ’to excite’, lõbustama ’to amuse’ etc. (cf.

Lindström, to appear, Lindström 2012).

Nevertheless, it seems that the passivized equivalents of this

construction type have developed into an independent construction type [NP-

NOM(Experiencer) be NP-ELA(Stimulus) V-PPP], since the same construction

is used with some predicates which cannot be used in the experiencer-object

construction (compare 19a and b).

15

(19) a. Passive:

Ma ole-n Esa-st sisse võe-tud.

I be-1SG Esa-ELA in take-PPP

‘I am fond of Esa’

b. Active:

*Min-d võtti-s Esa sisse.

I-PAR take-PST.3SG Esa.NOM in

≠‘Esa is dear to me’

(grammatical under literal interp.: ’Esa took me in’)

Although it was not observed from the corpus of fiction of the 1990s, Uuspõld

(1970) has mentioned another type of predicate which allows the use of elative

animate As: ümbritsema, piirama (‘to border, to surround’, ex. 20a). In this

case it is a collective A in an instrumental-like use; this is indicated by the

possibility of paraphrasing it with the comitative case (20b), a typical

instrument-marking device in Estonian. It is interesting to note that the active

equivalent of this construction type is somewhat similar to the experiencer-

object construction – the clause begins typically with a partitive NP [NP-PAR

V NP-NOM] (20c), cf. Rätsep 1978: 129). Nevertheless, the use of elative As

cannot be generalized to all passive equivalents of sentences containing the

[NP-PAR V NP-NOM] construction, as some of these can also be used with

the comitative (e.g. katma ’to cover’, ex. 7).

(20) a. Juubilar ol-i sõpra-de-st ümbritse-tud.

jubilarian.NOM be-PST.3SG friend-PL-ELA surround-PPP

‘The jubilarian was surrounded by friends'

b. Juubilar ol-i sõpra-de-ga ümbritse-tud.

jubilarian.NOM be-PST.3SG friend-PL-COM surround-PPP

‘The jubilarian was surrounded with friends'

c. Juubilari ümbritse-sid sõbra-d ja kolleegi-d.

jubilarian.PAR surround-PST.3PL friend-PL and colleage-PL

‘Friends and colleagues surrounded the jubilarian’

On the basis of the corpus of the 1990s, it appears that elative As denoting

animate participants can be used also while referring to Jumal ’God’. The

frequent use of this word in the elative case in the earlier texts gives reason to

assume that in current usage its elative use has become fixed, expecially in

connection with the predicate looma ’create’ (21).

16

(21) Inimese-d olla ju jumala-st loo-dud kahekesi

people-PL be.KVT PART god-ELA create-PPP by_two

koos ela-ma, teineteise-st rõõmu tund-ma.

together live-SUP,each_other-ELA joy.PAR feel-SUP

‘People are created by God to live together and enjoy each other’

(ILU1990\ilu0205)

Compared to animate As, the elative is used considerably more to mark

inanimate As, both concrete (22) and abstract (23). Some of them could be

described semantically as force (22). This is the most usual and the most

frequent usage of elative As in passive constructions in modern Estonian.

(22) ja kask maja otsas kohises endiselt,

nagu pole-ks ta oksa-d ja tüvi

like not_be-COND.CONNEG it-GEN branch-PL and bole

saa-nud-ki riku-tud raudse-st vihma-st...

get-APP-PART ruin- PPP iron-ELA rain-ELA

‘and the birch was still rustling, as if its branches and trunk had never

been ruined by the hard rain’ (ILU1990\ilu0028)

(23) See või-s olla tingi-tud ta mineviku-st.

it can-PST.3SG be.INF condition-PPP he/she.GEN past-ELA

‘It could have been conditioned by his past’ (ILU1990\ilu0216)

Thus, adding animate human As to the passive construction is possible in

modern Estonian only with a limited number of verbs or construction types.

Most verbs do not allow the use of elative to mark animate As; the elative

argument will then be interpreted as something else. For example, sentence 24

is only acceptable if the elative kokast ‘chef-ELA’ refers to the ingredients of

the soup.

(24) ?Supp on koka-st keede-tud.

soup be.3SG chef-ELA cook-PPP

≠‘The soup is cooked by the chef’ / =‘The soup is made of the chef’

The elative case cannot be used to mark higher animals as As either (25).

However, using the elative to express smaller animals and insects as As is

plausible if there is an unspecified number of them, just as in ex. (26) the

number of mice is unspecified and irrelevant.

17

(25) *Kont on koera-st näri-tud.

bone be.3SG dog-ELA chew-PPP

‘The bone has been chewed by the dog’

(26) Vaip ol-i hiir-te-st näri-tud.

carpet be-PST.3SG mouse-PL-ELA chew-PPP

‘The carpet was chewed by mice.’

Thus, there has been a considerable change in the use of the elative between

the early 19th century and late 20th century: the elative, which was once used

to express all kinds of As, can now only be used to mark inanimate As. In

modern Estonian, the elative case can be used to mark animate As only in

certain constructions (the passive experiencer construction, with a limited

number of predicates) and to mark smaller animals and insects in groups of

unspecified numbers.

Adessive

The constraints on the use of adessive to mark As are quite different from the

elative. I found a total of 13 instances of the adessive case being used to

express As in the 19th-century texts. All of these referred to an animate human

agent acting volitionally. Most of the instances were pronouns (27); NPs

referring to humans were rarer (28).

(27) Mei-l on jo hea koggo walmis wolista-tud.

we-ADE be.3SG PART good collection.NOM ready carve-PPP

‘We have a good amount carved already’ (1843-Schüdlöffel_1106)

(28) „Seal ta on!” ütle-s märra, jalga üllestöstes,

there it be.3SG say-PST.3SG broodmare, foot-PAR up_lift.GER,

mis eile seppa-l uest oll-i rauta-tud.

what.NOM yesterday blacksmith-ADE again be-PST.3SG shoe-PPP

‘“There it is!” said the broodmare, lifting up the foot which had been

shod again by the blacksmith yesterday.’ (1850a-Kreutzwald_lisa2_19)

During the second period under observation (the 1990s) the situation is the

same: the adessive case is only used to refer to animate human volitional

agents, which are usually expressed by a pronoun (29), on a few occasions also

by an NP referring to a human (30).

(29) einoh, ta-l on juhtme-d pan-tud noh

PART he/she-ADE be.3SG wire-PL.NOM put-PPP PART

‘Well, he has done the wiring’ (spoken)

18

(30) See on tudengi-l salaja teh-tud.

it.NOM be.3SG student-ADE secretly make-PPP

‘This has been made by the student secretly’ (ILU1990\ilu0214)

In modern Estonian, the adessive argument can only be used to refer to

humans, exceptionally also to animals depicted as having some human

characteristics such as volitional action, premeditation (31). Sentence 31 is

more acceptable if the adessive argument is located at the beginning of the

sentence in a typical subject (topic) position. The adessive case can never be

used to refer to smaller animals and inanimate agents (32-33).

(31) Koera-l on kont näri-tud./ ?Kont on koeral näritud.

dog-ADE be.3SG bone.NOM bite-PPP

‘The bone has been chewed by the dog’

(32) *Pluus on koi-de-l söö-dud. / *Koidel on pluus söödud.

shirt NOM be.3SG moth-PL-ELA eat-PPP

‘The shirt has been eaten by moths.’

(33) *Puu on külma-l võe-tud.

tree.NOM be.3SG cold-ADE take-PPP

‘The tree is frozen’, lit. ‘The tree has been taken by cold’

In conclusion, it appears that during the last 200 years, a specialization of the

means of marking A has taken place. In modern Estonian, the elative is mainly

used to express inanimate As such as stimulus, instrument, force, etc., located

at the end of the animacy hierarchy, whereas the adessive is used to express

volitional, animate As. Comparing the two time periods, it is clear that in the

early 19th century, this division of labor did not yet exist as both the elative

and adessive could be used with animate volitional As. It developed at some

time between the two periods.

Agentivity has been considered one of the properties of (canonical)

subjecthood – subjects are typically agents or initiators of events (Keenan

1976). Therefore, in modern Estonian the adessive A can be attributed at least

one subject-like property that the elative A does not have – agentivity, i.e. the

ability to be an energy source and act volitionally.

In the next sections I take a closer look at some other subject properties

of adessive and elative As in the passive construction such as topicality

(section 4.2.1) and control over reflexivization (4.2.2).

4.2. Other subject properties of adessive and elative As

4.2.1. Word order and topicality

19

Syntactic position can shed light on the possible behavior of the

argument as a non-canonical subject (see e.g. Andrews 1985, Onishi 2001,

Haspelmath 2001a, Barðdal & Eythórsson 2003, Barðdal 2006).

The basic word order of Estonian is SVO (Tael 1988). In interrogatives

and in some embedded clauses, the verb tends to occur in the final position

(Lindström 2005, 2006). In main clauses starting with the adverbial, the verb

usually takes the second position and the subject occurs in the 3rd position

(Remmel 1963, Tael 1988, Lindström 2005) – the V2-rule, well known from

Germanic languages, operates in Estonian as a result of long-lasting contacts

with German. In spite of the restriction, Estonian word order is relatively free

and very sensitive to the information structure (Lindström 2005).

There is a strong tendency in languages for subjects to also be topics

(Lambrecht 1994: 132, Chafe 1976). This is also the case in Estonian: the

nominative subject is typically in the beginning of the clause and functions as a

topic in basic non-inverted clause types.

Passivization is the syntactic process through which the P (normally the

object of the active clause) is foregrounded and the A (subject of the active

clause) is backgrounded. Foregrounding can be done syntactically by

promoting the P argument to the subject and backgrounding by demoting the A

argument to oblique or eliminating it from the clause. The P argument is also

foregrounded pragmatically: the passive prefers P to be the topic, while the

active prefers A to be the topic (see Comrie 2008: 9, Keenan & Dryer 2007).

Thus the most typical position of the P in passive clauses is in the beginning of

the clause – in the topic position.

If the A acquires the features of the subject (non-canonical subject), we

can presume that it also tends to take the topic position. Therefore in the

passive clause it may happen that two arguments – the P marked by the

nominative (syntactic subject) and the A marked by the oblique - may compete

for the topic position.

In the next paragraphs, the order of the A and P is discussed (see table

3). The position of the verb is not taken into account, as it varies also because

of V2 (Tael 1988, Lindström 2005)

time period 1800-1850 1990s

order of A and P PA AP Total PA AP Total

elative A 63 5 68 37 3 40

adessive A 6 6 12 12 15 27

Table 3. The ordering of A and P arguments in passive clauses.

Table 3 shows that in sentences with an elative argument the P is

overwhelmingly located before the A, even if the P is inanimate and the A is

animate. The P argument is typically located at the beginning of the sentence

20

(ex. 34). Thus, sentences with the elative exhibit the word order characteristic

of passive sentences. This is common to both the 19th-century and late 20th-

century texts; no significant changes have taken place. Elative A can occur at

the beginning of the sentence before P (35), but this is rare. It is important to

note that all examples of AP word order with elative A were found in sentences

where the A denotes an animate human participant (or God, as in 35).

(34) Kiskja-d on Jummala armu-st teis-te loma-de

predator-PL be.3PL god.GEN grace-ELA other-PL.GEN animal-PL.GEN

hea-ks lo-dud.

good-TRNSL create-PPP

‘Predators are created by God’s grace for the good of other animals’

(1818_Masing_163)

(35) /…/ Jssanda-st oll-i temma-l üks suur koorm

god-ELA be-PST.3SG he/she-ADE one big burden

peäle pan-dud, mis temma omma ello otsa-ni

on put-PPP, what he/she own life.GEN end-TER

kannatlikkult ja hea mele-ga kand-is.

patiently and good mind-COM carry-PST.3SG

‘A great burden had been put onto him/her by God, which he/she bore

patiently and happily until the end of his/her life.’ (1847-Kersten_1719)

The adessive A can be located equally before or after the P in both time

periods; there have been no relative changes in frequency of position of

adessive arguments during the past 150-200 years.

In sentences with adessive arguments (AdA), there is a difference in

interpretation between sentences beginning with A and with P: if the AdA is

located at the beginning of the sentence (before P), the sentence is more likely

to describe the (past) action of the A (27, 36). If P is located before A, the

sentence describes the state of P (which is the result of the past action of the A)

rather than A’s past action itself (37). Thus, the adessive A at the beginning of

the sentence clearly acts as the topic of the sentence, like a canonical subject in

a basic sentence (Keenan 1976: 318), and has an effect on the interpretation of

the whole sentence as an A-oriented event. Therefore, a passive sentence

beginning with AdA is similar to an active sentence – AdA behaves like a

“figure within the profiled relationship” (Langacker 1991: 316).

21

(36) Ruiso-l, muide, ol-i alati raamat kaasa võe-tud.

Ruiso-ADE by_the_way be-PST.3SG always book.NOM with

take-PPP

‘Ruiso, by the way, had always taken the book with him’

(ILU1990\ilu0021)

(37) See kena kahetoaline ol-i ju

this nice two_room.NOM be-PST.3SG PART

nei-l koos oste-tud /…/

they-ADE together buy-PPP

‘This nice two-room apartment was bought by them together’

(ILU1990\ilu0172)

The adessive A also behaves like a subject in a passive sentence by following

the V2 rule. If the adverbial is placed at the beginning of the sentence, the

subject moves from the beginning of the sentence to the position after the verb

(subject-predicate inversion). The same happens to the adessive A (but not to

the elative A): AdA can be inverted and moved to the 3rd

position in the clause

(38). At the same time the AdA continues to act as a topic (just like adessive A

in the beginning of the sentence): in example 38 the focus is on the action of A,

not the state of P. Moreover, P is a new referent in the discourse, it is affected

by the action of the A (like a typical object of a transitive sentence) and is

located in a typical object position in the sentence, although it is a syntactic

subject (by nominative case marking and agreement).

(38) Aga Romeo kõrvalegi on mu-l oma inimene sea-tud.

but Romeo.GEN next be.3SG I-ADE own person set-PPP

‘But I have also set my own person beside Romeo.’ (ILU1990\ilu0012)

4.2.2. Reflexivization

A well-known feature which is usually considered to be a subject

property is control over reflexivization – the reflexive pronoun is coreferential

with the subject rather than any other elements of the clause (cf. Keenan 1976,

Andrews 1985, Eythórsson & Barðdal 2005, Barðdal 2006). This seems to be

one of the most universal subjecthood tests – no language has been found

where the object argument only (and no A argument) controls reflexivization

(Onishi 2001: 15). Thus, it is the subject which controls the use of the reflexive

pronoun oma ’own’ or enese~enda ’self’ in Estonian.

Reflexivization is also suitable for testing the subjecthood of the A

argument in passive constructions where two subject candidates appear: a

nominative P argument and an elative or adessive A argument. As the corpus

data is not sufficient to draw any conclusions here, in the following section the

22

reflexivization test is conducted mainly on the basis of the author’s own

intuitions. Thus the results represent only the contemporary situation and say

nothing about the situation in the 19th century.

In agentless passives, the foregrounded P argument (syntactic subject,

lapsed in ex. 39) controls the use of the possessive reflexive pronoun oma

‘own, self’.

(39) Lapse-di ol-i-d omai laua taha istu-ma pan-dud.

child-PL.NOM be-PST-3PL own desk.GEN behind sit-SUP put-PPP

‘Children were seated at their desk’

Elative

In passive clauses with an elative A (ElA), the only controller over

reflexivization seems to be the nominative P argument, even if the A argument

refers to animate human As, as happens mainly with experiential predicates

(40). Even the topicalization of A does not change the interpretation of the

reflexive pronoun (41).

(40) Tai pole Etheli-st omai isekuse

he/she.NOM not_be.CONNEG Ethel-ELA own selfishness.GEN

tõttu kuigivõrd huvita-tud.

because very interest-PPP

‘He is not very interested in Ethel because of his selfishness.’

(41) Etheli-st pole tai omai isekuse

Ethel-ELA not_be:CONNEG he/she own selfishness.GEN

tõttu kuigivõrd huvita-tud.

because very interest-PPP

‘He is not very interested in Ethel because of his selfishness.’

Adessive

Generally, the adessive A (AdA) seems to control the use of the reflexive

pronoun, which marks the possessor of P (42). P, on the contrary, cannot

control the reflexive pronoun which syntactically belongs to AdA (43), thus,

AdA is higher on the hierarchy of grammatical relations than the P argument in

this construction.

(42) Aga Romeo kõrvalegi on mu-li omai inimene sea-tud.

but Romeo.GEN next be.3SG I-ADE own person set-PPP

‘But I have also set my own person beside Romeo./But next to Romeo,

my own person is set by me’ (ADE ILU1990\ilu0012)

23

(43) *See lapsi ol-i omai ema-l alati hästi riieta-tud.

this child be-PST.3SG own mother-ADE always well dress-PPP

‘This child has always been dressed well by his/her mother’

If we compare clauses where both the AdA and P arguments might be potential

antecedents of the possessive reflexive pronoun, the AdA controls the use of

the reflexive pronoun (44). However, this is not without exceptions. Example

45 is ambiguous: the most likely interpretation is a), but b) is possible as well.

Even the topicality of P does not prevent this ambiguity (46). Context seems to

play a crucial role here; with the help of context, it is also clear that in example

47, oma refers back to the table, and not to the mother, because typically the

table has its own place, but not the mother.

(44) Ema-li ol-i leib juba omai kotti piste-tud.

motheri-ADE be-PST.3SG bread already owni bag.ILL put-PPP

‘Mother had put the bread into her bag’

(45) Ta-li ol-i-d lapse-dj omaij laua taha istu-ma pan-dud.

he/she-ADE be-PST.3PL child-PL own desk.GEN behind sit-SUP put-

PPP

a) ‘She had seated the children at her desk.’

b) ‘She had seated the children at their desk.’

(46) Lapse-di ol-i-d pan-dud ta-lj omaij

child-PL be-PST.3PL put-PPP he/she-ADE own

laua taha istu-ma.

desk.GEN behind sit-SUP

‘The children were seated by her at her/their desk.’

(47) Ema-l ol-i laudi omai koha peale tõste-tud.

mother-ADE be-PST.3SG table own place on lift-PPP

‘Mother has moved the table to its place’

Thus, AdA generally controls the reference of the reflexive pronoun, but this

depends on the context; sometimes also the P argument may control

reflexivization.

4.3. Conclusion: properties of the subject

During the period between the early 19th and the late 20th century there

was a change in the marking of As in passive sentences in Estonian. While in

the 19th century the main agent-marking device was the elative, by the end of

the 20th century considerable restrictions had arisen in its use; now the use of

24

the elative is restricted to inanimate As. At the same time, usage of the adessive

argument, with animate volitional As only, has increased.

In addition to animacy, the adessive A has other properties of the

subject that the elative argument does not have in modern Estonian. One of the

most important of these is topicality: AdA is located in the topic position in a

sentence considerably more often than the elative argument, and similarly to

nominative (canonical) subjects, it follows the word-order changes caused by

the V2 principle. The topical AdA affects the interpretation of the whole

sentence as an A-oriented event, despite its passive morphology. Such a

sentence is thus similar to the basic sentence with a nominative subject. Elative

A – even if it occasionally occurs in a topic position – does not have this

property. In addition, AdA controls reflexivization, which is another property

that the elative does not have. We may ask why the use of the adessive as an

agent-marking device has increased and the use of the elative has decreased;

why has the adessive argument in a passive construction acquired more subject

properties than the elative? In the next sections, I will briefly look at two points

that may explain this: 1) the semantics of the adessive and the elative and their

use in different constructions and 2) the rise and spread of the possessive

perfect construction in Estonian and neighbouring languages.

5. Semantics of the adessive and the elative case

One possible explanation for the increase in the use of the adessive and

decrease of the elative as agent-marking devices might come from their

semantics. Both the adessive and the elative are very polysemous cases and can

be used to mark different semantic functions; both are also originally locative

cases.

The elative is one of the three internal locative cases, marking

separation (‘away from X’) or source (‘from’). Related meanings where this

component can easily be seen are source (48), material (49), partitive (part-

whole relations, 50), time (51), and cause (52), among others. A more abstract

use is marking the stimulus with the elative – the connection to the 'source'

meaning is not so obvious (53).

(48) source: Poiss tuleb kooli-st.

‘The boy comes from school’

(49) material: puu-st laud

‘the table made of wood’

(50) part-whole relations: üks poiste-st

‘one of the boys’

25

(51) temporal: Esmaspäeva-st hakkab sadama.

‘It will rain starting from Monday’

(52) cause: Käed läksid külma-st kangeks.

‘The hands were going numb with cold’

(53) Stimulus: Ma sain selle-st aru.

'I understood it' [lit.: ’I got understanding from it’]

(54) Inanimate A: Puu on külma-st võe-tud.

‘The tree is frozen’, lit ’taken by the cold’

Cause is especially closely related with inanimate As and sometimes it is even

difficult to distinguish cause and inanimate A superficially (compare 52 and

54). In both cases, the elative marks the source of the energy which causes the

change, the main difference being that arguments with the role of cause

typically do not occur as subjects in the active clause.

Thus, the elative case should be an appropriate means for marking the

A (including volitional agent), as it marks the source – and also the source of

energy. Other agent-marking devices in Estonian (the poolt-construction, the

ablative in South Estonian) also have the same meaning component (source,

directionality).

However, the meaning ‘source’ is present also in the German von-

construction (55), hence it has probably also influenced A marking in Estonian

passive constructions.

(55) Der Brief wird von mir geschrieben.4

‘The letter is being written by me.’

As German had a strong influence on Estonian during the first observed

period (1800-1850), there is a reason to assume that the use of the elative as a

marker of A in Estonian passive constructions is a contact-induced

phenomenon which is based on the SOURCE-metaphor and is strengthened by

the German von-construction. Also, Ross (1997) has explained the use of the

elative as a marker of agent in 17th

-century texts with German influence. As the

German von-construction typically refers to animate As (as in 55), it obviously

influenced the widespread use of the elative in referring to animate As during

the first observed period. Since then the German influence on Estonian has

decreased, and also the use of the elative has become restricted only to

inanimate As.

4 Source of the example: http://german.about.com/library/weekly/aa012901a.htm

26

The adessive, on the contrary, has a stative meaning which does not

include directionality. It is hard to find any semantic motivation why the

adessive is used to mark a volitional agent as it typically does in passive

constructions. The adessive is an external locative case, meaning ‘on the

surface/top of X’ (56) which can also be used to mark event time (57).

However, compared to the elative it has more grammaticalized functions, e.g.

possessor (58), experiencer (59), deontic agent (60), etc. (cf. Vainik 1995,

Lindström & Tragel 2006, Klavan et al. 2011 etc.) As we can easily see from

these examples, all the constructions used in examples 58-60 are based on

Location schema (56).

(56) Location

Riiuli-l on tolmu.

Shelf-ADE be.3SG dust.PAR

‘There is some dust on the shelf’

(57) Temporal (event time):

Esmaspäeva-l saja-b.

Monday-ADE rain-3SG

‘It is/will be raining on Monday’

(58) Possessor

Jaani-l on väike õde.

Jaan-ADE be.3SG little sister

‘Jaan has a little sister’

(59) Experiencer

Jaani-l on paha.

Jaan-ADE be.3SG bad

‘Jaan feels sick’

(60) Deontic agent in modal constructions

Mu-l on vaja tööta-da.

I-ADE be.3SG need work-INF

‘I have to work’

Thus, we can conclude that, semantically, the elative seems to be an

appropriate means for marking the A, as it has the meaning component

‘source’, which can also be used as a source of energy (agent). The adessive is

a polysemous static case and its different uses in different constructions are

more grammaticalized. As it also marks the deontic agent (in modal

constructions) and experiencer, the agent-like use can motivate its spread to the

27

passive constructions. As we can see in the next section, the adessive argument

has some subject properties also in other constructions.

6. Subject properties of the adessive argument in other constructions

The adessive argument has subject properties in many other

constructions as well. For example, the adessive argument controls the use of

the reflexive pronoun in the possessive construction (61, Erelt & Metslang

2006) and in the experiencer construction (62).

(61) Mu-l on oma venna-ga ühine maja.

I-ADE be.3SG own brother-COM common house

‘I share a house with my brother’

(62) Mu-l on enda-st kahju.

I-ADE be.3SG self-ELA pity

‘I feel sorry for myself’

According to Seržant (to appear), the use of the adessive argument in Estonian

experiencer clauses (as in ex. 62) is similar to what is found in Latvian and

Lithuanian, where the same constructions exploit the dative to mark the

experiencer; at the same time, Russian uses either the dative or (in some

constructions) the spatial schema (Russian u menja is a counterpart of the

Estonian adessive argument mul ‘on me’). The dative-like experiencers have

spread widely in the Eastern Circum-Baltic area; predicate classes where

dative-like experiencers are used show a significant structural parallelism in

Eastern Circum-Baltic languages (Seržant, to appear). At the same time,

dative-like experiencers are not characteristic of Standard Average European

(SAE) but are used more often on the periphery of SAE; in central SAE

languages, agent-like (nominative) experiencers dominate. (Haspelmath 2001a)

Thus, the use of adessive non-canonical subjects seems to be an areal feature.

In addition to the experiencer constructions, the adessive agent

argument is used in many modal constructions where it controls subject

deletion in infinitives (in 63, deletion of mina ‘I’ from the infinitival

construction endast kirjutada) and reflexivization (endast ‘from (my)self’ in

63).

(63) Mu-li on vaja enda-sti kirjuta-da.

I-ADE be.3SG need self-ELA write-INF

‘I have to write about myself’

Thus, adessive arguments are used as non-canonical subjects in many

constructions, and this may be the reason why the AdA has obtained the

behavioral properties of the subject in passive constructions as well, though

28

still in the early 19th century, the elative A was more common and more

frequent.

7. The spread of the possessive perfect

As we can see, a passive sentence with adessive A behaves in many

ways differently from other passive sentences: the adessive is typically

topicalized and bears some subject properties, and in this case the overall

construction designates the completion of the action of A rather than the state

of the P (which is typical of passives). Therefore, we have also called it the

possessive perfect construction (PPC; cf. Lindström & Tragel 2007, 2010).

The PPC is in many ways similar to the have-perfect found in several

European languages – it is based on the possessor construction (c.f. Heine &

Kuteva 2006) but uses different morphosyntactic means from the SAE

languages. The possible development of this construction in Estonian has been

discussed by Lindström & Tragel (2010). The PPC has evolved from a passive-

like construction where the topical (external) possessor has been reinterpreted

as an agent, e.g. in (64) the adessive argument sul may be interpreted as a

possessor (older interpretation) or an agent (newer interpretation). From there

the interpretation as an agent has been extended to intransitive verbs, which

cannot be used in the passive construction (examples 65-66). Thus, the

possessive perfect construction can also be used with intransitive verbs and it

cannot be simply considered a subcategory of the passive construction.

(64) Kas su-l on pilet oste-tud?

Q you-ADE be.3SG ticket.NOM buy-PPP

‘Have you bought a ticket?’

(65) Mu-l on poe-s käi-dud.

I-ADE be.3SG shop-INE go-PPP

‘I have done the shopping’ / ‘I have been to the store’

(66) Su-l on juba aasta otsa seal ol-dud,

you-ADE be.3SG already year long there be-PPP

peaks olema küll hollandi keel suus.

‘As you have been there for a year already, you should speak Dutch.’

In Estonian linguistics, sentences like in ex. 64 are usually regarded as

belonging to the passive paradigm (Pihlak 1993: 81, Rajandi 1999, Erelt (ed.)

2003: 102-103). Some studies have treated it as a separate construction

(Meerwein 1994, Holvoet 2001a, Lindström & Tragel 2007, 2010). Lindström

& Tragel 2010 have shown that the construction is not always clearly

distinguishable from the passive. The main difference concerns word order and

29

information structure: in passive clauses, typically the P argument is in the

topic position, while in the possessive perfect construction, the AdA is topical

(and P can be missing). As the distinction of the PPC and passive is not clear-

cut and using information structure as a main distinguisher may be problematic

(Erelt, to appear), the distinction is not made in the paper.

Nevertheless, it is rather obvious that the use of AdA in passive and

passive-like constructions in Estonian has been influenced by the PPC in

surrounding languages. As has been claimed, similar constructions exist in e.g.

North-Russian, Latvian, Votic, etc (Seržant 2012), i.e. are characteristic to

Eastern Circum-Baltic languages (Seržant 2012). According to Seržant, the

epicenter of the PPC in this area has been North-Russian, where the

construction is the most grammaticalized (Seržant 2012: 381). As Timberlake

has shown (1976), in North-Russian dialects the locative construction u + gen

is used to mark the agent in a possessive perfect construction – it is the

semantic equivalent of Estonian adessive arguments (’on; on the top of)’.

Additionally, the North-Russian A has many subject properties (Timberlake

1976, Seržant, this volume), similarly to Estonian. The possessive perfect

construction exists and exploits the adessive as an Agent-marking device in

some other Baltic-Finnic languages as well, for example in Ingrian (67) and

Veps (68), but not in Finnish.

(67) Ingrian5:

Minu-n tüdö-ll oppi on lobete-ttu.

I-GEN daughter-ADE school be.3SG finish-PPP

‘My daughter has finished school’

(68) Veps:

Ol’-i tsar’i-l sauba-tud t’ürm-ha zmei ulan.

be-PST.3SG tsar-ADE bolt-PPP prison-ILL snake Ulan

‘The snake Ulan had been imprisoned by the tsar’ (Grünthal 2003: 145)

The construction is also present in Latvian (a Baltic language) and in Livonian

(a Baltic-Finnic language which has been under the strong influence of

Latvian), but in these languages, the dative case has acted as the counterpart of

the Estonian adessive. (The dative is also used to mark the possessor in the

possessive construction).

(69) Latvian

Viņam viss jau bija izteikts.

he:DAT all:NOM:SG:MASC be:PST:3SG say:PPP:SG:MASC

‘He had already said everything’ (Holvoet 2001a: 375)

5 Example is obtained from a translation test, on the Soikkola peninsula (Valjanitsa, informant

Aleksandr Sergejev, 23.07.2009).

30

(70) Livonian

kouv mä’ddõn vo’ĺ gōstõt mōdõ tǟdõks

well we.DAT be.3SG belt_down land.PAR full.TRNSL

‘We have fulfilled the well by the land’ (Meerwein 1994: 172)

It is notable that in Veps and Livonian, the elative has also been used as a

marker of A (71-72), but at least in Veps (according to Grünthal 2003: 145) the

use of the elative in this function is quite infrequent and atypical. Thus, the

adessive dominates also in Veps.

(71) Veps

leśk om mužika-spei jät-tut

widow.NOM be.3SG husband-ELA leave-PPP

‘The widow has been left by her husband’ (Grünthal 2003: 146)

(72) Livonian

sǟl um võ-nd ikš kurēst apsǟs-tõd neitst

there is be-APP one devil.ELA possess-PPP maid

‘There was a maid who was possessed by the devil’

(Grünthal 2003: 147)

Thus, it seems that the elative is or has been a means of A marking also in

Veps and Livonian and it has been replaced by the adessive (in Veps) or dative

(Livonian), but there is no evidence in these languages of the kind of

semantics-based division of functions that has taken place in Estonian during

the past 200 years. There is a reason to assume that in both Veps and Livonian

the increase of the use of adessive A has happened at least partially because of

the contacts with Russian (in the case of Veps) or Latvian (Livonian), i.e. with

Eastern Circum-Baltic languages.

It seems possible that the reason the adessive dominates and the elative

has lost ground as an agent-marking device in passive constructions in

Estonian (and probably also in Veps and Livonian) is the spread of the PPC

construction in Eastern Circum-Baltic languages. On the one hand, Estonian

adessive A is a semantic counterpart of the same construction in Russian and

other languages in the Eastern Circum-Baltic area; on the other hand, AdA has

some subject behavioral properties already in these languages.

8. Conclusion

A is not commonly expressed in the Estonian passive, but when it is

expressed there are several ways to do it. The two agent-marking devices

discussed in this paper – the elative and adessive – have gone through a great

deal of change since the early 19th century: while the elative was the main

agent-marking device in the passive sentence in the 19th century, by the end of

31

the 20th century it could only be used to refer to inanimate As. In the early

19th century, the adessive was rarely used to express the A, but by the end of

the 20th century its use had become more frequent. The use of the adessive A

is especially profuse in spoken Estonian. However, the adessive is used only

for marking animate volitional agents.

In addition to the specialization in animate volitional agents, adessive A

has gained some other subject properties. Above all, these include topicality (a

tendency to occur in the topic position and thus affect the interpretation of the

whole sentence) and control over the use of the reflexive pronoun. The elative

A does not have these properties.

There are several reasons for these changes in marking the A. Firstly,

this change is likely to have been caused by the subject-like properties of the

adessive argument in other constructions – the adessive also tends to behave as

a non-canonical subject in possessive constructions, experiencer constructions,

and modal constructions. The use of the adessive in these constructions is

highly grammaticalized, and its meaning has shifted away from the original

locative meaning ‘on, on the top of’. Although the elative is also a polysemous

case, its meanings in different constructions are more closely connected to the

original ‘source, separation from’ meaning.

Secondly, the elative is a semantic counterpart of the German von-

construction and this is a fact which may have affected the large use of the

elative as an A marking device in the beginning of the 19th

century. At that

time, the Estonian literary language was heavily influenced by German. By the

end of the 20th

century, German influence on Estonian had weakened

remarkably, and this may serve as one of the reasons why the use of the elative

has become restricted such that nowadays it may refer mainly to inanimate As,

low in the animacy hierarchy, i.e. untypical As.

Thirdly, the spread of the adessive argument is undoubtedly connected

to the rise and spread of the possessive perfect construction in Eastern Circum-

Baltic languages.

Thus, the elative and adessive as markers of the A in Estonian passive

sentences have undergone a change which can be described as a specialization

of functions. However, this change cannot be explained with a single factor but

rather with a bundle of factors, above all language contacts. Which is, of

course, not surprising at all.

Abbreviations

1,2,3 person

A agent

AdA adessive argument

ADE adessive

ALL allative

32

APP active past participle

COM comitative

COND conditional

CONNEG connegative

DAT dative

JUS jussive

ElA elative argument

ELA elative

GEN genitive

GER gerundive

ILL illative

INE inessive

INF infinitive

IPS impersonal

KVT quotative (evidential)

MASC masculinum

NEG negation

NOM nominative

P patient

PTCL particle

PL plural

PPP passive past participle

PAR partitive

PST simple past

Q question particle

SG singular

SUP supine

TER terminative

TRNSL translative

REFERENCES

Andrews, Avery 1985. “The major functions of the noun phrase.ˮ In Language

typology and syntactic description ed. by T. Shopen, 62–154. Vol. 1:

Clause structure. Cambridge: Cambridge University Press.

Barðdal, Jóhanna & Thórhallur Eythórsson 2003. “The change that never

happened: the story of oblique subjects.ˮ – Journal of Linguistics 39, 439–

472.

Barðdal, Jóhanna 2006. “Construction-specific properties of syntactic subjects

in Icelandic and German.ˮ – Cognitive Linguistics, 17–1, 39–106.

33

Beck, David 2000. “Semantic agents, syntactic subjects, and discourse topics:

How to locate Lushootseed sentences in space and time.ˮ – Studies in

Language 24:2, 277–317.

Blevins, James P. 2003. “Passives and Impersonals.ˮ – Journal of Linguistics

39: 473–520.

Chafe, Wallace L. 1976. “Givenness, contrastiveness, definiteness, subjects,

topics, and point of view.ˮ In Subject and Topic ed. by Charles N. Li, 25–

55. New York: Academic Press.

Comrie, Bernard 2008. “What is a Passive?ˮ In Studies in Voice and

Transitivity ed. by Z. Estrada Fernández, S. Wichmann, C. Chamoreau, A.

Álvarez González, 1–18. München: Lincom.

Croft, William 2003. Typology and universals. Second edition. Cambridge:

Cambridge University Press.

EKG II = Erelt, Mati, Reet Kasik, Helle Metslang, Henno Rajandi, Kristiina

Ross, Henn Saari, Kaja Tael & Silvi Vare. 1993. Eesti keele grammatika

II: Süntaks. Lisa: Kiri. Tallinn, Estonia: Eesti Teaduste Akadeemia Keele

ja Kirjanduse Instituut.

Erelt, Mati 1979. Eesti lihtlause probleeme. Tallinn: Eesti Raamat.

Erelt, Mati 2003. “Syntax.ˮ In Estonian Language ed. by M. Erelt, 93–129.

Linguistica Uralica Supplementary Series. Vol. 1. Tallinn: Estonian

Academy Publishers.

Erelt, Mati 2004. “Lauseliigendusprobleeme eesti grammatikas.ˮ In

Lauseliikmeist eesti keeles ed. by L. Lindström, 7–15. Tartu Ülikooli eesti

keele õppetooli preprindid 1. Tartu.

Erelt, Mati, to appear. Eesti keele lauseõpetus. Sissejuhatus. Öeldis.

Manuscript.

Erelt, Mati & Helle Metslang 2006. “Estonian Clause Patterns — from Finno-

Ugric to Standard Average European. ˮ – Linguistica Uralica 4: 254–266.

Grünthal, Riho 2003. Finnic adpositions and cases in change. Suomalais-

ugrilaisen Seuran toimituksia 244. Helsinki: The Finno-Ugric Society.

Haspelmath, Martin 2001a. “Non-canonical marking of core arguments in

European languages.ˮ In Non-canonical marking of subjects and objects

ed. by A.Y. Aikhenvald, R.M.W. Dixon, M. Onishi, 53–83.

Amsterdam/Phildadelphia: Benjamins.

Haspelmath, Martin 2001b. “The European linguistic area: Standard Average

European.ˮ In Language typology and language universals: An

international handbook ed. by M. Haspelmath, E. König, K. Oesterreicher

& W. Raible,1492–1510. Vol. 2. New York: Walter de Gruyter.

Heine, Bernd & Tania Kuteva 2006. The Changing Languages of Europe.

Oxford: Oxford University Press.

Holvoet, Axel 2001a. “Impersonals and passives in Baltic and Finnic.ˮ In The

Circum-Baltic Languages. Typology and contact ed. by Ö. Dahl & M.

Koptjevskaja-Tamm, 363–389. Vol. 2. Grammar and typology.

Amsterdam/Phildadelphia: Benjamins.

34

Holvoet, Axel 2001b. Studies in the Latvian Verb. Kraków: Wydawnictwo

Uniwersytetu Jagiellońskiego.

Keenan, Edward L. 1976. “Towards a universal definition of “subject”.” In

Subject and Topic ed. by Charles N. Li, 303–333. New York: Academic

Press.

Keenan, Edward L. & Matthew S. Dryer 2007. “Passive in the world’s

languages.ˮ In Language Typology and Syntactic Description ed. by

Timothy Shopen, 325–361. Second edition. Vol. I: Clause Structure.

Cambridge: Cambridge University Press.

Klavan, Jane; Kaisa Kesküla, Laura Ojava 2011. “The division of labour

between synonymous locative cases and adpositions: the Estonian

adessive and the adposition peal ’on’.” In Studies on Case, Animacy and

Semantic Roles, ed. by Seppo Kittilä, Katja Västi, Jussi Ylikoski, 111-134.

Amsterdam/Phildadelphia: Benjamins.

Koks, Helen 2004. “Subjekti ja objekti käitumisreeglid komplekslauses.ˮ In

Lauseliikmeist eesti keeles ed. by L. Lindström, 34–39. Tartu Ülikooli

eesti keele õppetooli preprindid 1. Tartu.

Lambrecht, Knud 1994. Information structure and sentence form. Topic, focus

and the mental representations of discourse referents. Cambridge:

Cambridge University Press.

Langacker, Ronald W. 1991. Foundations of Cognitive Grammar. Vol. II,

Descriptive Application. Stanford: Stanford University Press.

Lindström, Liina 2005. Finiitverbi asend lauses. Sõnajärg ja seda mõjutavad

tegurid suulises eesti keeles. Dissertationes philologiae estonicae

universitatis tartuensis 16. Tartu: Tartu Ülikooli Kirjastus.

Lindström, Liina 2006. “Infostruktuuri osast eesti sõnajärje muutumisel.ˮ –

Keel ja Kirjandus, 11, 875–888.

Lindström, Liina & Ilona Tragel 2006. “Mul on saba. Adessiivsest

omajakonstruktsioonist.ˮ In Finest Linguistics ed. by K. Kerge & M.-M.

Sepper, 386–399. Proceedings of the Annual Finnish and Estonian

Conference of Linguistics.

Lindström, Liina & Ilona Tragel 2007. “Eesti keele impersonaali ja

seisundipassiivi vahekorrast adessiivargumendi kasutuse põhjal.ˮ - Keel

ja Kirjandus 7, 532–553.

Lindström, Liina & Tragel, Ilona 2010. “Possessive Perfect Construction in

Estonian.ˮ – Folia Linguistica 44/2 , 371–400.

Lindström, Liina 2012. “Tundekausatiivikonstruktsioon eesti moodi.” – Keel ja

Kirjandus, 1, 30-47.

Lindström, Liina, to appear. “Between Finnic and Indo-European: Variation

and change in the Estonian experiencer-object construction.ˮ In

Diachronic Typology of Non-canonical Subjects ed. by Ilja A. Seržant &

Leonid Kulikov. Amsterdam/Phildadelphia: Benjamins.

35

Meerwein, Georg 1994. “Einige Anmerkungen zu Gemeinsamkeiten in den

Tempussystemen der Sprachen des Ostseeraums.ˮ – Linguistica Uralica,

3, 168–176.

Metslang, Helena, to appear. “Coding and behaviour of Estonian Subjects.” –

Submitted to Journal of Estonian and Finno-Ugric Linguistics.

Ney, A. 1928. Kentütlev tehtaviku tegijanimetuseks! – Manuscript in the

Archive of the Estonian Dialect and Kindred Languages, University of

Tartu.

Onishi, Masayuki 2001. “Non-canonically marked subjects and objects:

Parameters and properties.ˮ Non-canonical marking of subjects and

objects ed. by A.Y. Aikhenvald, R.M.W. Dixon, M. Onishi, 1–51.

Amsterdam/Phildadelphia: Benjamins.

Pajusalu, Renate, this volume. “Hidden subjects in conversation: Estonian

personless verb forms as referential devices.“

Pihlak, Ants 1993. A comparative study of voice in Estonian. Eesti Sisekaitse

Akadeemia toimetised. Tallinn.

Rajandi, Henno 1999 [1968]. Eesti impersonaali ja passiivi süntaks. Eesti

Keele Instituudi toimetised 3. Tallinn.

Raun, Alo & Andrus Saareste 1965. Introduction to Estonian Linguistics. Ural-

Altaische Bibliothek XII. Wiesbaden: Otto Harrassowitz.

Remmel, Nikolai 1963. “Sõnajärjestus eesti lauses.” – In Eesti keele süntaksi

küsimusi. Keele ja Kirjanduse Instituudi uurimused VIII. Tallinn: Eesti

Riiklik Kirjastus. 216‒389.

Ross, Kristiina 1997. “Kohakäänded Georg Mülleri ja Heinrich Stahli eesti

keeles.” In Pühendusteos Huno Rätsepale. Tartu Ülikooli eesti keele

õppetooli toimetised 7, ed. by M. Erelt, M. Sedrik, E. Uuspõld, 184-201.

Rätsep, Huno 1978. Eesti keele lihtlausete tüübid. Tallinn: Valgus.

Sahkai, Heete 2011. ”Eesti keele genitiivse agendifraasi süntaks.” – Keel ja

Kirjandus 1, 12–30.

Seržant, Ilja A., to appear. “Dative experiencer constructions as a Circum-

Baltic isogloss.ˮ In Tipologija slav’anskix, baltijskix i balkanskix jazykov

ed. by P. Arkadiev. Moscow.

Seržant, Ilja A. 2012. “The so-called possessive perfect in North Russian and

the Circum-Baltic area. A diachronic and areal account.” – Lingua 122,

356-385.

Seržant, Ilja A., this volume. “Categorization and semantics of subject-like

obliques“.

Silverstein, Michael 1976. “Hierarchy of Features and Ergativity”. In

Grammatical Categories in Australian Languages, ed. by R.M.W. Dixon,

112-171. Canberra: Australian Institute of Aboriginal Studies.

Tael, Kaja 1988. Sõnajärjemallid eesti keeles (võrrelduna soome keelega).

Tallinn: Eesti NSV Teaduste Akadeemia Keele ja Kirjanduse Instituut.

Preprint KKI-56.

36

Timberlake, A. 1976. Subject properties in the North Russian passive. In

Subject and Topic ed. by Ch.N. Li, 545–570. New York: Academic Press.

Torn, Reeli 2002. “The Status of the Passive in English and Estonian.ˮ RCEAL

Working Papers in English and Applied Linguistics 7, ed. by H. Hendriks,

81–106. Cambridge: Research Centre for English and Applied Linguistics.

Torn, Reeli 2006. “poolt-tarind eesti keele impersonaalis ja passiivis.ˮ In Lause

argumentstruktuur, ed. by P. Penjam, 108-121. Tartu Ülikooli eesti keele

õppetooli preprindid 2. Tartu.

Torn-Leesik, Reeli 2009. “The voice system of Estonian.” - Language

Typology and Universals (Sprachtypologie und Universalienforschung)

62, 1/2, 72–90.

Torn-Leesik, Reeli & Vihman, Virve-Anneli 2010. “The uses of impersonals in

spoken Estonian. “ - SKY Journal of Linguistics, 23, 301-343.

Uuspõld, Ellen 1970. “On the agent adverbial in the tud-construction.” In

Annual meeting of the research group for generative grammar.

Abstracts.Tartu State University. Department of Estonian, 38–43.

Vainik, Ene 1995. Eesti keele väliskohakäänete semantika kognitiivse

grammatika vaatenurgast. Eesti Teaduste Akadeemia Eesti Keele

Instituut. (Ars grammatica.) Tallinn.

Vihman, Virve-Anneli 2007. “Impersonaliseeritud impersonaal kui

konstruktsioonitasandi grammatisatsioon.” Emakeele Seltsi aastaraamat

52: 158–178.

Vihman, Virve-Anneli 2008. “Construction-based interpretation of implicit

arguments.” In CLS 40: The Panels (2004): Annual Meeting of the

Chicago Linguistic Society ed. by N. Adams, A. Cooper, F. Parrill, T.

Wier, 225-239. Chicago: Chicago University Press.

Viitso, Tiit-Rein 2003. “Rise and development of the Estonian language.” In

Estonian Language ed. by M. Erelt, 130–230. Linguistica Uralica

Supplementary Series, Vol. 1. Tallinn: Estonian Academy Publishers.

Wiedemann, Ferdinand Johann 1875. Grammatik der ehstnischen Sprache,

zunächst wie sie in Mittelehstland gesprochen wird, mit Berücksichtigung

der anderen Dialekte. St. Pétersbourg.