thinking out of the transitivity box -...

74
Universiteit Gent Faculteit Letteren & Wijsbegeerte Academiejaar 2012-2013 Thinking out of the transitivity box An exploration of the causative-inchoative verb alternation in English and German Masterproef voorgelegd tot het bekomen van de graad van Master in de Taal- en Letterkunde: Engels-Duits Joren Somers Promotor: Prof. Dr. Miriam Taverniers

Upload: truongkiet

Post on 31-Mar-2018

224 views

Category:

Documents


0 download

TRANSCRIPT

Universiteit Gent

Faculteit Letteren & Wijsbegeerte

Academiejaar 2012-2013

Thinking out of the transitivity box

An exploration of the causative-inchoative verb alternation

in English and German

Masterproef voorgelegd tot het bekomen van de graad van

Master in de Taal- en Letterkunde: Engels-Duits

Joren Somers

Promotor: Prof. Dr. Miriam Taverniers

In loving memory of Annick Craeye

Universiteit Gent

Faculteit Letteren & Wijsbegeerte

Academiejaar 2012-2013

Thinking out of the transitivity box

An exploration of the causative-inchoative verb alternation

in English and German

Masterproef voorgelegd tot het bekomen van de graad van

Master in de Taal- en Letterkunde: Engels-Duits

Joren Somers

Promotor: Prof. Dr. Miriam Taverniers

Preface

The development of this thesis has in many ways been a frustrating and disappointing process, so it

should not come as a surprise that I am feeling tremendously relieved at finally being able to close

this chapter off. Before I proceed to the order of the day, it seems only natural that I take a brief

moment to thank some people who have greatly helped me in shaping this piece of work. First and

foremost, I would like to genuinely thank my parents. The gratitude I feel towards them can hardly

be put to words, not to mention the help and support they have given me over the past few months.

Secondly, my supervisor, professor Miriam Taverniers, for her interesting introduction to the

phenomenon of ergativity and the helpful feedback. Thirdly, I would also like to thank Drs. Stavros

Kelepouris. Despite the fact that I was quite lenient in adhering to my self-imposed deadlines, the

days and hours before our weekly meetings always seemed to cause some very productive writing.

Thanks as well to my brother Hans, who turned my amateur lay-out into something far more

professional, and my sister Sara, for providing me with sugar. A massive thank you goes to my dear

friend and partner in ergative crime, Evi. I can only say that if it was not for her, this thesis would

simply not have been there (and that is by no means an overstatement). A very special word of

thanks goes to professor Benjamin Biebuyck, probably one of the most dedicated, genuine and

brilliant people I have ever had the pleasure of meeting. Not only did he teach me the importance of

critical thinking and solid argumentation, the extremely self-evident way in which he was willing to

listen and spontaneously offer his help when no one else seemed to understand the problem, is

something I have never experienced before. Much appreciated was also professor Luc de Grauwe’s

syntactic elucidation on a stubborn particle verb and Wim’s inside information.

My final word of thanks goes to Charlotte, Meltem and Evi (once again), for making the past four

years an unforgettable experience. The prospect of graduating together with them has probably

been the main incentive to bring this thesis to a successful conclusion.

Joren Somers,

Friday, 9 August 2013.

5

Table of contents

PREFACE ...................................................................................................................................... 4

TABLE OF CONTENTS .................................................................................................................... 5

INTRODUCTION ............................................................................................................................ 7

1 SECTION I – LITERATURE REVIEW ........................................................................................... 10

1.1 WAYS OF EXPRESSING THE CAUSATIVE-INCHOATIVE VERB ALTERNATION .............................................. 10

1.1.1 MORPHOLOGICALLY-BASED ALTERNATIONS ........................................................................................ 10

1.1.1.1 Relic forms .............................................................................................................................. 10

1.1.1.2 Suppletion .............................................................................................................................. 11

1.1.1.3 Affixation ................................................................................................................................ 12

1.1.1.4 Lability .................................................................................................................................... 14

1.1.2 SYNTACTICALLY-BASED ALTERNATIONS: CAUSATIVE/GRAMMATICAL PERIPHRASES ..................................... 19

1.2 ERGATIVITY AND TRANSITIVITY COMPARED: DAVIDSE’S JANUS-HEADED GRAMMAR OF ACTIONS AND EVENTS 23

1.3 TESTS FOR ERGATIVITY ............................................................................................................. 27

1.4 CONSTRAINTS ON ERGATIVISATION ............................................................................................. 31

1.5 UNACCUSATIVITY AND UNACCUSATIVE DIAGNOSTICS ....................................................................... 35

1.5.1 UNACCUSATIVITY ........................................................................................................................... 35

1.5.2 UNACCUSATIVE DIAGNOSTICS........................................................................................................... 36

1.5.2.1 Possibility to appear in resultative constructions .................................................................. 36

1.5.2.2 Prenominal perfect/passive participles ................................................................................. 37

1.5.2.3 German-exclusive diagnostics ................................................................................................ 37

1.5.2.3.1 Auxiliary selection ............................................................................................................... 37

1.5.2.3.2 Impersonal passives ............................................................................................................ 38

1.5.2.3.3 Split phrases ........................................................................................................................ 38

1.5.2.3.4 Nominal attributability and agent nominals ....................................................................... 39

1.6 A NOTE ON THE MIDDLE VOICE ................................................................................................... 40

2 SECTION II – ANALYSIS .......................................................................................................... 43

2.1 AIM AND METHODOLOGY ........................................................................................................ 43

2.2 DISCUSSION AND RESULTS ........................................................................................................ 46

2.2.1 ERGATIVE VERBS ............................................................................................................................ 46

2.2.2 REFLEXIVE VERBS ........................................................................................................................... 47

2.2.3 VERBS WITH PREVERBAL ELEMENTS ................................................................................................... 50

2.2.3.1 Prefix verbs (‘Präfixverben’) ................................................................................................... 51

2.2.3.1.1 ver- ...................................................................................................................................... 51

2.2.3.1.2 be- ........................................................................................................................................ 53

2.2.3.1.3 er- ........................................................................................................................................ 55

2.2.3.2 Particle verbs (‘Partikelverben’) ............................................................................................. 55

2.2.3.2.1 ab- ....................................................................................................................................... 56

2.2.3.2.2 an- ....................................................................................................................................... 57

2.2.3.2.3 auf- ...................................................................................................................................... 58

6

2.2.4 PASSIVES ...................................................................................................................................... 58

2.2.5 FUNCTION VERB PHRASES (‘FUNKTIONSVERBGEFÜGE’) ........................................................................ 61

2.2.6 VERBS WITH A PREPOSITIONAL COMPLEMENT ..................................................................................... 65

2.2.7 SYNTACTIC CAUSATIVES ................................................................................................................... 66

2.2.8 REMAINING VERBS ......................................................................................................................... 66

CONCLUSION .............................................................................................................................. 68

REFERENCES ............................................................................................................................... 70

7

Introduction

Although the term ‘ergativity’ has been around since at least the late 1920s (Dixon 1994: 3), the

study of this phenomenon only truly took off half a century later with the publication of Dixon’s

appropriately entitled paper Ergativity. In this seventy-page long essay, he tries to explore “the

various properties that have been taken as examples of ‘ergativity’” and attempts “to arrange and

explain them” (Dixon 1979: 60). Before moving on, it is important to realise that regardless of the

way in which a specific author defines the concept of ergativity, it always boils down to a contrast of

the following sort:

The term ergativity or ergative patterning refers to the situation in which, in a given language, the

agent or ‘subject’ of a transitive clause [...] shows patterning distinct from the actor or ‘subject’ of an

intransitive clause [...], which patterns like the patient or undergoer (‘object’) of a transitive clause [...].

(McGregor 2009: 480)

The lion’s share of Dixon’s aforementioned paper is dedicated to so-called ‘morphological ergativity’,

which means that the ergative properties of a given language are expressed through specific

grammatical forms, e.g. by using affixes or case inflections. This results in a system in which –

contrary to a nominative-accusative-based system – the subject of the transitive clause (= A) appears

in the marked ergative case and both the subject of the intransitive clause (= S) and the object of the

transitive clause (= O) occur in the unmarked absolutive1:

Figure 1: The marking of S, A and O in accusative and ergative case systems (source: Dixon 1994: 9)

Next to using a morphological marking system, Dixon argues that languages can also display ‘ergative

syntax’, meaning that “some rules of coordination and/or subordination will treat O and S in the

1 It has been estimated by Siewierska 2008 (as cited in McGregor 2009) that ergative case systems are

operative in approximately 20% of the world’s languages. Among these are e.g. Basque and Dyirbal (McGregor 2009: 483).

8

same way, and A rather differently”2 (Dixon 1979: 62). However, he explicitly rejects the use of the

term ‘ergative’ with respect to phenomena on the lexical-semantic level of a given language, since he

argues that

if the term ‘ergative’ is used in this way [i.e. the lexical-semantic way], then every language would

show ‘ergativity’ (surely every language has some sort of causative construction). In contrast, only

about a quarter of the world’s languages (not including German and English) show morphological or

intra-clausal ergativity. (Dixon 1994: 20)

In order to fully grasp what is precisely meant by this lexical-semantic interpretation, let us consider

the following sentence pair:

(1) a. Jill breaks the vase.

b. The vase breaks.

If we take a closer look at these alternating constructions, it strikes us that the direct object of (1)a,

notably the vase, ‘moves’ to subject position in (1)b. In other words, the object of the transitive

clause ‘becomes’ the subject of the intransitive clause. Thus, the alternation exemplified above

instantiates the exact same principle at work in ergative languages, albeit on a different level, for in

this case, the ergativity solely depends on the lexical-semantic properties of the verb. This is the

reason why this phenomenon is often termed LEXICAL-SEMANTIC ERGATIVITY and the verbs allowing for

such an alternation ERGATIVE or LABILE VERBS (Davidse 1992; Dixon 1994; Garcia Garcia 2012;

Haspelmath 1993; McGregor 2009; Van Gelderen 2011). It has been argued that, semantically,

ergative verbs allow us “to talk about the world in different ways” (Francis et al. 1996). In the one-

participant construction (as in The vase breaks), a speaker can present an event as a “natural

occurrence”, an action which seems to happen of its own accord. Many terms have been proposed to

describe this type of construction, but in line with Haspelmath 1987, I will adhere to the term

INCHOATIVE ALTERNANT. The two-participant construction, on the other hand, always identifies the

cause responsible for initiating the action, so it will henceforth be referred to as the CAUSATIVE

ALTERNANT. Thus, an ergative verb is a verb which can appear in a CAUSATIVE-INCHOATIVE VERB

ALTERNATION. It is important to realise that causative-inchoative alternations are by no means limited

to ergative verbs alone, but the reason why ergatives are so special in this respect is because they

can express this alternation without changing their external form.

2 Recall that O = transitive object, S = intransitive subject and A = transitive subject.

9

The aim of the current paper is twofold. First of all, I will try to provide a succinct overview of

ergativity research so far. As has already been pointed out above, ergativity refers to a phenomenon

where a causative-inchoative alternation is expressed by using one and the same verb, but such an

alternation can be expressed in other ways as well: ergative verbs are, in other words, part of a

‘bigger picture’. Therefore, I will devote the first chapter to a discussion of ergativity on a par with

other causative-inchoative alternations. The structure of this overview will mainly be based on Baron

(1974), but it is evident that evidence was taken from an array of other sources as well. Next to that,

the literature review will also pay attention to the relation between ergativity and transitivity

(Chapter 2), tests for determining the ergative nature of a specific verb (Chapter 3), constraints on

ergativisation (Chapter 4) and the way ergativity is linked to what in formal traditions has been

termed ‘Unaccusativity’ (Chapter 5). I will conclude the first section with a note on the middle voice,

a construction which is often mentioned in the same breath as ergativity (and sometimes even

equated with it), but which needs to be clearly distinguished from it. The second part of this paper

will be devoted to a contrastive study of ergativity. More specifically, I will try to translate a semantic

set of English ergative verbs into German and take a closer look at the strategies the target language

uses to express the causative-inchoative alternation inherently present in the original set.

10

1 Section I – Literature review

1.1 Ways of expressing the causative-inchoative verb alternation

1.1.1 Morphologically-based alternations

1.1.1.1 Relic forms

The first subtype of morphologically-based causative-inchoative verb alternations encompasses a

handful of fossilised verb pairs in which the causative alternant was once derived from its inchoative

base form through (palatal) vowel mutation or umlaut (Garcia Garcia 2012: 126; Van Gelderen 2011:

115). Lass argues that this phonological process, which was caused by /i/ or /j/ in the following

syllable, essentially boiled down to the “’attraction’ of vowels towards the upper front corner of the

vowel space” (Lass 1994: 60). In this case, the trigger was the once productive but now obsolete

Germanic suffix *-(i)ja-, which “was used profusely to derive causatives from a non-causative base”

(Garcia Garcia 2012: 126). As can still be observed in the modern Germanic languages, the inchoative

base forms were always strong, the derived causatives weak. The total amount of original Germanic

causatives is controversial, but throughout history, their number is sure to have dropped

substantially. According to Baron 1974, Modern English only has four witnesses to this process left,

but Garcia Garcia makes mention of three more3:

Inchoative base-form Derived causative verb

bite bait

drink drench

fall fell

lie lay

sit set

rise rear

sing singe

Table 1: English causatives derived by i/j-mutation along with their inchoative base forms (sources:

Baron 1974: 303; Garcia Garcia 2012: 131).

3 The three additional entries only mentioned by Garcia Garcia 2012 (131) are marked in bold.

11

As can be deduced from the table below, the number of derived causatives in Modern German is

considerably higher:

Inchoative base-form Derived causative verb

ersaufen ersäufen

fallen fällen

fließen flößen

liegen legen

sinken senken

saugen säugen

schwingen schwenken

sitzen setzen

springen sprengen

trinken tränken

wachen wecken

verschwinden verschwenden

singen sengen

Table 2: German causatives derived by i/j-mutation and their inchoative base forms (source: Ide

1996: 50)

Haspelmath argues that for instances in which base and derivative differ minimally (e.g. with fall-fell

or sinken-senken), the vowel alternation is iconically motivated. Syntactically speaking, vowel

mutation is an instance of a valence-changing operation, as it allows a certain verb to change the

number of participants it takes. Haspelmath observes that crosslinguistically, “valence-changing

morphemes tend to occur very close to the stem”, since they are perceived as highly relevant

categories (Haspelmath 1987: 17). In fact, by causing a change to the stem itself, vowel mutation

brings this principle to the extreme (Haspelmath 1987: 17).

1.1.1.2 Suppletion

A second subtype within this category is constituted by so-called “suppletive” verb pairs, i.e. verbs

bearing no morphological resemblance to one another, due to different underlying root morphemes

(Baron 1974: 303; Comrie 2006: 304; Garcia Garcia 2012: 126; Haspelmath 1993: 92). Baron 1974

cites the English examples believe-persuade, learn-teach, eat-feed, die-kill and many more, in which

the latter always represents the causative member of the alternation, the former the inchoative. For

12

German, such a list does not seem to have yet been drawn up, but at least with respect to the verb

pairs cited above, it seems to behave in a way similar to English: glauben-überzeugen, lernen-lehren,

essen-nähren and sterben-töten.

1.1.1.3 Affixation

Thirdly, morphological causatives can also be formed through affixation. Baron argues that there are

“at least four suffixes and two prefixes for deriving causative verbs from adjectives and nouns [...]”,

notably -ify, -ize, -en, -ø, en- and dis-(Baron 1974: 306). The following table provides a brief overview

of these affixes , as well as some examples to illustrate the derivational processes:

process affix example

adjective > verb -ify just > justify

-ize legal > legalize

-en ripe > ripen

-ø empty > empty

en- rich > enrich

dis- able > disable

noun > verb -ify glory > glorify

-ize American > Americanize

-en strength > strengthen

-ø wine > wine and dine

en- code > encode

dis- courage > discourage

Table 3: Causativising affixes in English (source: Baron 1974: 306-307).

However, according to Levin & Rappaport Hovav 1994, the suffixes -ify, -ize and -en are not just

typical of causative verbs, but also of inchoative verbs, thus giving rise to ergatives. The following

sentences clearly exemplify the labile nature of the aforementioned suffixes:

(2) a. I solidified the mixture. (Levin & Rappaport Hovav 1994: 44, ex. 34a)

b. The mixture solidified. (Ibid., ex. 34b)

(3) a. The cook caramelized the sugar. (Ibid., ex. 35a)

b. The sugar caramelized. (Ibid., ex. 35b)

13

(4) a. John thickened the sauce. (Ibid., ex. 36c)

b. The sauce thickened. (Ibid., ex. 36c)

Levin & Rappaport Hovav argue that the original function of the suffixes -ify, -ize and -en was indeed

that of a causativiser, but that they gradually started to develop an inchoative meaning component

as well. Based on figures provided by Fontenelle & Vanandroye, Levin & Rappaport conclude that,

compared to the other two post-verbal elements, -en has shown a distinctly higher inclination

towards ergativisation, since “only 14 out of 82 -ify verbs [from the Longman Dictionary of

Contemporary English] participated in [a causative-inchoative alternation], contrasting with 46 out of

84 -en verbs” (Levin & Rappaport Hovav 1994: 44). Data for -ize verbs were not included in

Fontenelle & Vanandroye’s aforementioned study, but Levin & Rappaport Hovav bridged this hiatus

by calculating that the number of ergative verbs in the Oxford Advanced Learner’s Dictionary showing

this suffix only amounted to 14 out of 78. However, these figures need to be nuanced somewhat.

Fontenelle & Vanandroye’s and Levin & Rappaport Hovav’s calculations were based on dictionaries

compiled in 1978 and 1974 respectively. It is possible that the ergativisation of verbs in -ize was still

in a premature stage back then, potentially resulting in these relatively low numbers. Keyser and

Roeper seem to confirm this tentative explanation by arguing that the productivity of this suffix is

clearly demonstrated by the fact that “new forms constantly arise”, especially in bureaucratic and

scientific language (Keyser & Roeper 1984: 389-390)4. They provide the following list of ergative -ize

verbs:

alkalinize alkalize Americanize anatomize

automatize capitalize centralize channelize

demagnetize demilitarize demobilize equalize

federalize generalize harmonize hybridize

liberalize localize magnetize materialize

mechanize militarize mobilize neutralize

normalize organize oxidize polarize

pressurize regularize reorganize revitalize

stabilize standardize synchronize urbanize

Table 4: Non-exhaustive list of ergative -ize verbs typical of bureaucratic and scientific language use

(source: Keyser & Roeper 1984: 390)

4 This could also be the reason why Baron only recognised these affixed verbs as causatives, rather than as

ergatives.

14

Keyser & Roeper also cite a number of verbs in -ize which do not allow for ergative formation, which

leads them to conclude that ergativisation apparently not only hinges on the properties of the suffix,

but also on those of the noun with which they can appear in a collocation. These include (among

others) capitalize, characterize, sympathize and visualize (Keyser & Roeper 1984: 390).

A diachronic account of English causativisers is offered by Van Gelderen, who argues that -en, -ize, -

(i)fy and -ate all emerged in the Middle English period. However, these suffixes need to be traced

back to different sources, since -ize, -(i)fy and -ate “came to English mostly through Latin and French

loans that were themselves based on Greek and Latin”, whereas-en clearly is Germanic in origin (Van

Gelderen 2011: 126). Another difference between these elements is that Van Gelderen only

considers the en-suffix as productive, but not the other three. This observation is in line with Levin

and Rappaport Hovav 1994, but (at least partially) in contradiction with Keyser and Roeper 1984,

who consider the -ize-causativiser a productive affix.

1.1.1.4 Lability

As was already pointed out in the introduction, ergative or labile verbs are grammatical forms which

can be used both in a causative as well as in an inchoative constructions without changing their

external form. According to Francis et al., an ergative verb can be ascribed three specific features:

“[1] it has two patterns, [2] only one of these patterns has a noun group following the verb and [3]

the person or thing indicated by that noun group may also be indicated by the Subject of the other

pattern” (Francis et al. 1996). Moreover, ergative verbs come in two types of clausal patterns. Most

often, the alternation is “symmetrical” in nature, which means that “the only difference between the

two patterns is that one has a noun group following the verb and the other does not”. Symmetrical

combinations come in six different patterns:

Pattern number Structure number Inchoative alternant Causative alternant

Pattern 1

Verb; Verb + NP

Structure (i) Verb

e.g. The door closes

Verb + NP

e.g. Jill closes the door

Pattern 2

Verb + PP/AdvP; Verb +

NP + PP/AdvP

Structure (i) Verb + Prepositional

Complement

e.g. The prince changed

into a frog

Verb + Object +

Prepositional Object

Complement

e.g. The magician

changed the prince

into a frog

Structure (ii) Verb + Prepositional Verb + Object +

15

Object

e.g. Beauty equates

with goodness

Prepositional Object

e.g. He equated

beauty with goodness

Structure (iii) Verb + Prepositional

Object

e.g. She converted to

Christianity

Verb + Object +

Adjunct

e.g. He converted

them to Christianity

Structure (iv) Verb + Adjunct

e.g. The coach halted in

front of the ballroom

Verb + Object +

Adjunct

e.g. The footman

halted the coach in

front of the ballroom

Pattern 3

Verb + AdjP; Verb + NP +

AdjP

Structure (i) Verb + Complement

e.g. The door slammed

shut

Verb + Object +

Object Complement

e.g. She slammed the

door shut

Pattern 4

Verb + PP*; Verb + NP +

PP*

(*PP = as + AdjP)

Structure (i) Verb + Prepositional

Complement

e.g. He qualified as

unemployed

Verb + Object +

Prepositional Object

Complement

e.g. This qualified him

as unemployed

Pattern 5

Verb + to-infinitive; Verb

+ NP + to-infinitive

Structure (i) Verbs in phase

e.g. I incline to think he

is wrong

Verbs with 2 Objects

e.g. This inclined me

to think he was wrong

Structure (ii) Verb + Adjunct

e.g. She qualified to

teach children

Verb with 2 Object

e.g. Her course

qualified her to teach

children

Pattern 6

Verb + Ordinal Number +

PP; Verb + NP + Ordinal

Number + PP

Structure (i) Verb + Adjunct +

Adjunct

e.g. Michael Adams

already ranks 20th in

the world

Verb + Object +

Adjunct + Adjunct

e.g. The junior team is

ranked third in the

world

16

Table 5: Symmetrical pattern combinations for ergative verbs (source: Francis et al. 1996)

Asymmetrical pattern combinations seem to be less numerous, displaying only four different

possibilities:

Pattern number Structure number Inchoative alternant Causative alternant

Pattern 1

Verb + PP/AdvP; Verb +

NP; Verb + NP + PP/AdvP

Structure (i) Verb + Adjunct

e.g. The boat rocked up

and down

Verb + Object (+

Adjunct)

e.g. Huge waves rocked

the boat (up and down)

Pattern 2

Verb + PP/AdvP; Verb +

NP

Structure (i) Verb + Prepositional

Complement

e.g. The glass splintered

into pieces

Verb + Object

e.g. The blow splintered

the glass

Structure (ii) Verb + Prepositional

Object

e.g. His heart hardened

against her

Verb + Object

e.g. The years hardened

my heart

Structure (iii) Verb + Adjunct

e.g. Water gushed out

of the hole

Verb + Object

e.g. The hole gushed

water

Pattern 3

Verb + AdvP; Verb + NP

Structure (i) Verb + Adjunct

e.g. The dress washed

easily

Verb + Object

e.g. She washed the

dress

Pattern 4

Verb + AdjP; Verb + NP

Structure (i) Verb + Adjective

e.g. These racks fold flat

for easy storage

Verb + Object

e.g. Brian folded the

lawn chair

Table 6: Asymmetrical pattern combinations for ergative verbs (source: Francis et al. 1996)

With respect to the number of labile verbs in English, it has been repeatedly argued (e.g. by Baron

1973: 305; Garcia Garcia 2012: 138; Haspelmath 1987: 17; Van Gelderen 2011: 106) that Modern

English is somewhat of an outsider within the Germanic language group, since it contains a peculiarly

high amount of them. In a study carried out in 1993, Haspelmath translated a total of 31 causative-

17

inchoative verb pairs into 21 languages and subsequently classified the retrieved data according to

the way in which the causative-inchoative alternation was represented. In total, he distinguished

between five different translation strategies, namely ‘anticausative’, ‘causative’, ‘equipollent’, ‘labile’

and ‘suppletive’. Consult Table 7 for an overview of the studied verb pairs and Table 8 for the results

for English and German according to the respective strategies:

wake up break burn die/kill

open close begin learn/teach

gather spread sink change

melt be destroyed/destroy get lost/lose develop

connect boil rock go out/put out

rise/raise finish turn roll

freeze dissolve fill improve

dry split stop

Table 7: Sample of causative-inchoative verb pairs studied in Haspelmath 1993.

Anticausative Causative Equipollent Labile Suppletive

English 2 0 1 25 3

German 15,55 1 4 9,56 1

Table 8: English and German strategies to represent the causative-inchoative verb alternation

(source: Haspelmath 1993 as cited in Comrie 2006: 306)

As can be deduced from the figures in the table above, the number of labile verbs in English is indeed

remarkably high, covering up to 80% of the total outcome. In German, this verb type proves to be

substantially rarer, since only 30% of the German translations satisfied the criteria for this class. Van

Gelderen argues that “English presumably started out with valency alternations similar to the ones in

present-day German, [but] changed drastically to lability” (Van Gelderen 2011: 115). The question

this hypothesis logically puts forward is of course: What caused English to deviate from the

‘Germanic standard’? In fact, the answer is only partly satisfying, since the number of labile verbs

was already “quite high” in Old English and it remains unclear what might have caused its deviation

from the other Germanic languages in the first place (Van Gelderen 2011: 137). What we do know is

5 The reason why the number of anticausatives and labile verbs in German is not a round figure, is because

English ‘rock’ could both be translated as ‘schaukeln (intr./tr.)’ as well as ‘sich schaukeln/schaukeln’ (Haspelmath 1993: 114). 6 Cf. supra.

18

that from the Old English period onwards, the English verbal system has shown an unabated

tendency towards what Van Gelderen terms ‘morphological opacity’. On the one hand, this resulted

in the obsolescence of transitivising and causativising affixes, which in turn led to verbs being

reinterpreted as labile, since base and derivative could not be distinguished anymore (Van Gelderen

2011: 134, 138). Garcia Garcia argues that this ‘syntactic melting’ affected at least 13 out of 57 Old

English causative-inchoative verb pairs (Garcia Garcia 2012: 137). On the other hand, a handful of

new causativisers (such as -en, -ise, -(i)fy and -ate), which were introduced during the Middle English

period, came to develop an inchoative meaning component as well (Van Gelderen 2011: 125, 126). It

has been argued that this ‘morphological opacity’ resulted from the fact that English is essentially a

transitivising language (contrary to German, which has been characterised as detransitivising), but

further research within this area is required to reach a higher degree of certainty with respect to this

hypothesis (Van Gelderen 2011: 111, 137).

19

1.1.2 Syntactically-based alternations: causative/grammatical periphrases

Apart from the morphologically based alternations discussed above, causatives can also be expressed

periphrastically through a verb + complement construction. According to Baron 1974, English has at

least five verbs at its disposal capable of functioning as such causative auxiliaries, but “not every

periphrastic may be used with every complement” (Baron 1974: 308). For example, in constructions

with a present participle as complement, only have and get allow for a grammatical periphrasis:

(5) a. The actress had/got her director eating out of her hand. (Baron 1974: 308)

b. *The actress made/caused/let the director eating out of her hand. (Baron 1974:

308)

Baron also observes that “all periphrastic constructions are potentially more ‘indirect’ than

morphological causatives”, by which she means that in constructions involving morphological

causatives, the instigation of the process and the resultant state of affairs always have to coincide in

time and space (Baron 1974: 333, 327; Fodor 1970 in Davidse 1992: 116). As indicated by the

following examples, periphrastic causatives are not subject to this restriction:

(6) a. *On Monday, Nixon activated the mines in Haiphong Harbor on Thursday.

(Baron 1974: 327, ex. 122)

b. On Monday, Nixon caused the mines in Haiphong Harbor to activate on Thursday.

(Baron 1974: 327, ex. 123)

(7) a. *In Washington, the State Department dropped three hundred leaflets in

Bhutan. (Baron 1974: 327, ex. 124)

b. In Washington, the State Department had three hundred leaflets dropped in

Bhutan. (Baron 1974: 327, ex. 125)

However, the degree of directness still varies among causative auxiliaries as well, making it possible

to arrange (at least some of them) in a hierarchical order. Baron argues that make displays a higher

degree of directness than get, which in turn overrules have in this respect. The hierarchy is thus as

follows: make > get > have (Baron 1974: 333). This is also exemplified by the following constructions:

(8) a. The trainer made the lion enter the cage. (Baron 1974: 333, ex. 154)

b. The trainer got the lion to enter the cage. (Baron 1974: 333, ex. 155)

20

c. The trainer had the lion enter the cage. (Baron 1974: 333, ex. 156)

According to Baron, “[6a.] normally implies the direct force of an Agentive trainer, [6b.] potentially

involves the trainer using devious means [e.g. a whip], and [6c.] suggests that the trainer was never

in a position to act directly” (Baron 1974: 333). A similar idea is developed in Davidse 1992, who,

building on Fodor 1970, also argues against the synonymy of lexical and grammatical causatives.

Consider the examples below:

(9) a. John caused Bill to die on Sunday by stabbing him on Saturday (Davidse 1992:

116, ex. 1).

b. *John killed Bill on Sunday by stabbing him on Saturday. (Davidse 1992: 116, ex. 2).

Just like examples (6)a and (7)a, (9)b is ungrammatical because lexical causatives cannot separate the

instigation of given event from the resultant state of affairs, neither temporally, nor locally. However,

Davidse 1992 argues that the class of lexical causatives also contains causative ergatives, and, for this

subgroup, a characterisation in terms of directness is completely pointless. She even describes it as

“one of the most stubborn fictions in descriptive linguistics” (Davidse 1992: 119). Observe the

examples below:

(10) The boss works his secretaries from eight till six7. (Davidse 1992: 119, in-text example)

(11) [through the intercom of an apartment building] Could you open the door, please?

(Davidse 1992: 119, in-text example)

Both of these constructions hinge on a lexical causative of the ergative type, so according to the

claim put forward in Baron 1974 – which does not distinguish between transitive and ergative

causatives – they should display a high degree of directness. However, example (10) does not in any

way imply that the boss “physically manipulates his secretaries” (Davidse 1992: 119). Construction

(11) also runs counter to the traditional view, as it applies to a context of use in which the degree of

directness is rather low, more precisely to a situation in which a person on the ground floor of an

apartment complex is asking someone living several floors higher to open the door downstairs by

pushing the appropriate button. Indeed, the lexical causative construction is perfectly acceptable,

whereas the use of the causative periphrasis Could you cause the door to open? would indeed be a

7 Keen observers might have noticed that this example is founded on a causatived intransitive verb (rather than

on an ergative verb), a pattern which Davidse & Heyvaert 1997 term the “ergative causativization of an intransitive” or ECI. This phenomenon will be discussed in greater detail in Section 1.3.

21

rather odd choice in this case (Davidse 1992: 119). But if the difference between ergative and

analytical causatives does not reside in the feature of directness, then clearly we need to look for

another explanation. Davidse argues that the main difference between these two construction types

is to be found in the behaviour of the Medium-Process complex:

The principle is as follows: ergative structures ALTER the inherent voice of the Medium-Process

complex; analytical causation does NOT alter the inherent voice of the process-participant(s) complex

to which it adds a feature of ‘causation’. (Davidse 1992: 119)

Let us take a look at the examples below:

(12) a. The bal rolls. (Davidse 1992: 120, ex. 1a.)

b. Peter made the ball roll. (Davidse 1992: 120, ex. 1b.)

c. Peter rolled the ball. (Davidse 1992: 120, ex. 1c.)

Davidse argues that construction (12)a is characterised by a fundamental vagueness, as it remains

unclear whether the balls rolls due a natural inclination, or whether the rolling is brought about by an

external force (Davidse 1992: 121). In accordance with the principle defined above, this (ambiguous)

voice character remains unchanged in (12)b, but (12)c allows for one interpretation only, since Peter

is in that case the uncontested instigator of the process (Davidse 1992: 121).

For German, Gunkel proposes that causative periphrases can be expressed by means of two different

constructions, namely by using a plain causative or with the help of a so-called “causative passive”

(Gunkel 1998: 65). The following sentences, based on two examples from the study in question,

clearly exemplify the suggested dichotomy:

(13) a. Ich lasse den Maler die Wand neu streichen. (based on Gunkel 1999: 65, ex. (2))

b. Ich lasse die Wand (vom Maler) neu streichen. (based on Gunkel 1999: 65, ex. (2))

Whereas (13)a illustrates the plain causative construction, (13)b is an instance of a causative passive.

A typical feature of the former is that the subject of the infinitival construction needs to be realised

as an accusative NP (here: ‘den Maler’), a syntactic phenomenon also known as accusativus cum

infinitivo. This does not hold for the causative passive, which, just like ordinary passives, expresses

the agent by means of an omissible prepositional phrase. A second feature both ordinary and

causative passives have in common is the fact that they can only be derived from agentive verbs,

22

whereas an ordinary causative can take any type of verb as its input8 (Gunkel 1998: 65). In fact,

Gunkel’s plea for causative passives is reminiscent of Haspelmath’s claim that “a causative can

become a passive, but there is no evidence for a case of a passive becoming a causative”

(Haspelmath 1990: 49 as cited in Ogawa 2002: 89). Consequently, as Ogawa argues, if a given

language expresses both causatives as well as passives in morphologically identical ways, “such

constructions are to be interpreted either as causative, or as both causative as well as passive”

(Ogawa 2002: 89). A further restriction lassen-periphrases are subject to, is the fact that “the causee

nominal [...] cannot be made the subject of a passive clause” (Shibatani 2002: 155). This contrary to

causative constructions hinging on the auxiliary zwingen ‘force’, which, because of the fact that it is

“less grammaticalised” than lassen, does allow for an alternation of the aforementioned type

(Shibatani 2002: 155). The following examples clearly illustrate this contrast:

(14) a. Man ließ den Studenten abreisen. (Shibatani 2002: 155, ex. 28a.)

b. *Der Student wurde abreisen gelassen. (Ibid., ex. 28b.)

(15) a. Man zwang den Studenten abzureisen. (Ibid., ex. 29a.)

b. Der Student wurde gezwungen abzureisen. (Ibid., ex. 29b.)

8 Gunkel notes that “certain psych and experiencer verbs” also allow for a causative passive, but she does not

further discuss this claim (Gunkel 1998: 65).

23

1.2 Ergativity and transitivity compared: Davidse’s Janus-headed grammar of actions and events

According to Davidse 1992, the transitive paradigm boils down to a PROCESS AND EXTENSION model.

Here, the minimum requirement is that a certain Actor initiates a specific Process, which will result in

an intransitive clause with an Actor-Process structure (e.g. The water is flowing). This basic pattern

may subsequently be extended to including a Goal, thus giving rise to an Actor-Process-Goal

structure, which is realised by means of a transitive clause. Davidse argues that, in English, the

process and extension model for semantically related constructions is typically realised by means of

different verb roots (Davidse 1992: 108):

Actor-Process Actor-Process-Goal

The water is flowing. The water is flooding the river banks.

The baby is playing. The baby is manipulating her toys.

Table 9: The Process and Extension model exemplified (source: Davidse 1992: 108)

The ergative system, on the other hand, hinges on an INSTIGATION OF PROCESS model, whose semantic

core corresponds to a Medium-Process structure (e.g. The vase breaks). Davidse claims that this

construction type, which we have termed ‘inchoative’, is essentially vague in nature, “[since] it leaves

open whether or not a process is self-instigated or instigated by an external agent” (Davidse 1992:

109). One could thus say that a one-participant ergative construction ‘floats’ somewhere in between

these two options. This is clearly a more nuanced version of the widely-accepted view that inchoative

constructions (solely) present actions as coming about spontaneously (e.g. Francis et al. 1996;

Haspelmath 1987: 15; Haspelmath & Müller-Bardey 1991: 5; Haspelmath 1993: 93). Just like in the

transitive paradigm, the ergative system also allows for an additional ‘participant shell’ to its nucleus,

yet one of a totally different kind than the transitive Goal. Here, the one-participant construction can

be complemented with an Instigator, i.e. an entity which explicitly adds to the proposition the

feature of external causation, resulting in an Instigator-Process-Medium structure (Davidse 1992:

109). An example of this type of construction, which we have termed the causative alternant, is Jill

breaks the vase. These observations lead Davidse to conclude that transitive and ergative

constellations essentially have “different grammatical centres and different ‘directionalities’”

(Davidse 1992: 110):

The transitive system is Actor-centred: its most central participant is the Actor, and the Actor-Process

complex is grammatically more nuclear and relatively more independent. The basic Actor-Process

frame can be extended only to the RIGHT to include a goal. The ergative system, in contrast, is Medium-

24

centred, with the Medium as most nuclear participant and with a more nuclear, independent Medium-

Process complex. Here, the basic Medium-Process constellation is opened up to the left to incorporate

the Instigator. (Davidse 1992: 110)

However, the most important aspect of Davidse’s study is probably the fact that she abandons the

traditional dichotomy between causative and inchoative constructions and proposes a new, tripartite

approach to ergativity. In her opinion, the ergative paradigm can be broken down into (1) middles,

(2) effectives and (3) pseudo-effectives. The existence of these three subtypes is also postulated for

the transitive system. Perhaps the most straightforward of Davidse’s categories is that of the ergative

middle9, since it shows a complete overlap with what we have been terming inchoative

constructions, i.e. one-participant constructions of the type The vase breaks or The windows crack. In

the transitive paradigm, middles are realised by means of intransitive clauses, like Jack is walking.

What is common to both of these construction types then, is the fact that they represent their

respective paradigms in the most nuclear way. For the ergative paradigm, this means that the action

is not instigated by an external source, for the transitive paradigm, this results in an action not

projected onto a Goal (Davidse 1992: 123). The second subgroup, notably that of effectives, can be

broken down (at least for ergative constructions) into instigation-of-process constructions on the one

hand, and instigation-of-action constructions on the other. Davidse argues that an instigation-of-

process structure is the prototypical instance of ergativity, since it is the natural, two-participant

counterpart corresponding to the ergative middle (which is a one-participant construction). An

instance of such an alternation can be found below:

(16) a. Ergative effective (Instigation-of-process): Jill opens the door.

b. Ergative middle: The door opens.

Instigations of actions, however, should be seen as situated “on the border between the transitive

and the ergative paradigms” (Davidse 1992: 115). They always hinge on an intransitive verb (which is

the reason why they have been termed ‘ergative causativisations of intransitives’ by Davidse &

Gheyskens 1998), are typically characterised by human causers and require “active co-participation

of the causee” (Davidse & Gheyskens 1998: 10). An example of the construction type in question can

be found below:

(17) The general marched the soldiers. (Davidse & Gheyskens 1998: 10, in-text ex.)

9 Davidse’s use of the term ‘middle’ should not be mixed up with what has traditionally been termed a middle,

i.e. a construction of the type That book reads well or These vases break easily.

25

For a more detailed account of their properties the reader is referred to Section 1.4 of this paper.

The transitive:effective paradigm, Davidse argues, can best be described in terms of ‘intentionality’.

The primary distinction that needs to be made in this respect, is one between animate and inanimate

actors. Inanimate actors are always characterised by an ‘absence of intentionality’, but for animate

actors there are again two options, since their actions can be either fully intentional or accidental

(meaning they are characterised by a ‘lack of control’). As a result, the transitive:effective paradigm

can be represented in three different ways:

(18) Intentionality The teacher hit the child. (Davidse 1992: 111, ex. 1)

(19) Lack of control John accidentally hit Mary in the face. (Ibid., ex. 2)

(20) Absence of intentionality The arrow hit the target. (Ibid., ex. 3)

The last ergative construction type Davidse distinguishes, is the so-called ‘pseudo-effective’. Here,

the participant which behaves in an essentially different way than in effective structures, is the

Subject, since it “is reduced in participant status to the point of no longer being an Agent” (Davidse

1992: 127). Consider the following examples:

(21) He fractured an arm in the accident. (Davidse 1992: 127, ex. 1)

(22) The house blew a fuse. (Ibid., ex. 4)

The reason why these constructions cannot be regarded as (explicit) instances of middles is obvious:

ergative middles always have one participant only, whereas the examples above clearly have two. A

characterisation as an effective structure is out of the question too, since diagnostic tests like

passivisation and pseudo-clefting yield ungrammatical results:

(23) a. *An arm was fractured by him in the accident. (Davidse 1992: 127, ex. 1)

b. *What he did to his arm was fracture it in the accident. (Ibid., ex. 1)

(24) a. *A fuse was blown by the house. (Ibid., ex. 4)

b. *What the house did to a fuse was blow it. (Ibid., ex. 4)

26

As has already been pointed out above, the defining feature of these constructions lies in the nature

of the subject, since it does not function as the agent of the action, but rather as a “circumstantial”

participant (Davidse 1992: 128). A more explicit rendering of this circumstantial subject role results in

constructions of the following type:

(25) His arm fractured on him in the car accident (Davidse 1992: 128, ex. 1b)

(26) A fuse blew in the house (Ibid., ex. 4b).

In fact, we now have an ergative middle supplemented with an adverbial of place (marked in italics),

which means that “the ergative pseudo-effective construction is [...] basically a middle construction,

semantically as well as grammatically”, but one that differs from a prototypical middle in its mapping

of the involved participants (Davidse 1992: 128).

Pseudo-effectives can also be distinguished on the transitive level and, again, passivisation and

pseudo-clefting lead to ungrammatical results:

(27) They danced an energetic jig. (Davidse 1992: 124, unnumbered ex.)

(28) *An energetic jig was danced (by them). (Ibid., unnumbered ex.)

(29) *What they did to an energetic jig was dance it. (Ibid., unnumbered ex.)

Just like in ergative pseudo-effective structures, transitive pseudo-effectives are also characterised by

a subject that does not fulfil its prototypical agent-role. However, contrary to their ergative

counterparts, this neglect of duty is not due to the subject itself, but because of the fact that another

constituent prevents it from carrying out its function. This element, which corresponds to an

energetic jig in the example above and is also known as the Range, floats somewhere “in between

participant and circumstance” and can therefore not be ‘acted on’ (Davidse 1992: 124-125).

27

1.3 Tests for ergativity

One of the tests most frequently appealed to to identify ergative verbs is the so-called ‘sanctioning of

from-phrases’ (Alexiadou et al. 2005; Kalluli 2006). It seems to be universal property of English

inchoative constructions that they can express “the external cause of an event” by means of a

prepositional phrase introduced by from (Kalluli 2006: 202-203; Kalluli 2007: 771-772). By-phrases,

on the other hand, are only grammatical in passive constructions:

(30) a. The window cracked from the pressure. (Kalluli 2006: 203, ex. 6b.)

b. *The window cracked by the pressure. (Ibid., ex. 6a.)

(31) a. *The window was cracked from the pressure.

b. The window was cracked by the pressure.

However, there are a few important restrictions in the occurrence of from-phrases. First of all, Kalluli

argues that these complements are only grammatical “with what Levin and Rappaport Hovav (1995)

refer to as external causation verbs” (Kalluli 2006: 203) and secondly, from-phrases can only express

causers and causing events, as in (30a) (which is repeated here as (32)) and (33) respectively, but not

agents or instruments, as in (34) (Alexiadou 2005: 8; Kalluli 2006: 203):

(32) The window cracked from the pressure. (Kalluli 2006: 203, ex. 6b.)

(33) The window cracked from the explosion. (Alexiadou et al. 2005: 8, ex. 27)

(34) The door opened *from Mary / *from the key. (Ibid., ex. 28)10

Moreover, this diagnostic test is also valid for German anticausatives, since they behave in the exact

same way as their “English counterparts” (Alexiadou et al. 2005: 10). Alexiadou et al. argue that “[i]n

German, agents are introduced by von [...], instruments by mit [...], causers/natural forces by durch

[...], and causing events by durch [...]” (Alexiadou et al. 2005: 9). Thus, the diagnostic status fulfilled

by from-phrases in English is carried out by durch-complements in German:

(35) Die Vase zerbrach durch ein Erdbeben. (Alexiadou et al. 2005: 11, ex. 35a.)

10

As illustrated by this example, Kalluli 2006’s generalisation that “[f]rom-phrases identifying the external cause in anticausatives cannot combine with participants capable of wilful agency” is too narrow, since it does not take into account the ungrammaticality of instruments expressed in from-phrases (Kalluli 2006: 448).

28

(36) Die Luftqualität im Raum verschlechtert sich durch das Rauchen von Zigaretten massiv.

(Ibid., ex. 36)

(37) Die Vase zerbrach *von Peter / *mit dem Hammer. (Alexiadou et al. 2005: 10, ex. 34a.)

A test related to the sanctioning of from-phrases is the impossibility for the Agent of a causative

structure “to appear as an of-Complement”, just like in (38) (Davidse 1992: 112). However, this

diagnostic is not completely watertight, since what Davidse terms ‘accidental transitive clauses’

equally disallow such adjuncts (see (39)):

(38) a. Jill started the uproar.

b. *The starting of Jill.

c. The starting of the uproar.

(39) a. Jill grazed her hand. (Davidse 1992: 112)

b. *The grazing of Jill. (Ibid.)

c. The grazing of her hand. (Ibid.)

Another way of finding out whether a specific verb can be used inchoatively (and, thus, perhaps

ergatively) is by submitting it to the by itself11-probe (Alexiadou et al. 2005: 8). This diagnostic can

also by applied in German, by using the adverbial complement von selbst (Alexiadou et al. 2005: 11):

(40) The uproar started by itself.

(41) Der Aufruhr fing von selbst an.

Other tests include the fact that the causative alternant of an ergative verb can always be converted

to an analytical causative (as in (42)), whereas a traditional transitive cannot (as in (43)) and that an

ergative verb of course allows for its causer to be left unspecified in an inchoative construction (see

(44b)) (Davidse 1998: 58). However, as Davidse correctly points out, this does not hold for the causee

of a causative construction, since its elision will inevitably lead to a disruption of the original

semantics. For example, a reduction of a sentence like Jill broke the vase to Jill broke will inevitably

come to mean that Jill herself underwent the breaking (rather than the vase). Thus, as illustrated by

11

“in the interpretation of ‘without outside help’” (Alexiadou et al. 2005: 8).

29

(44c), “[in] ergative processes, it is totally impossible to isolate the Agent-Process complex in an

absolute construction” (Davidse 1992: 113).

(42) a. Jill broke the vase.

b. Jill caused the vase to break.

(43) a. Jack failed his exam.

b. *Jack caused his exam to fail.

(44) a. Jill broke the vase.

b. The vase broke.

c. *Jill broke.

The last diagnostic I will discuss in this chapter, namely “the derivation of agentive nominals through

the -er suffix”, allows for the identification of ergatives through negative deduction, since it generally

only applies to transitive verbs (Lemmens 1998: 127). The transitive inclination displayed by this

derivational morpheme should not come as a surprise, for transitive verbs are, according to Davidse’s

framework, essentially agent-centred. Ergatives, on the other hand, focus more on the affected

participant in the action, the so-called Medium. In order to provide an empirical basis for this

theoretical premise, Lemmens carried out a pilot study based on more than 45,000 -er-nominals,

which he divided into either of the following classes: (1) transitive, (2) intransitive, (3) ergative or (4)

transitive & ergative12. The results his study yielded, can be found below:

Verb type Absolute frequency Relative frequency

transitive 35395 77.6%

intransitive 8065 17.7%

ergative 1182 2.6%

transitive & ergative 974 2.1%

TOTAL 45616 100.0%

Table 10: Overview of verb types according to their number of -er-nominalisations (source: Lemmens

1998: 133)

12

The ‘transitive & intransitive’ category was attributed verbs showing “mixed characteristics, like sell” (Lemmens 1998: 133).

30

As indicated by the table above, the results Lemmens’s study yielded were highly significant: whereas

transitive -er-nominalisations were found in more than three quarters of the cases, their ergatively-

based counterparts were nearly inexistent. Among the rare exceptions were nominals, like choker,

“someone (or something) that chokes somebody else” (Lemmens 1998: 131), but given their

extremely low number of occurrence, I do not see why this diagnostic cannot be used at least as

supplementary evidence to negatively identify ergative verbs.

31

1.4 Constraints on ergativisation

It has been argued in the literature that ergativisation (i.e. the possibility for a verb to be used both

in an inchoative as well as in a causative alternation without changing its external form) is subject to

a number of syntactic and semantic restrictions. Since inchoative constructions present a change as

coming about spontaneously, it is important that a potential ergative verb does not imply the use of

all too specific “instruments or methods” (Haspelmath 1987: 15; Haspelmath & Müller-Bardey 1991:

5; Haspelmath 1993: 93). In other words, an ergative verb needs to be clear of so-called “agent-

oriented meaning components” (Haspelmath 1993: 93). Below, two examples are provided (both for

English and for German) to illustrate this principle:

(45) a. The girl tore her pants. (Haspelmath 1993: 93, ex. 11a.)

b. The pants tore. (Ibid., ex. 11b.)

(46) a. The tailor cut the cloth. (Ibid., ex. 12a.)

b. *The cloth cut. (Ibid., ex. 12b.)

(47) a. Florian zerreißt das Papier.

b. Das Papier zerreißt. (Haspelmath 1987: 16, ex. 28)

(48) a. Ayse schneidet das Papier. (Ibid., ex. 25)

b. *Das Papier schneidet (sich). (Ibid.)

As can clearly be deduced from examples above, the verbs ‘cut’ and ‘schneiden’ disallow inchoative

alternations, since they imply the use of a “sharp instrument”. ‘Tear’ and ‘zerreißen’, on the other

hand, lack this additional meaning component and consequently do allow for an inchoative use

(Haspelmath 1993: 93). However, Haspelmath argues that his proposition on agent-oriented features

probably needs to generalised, since verbs like ‘decapitate’, which do not seem to presuppose any

direct participation of a particular agent, equally disallow an inchoative use. The adjusted hypothesis

thus reads as follows (Haspelmath 1993: 93):

A verb meaning that refers to a change of state or a going-on may appear in an inchoative/causative

alternation unless the verb contains agent-oriented meaning components or other highly specific

meaning components that make the spontaneous occurrence of the event extremely unlikely.

(Haspelmath 1993: 94)

32

A second semantic restriction on ergativisation, Haspelmath argues, is the fact that ergative verbs

(and causative-inchoative pairs more generally) should always denote “a change of state or a going-

on” (Haspelmath 1993: 92-93). This means that verbs like ‘help’, ‘read’ or ‘invite’, since they do not

refer to an agent bringing about a certain change to a specific situation, can never be used as

ergatives. According to Haspelmath 1993, this also goes for so-called ‘agentive intransitive verbs’, like

‘work’ or ‘dance’, since “they are not conceived of as occurring spontaneously” (Haspelmath 1993:

93). However, Davidse 1992 clearly states the example of The boss works his secretaries from eight

till six in her discussion on direct and indirect causation, so it seems that Haspelmath’s last restriction

needs to be at least partially reformulated or nuanced (Davidse 1992: 119). This brings us to yet

another misconception about ergatives, namely that they can only be derived from a transitive base

(Alexiadou et al. 2005: 4; Comrie 1981: 161; Garcia Garcia 2012: 123; Haspelmath 1987: 14). Davidse

& Gheyskens 1997 provide ample counterevidence to this claim by arguing that there exists a certain

construction which they term an ‘ergative causativization of an intransitive’ (Davidse & Gheyskens

1997; henceforth ECI). It appears that ergative causativizations “typically designate ‘deliberate’

actions, in which an animate Actor is in control of the action [...]” (Davidse & Gheyskens 1997: 9; see

ex. (49)). However – albeit much less frequently – it may also concern ‘superventive’ verbs, meaning

that the Actor is not in control of the action described (Davidse 1992: 120; see ex. (50)):

(49) a. The horse walked (Lyons 1968: 365 as cited in Davidse & Gheyskens 1997: 8, ex.

19).

b. John walked the horse (Lyons 1968: 365 as cited in Davidse & Gheyskens 1997: 8,

ex. 20).

(50) a. The baby burped (Smith 1970: 107 as cited in Davidse & Gheyskens 1997: 9, ex.

26).

b. The nurse burped the baby (Smith 1970: 107 as cited in Davidse & Gheyskens

1997: 9, ex. 27).

As can be deduced from the examples cited above, it is a typical feature of ECIs that they exhibit two

causers instead of one. The main causer in such constructions – i.e. the one occupying the syntactic

slot associated with the subject – always needs to be “human and deliberate” in nature (Davidse &

Gheyskens 1997: 10). Davidse & Gheyskens argue that he is the one who ultimately instigates the

action “by transmitting his will, which goes hand-in-hand with a loss of volitional control on the part

of the causee”. However, the causee’s willingness is not completely reduced to zero, since his co-

participation is required to bring about the desired effect (Davidse & Gheyskens 1997: 10). Going

33

back to the examples above, it would be impossible for John to actually ‘walk’ his horse if the latter

would obstinately refuse to move its feet, just like it would be impossible for the nurse to make the

baby burp if the infant would not want to push out the excess air out of its lungs. In the tradition

initiated by M.A.K. Halliday, the causee (or ‘Medium’ in the terminology of Davidse 1992) is therefore

also termed the ‘enforced Actor’ (Halliday 1968: 185 as cited in Davidse & Gheyskens 1997: 10). It

needs to be noted that, in this respect, ECIs behave in the exact same way as the so-called

accusativus cum infinitivo (cf. supra), a construction in which the object of the main clause also fulfils

the role of agent in the embedded clause and whose verb appears in the infinitive. Typical examples

include She wants you to go home and I let her speak. Going back the the ECI, Davidse & Gheyskens

point out that their enforced Actors typically concern “subjugated animals and humans, and

incapable humans, such as infants” (Davidse & Gheyskens 1997: 9). Given this high number of

semantic restrictions, it should not come as a surprise that ECIs are in fact “statistically very

infrequent” (Davidse & Gheyskens 1997: 10). A corpus study based on what Levin 1993 terms ‘run

verbs’13 as carried out by Davidse & Gheyskens only yielded 26 ECIs out of a total of 1940 sample

sentences (Davidse & Gheyskens 1997: 19; Levin 1993: 31), or a mere 1.34%. Despite this meagre

turnout, it was nevertheless possible to distinguish between six different subtypes, but for briefness’

sake, those will not be discussed here.

Finally, it should also be noted that there are not only selection restrictions on a potential ergative

verb, but also on its Medium. Gheyskens and Davidse argue that it can typically be found somewhere

in between a prototypical animate entity bringing about a certain change and an inanimate entity

undergoing the action, meaning that

[...] nominals functioning as Medium should designate entities which can be conceptualized as ‘being

inherently inclined to V-ergative’ and as ‘quasi-autonomously manifesting V-ergative’. (Davidse &

Gheyskens 1997: 7)

In fact, this selection principle is so strong, that it can even determine the ergative character of a

specific verb altogether (Alexiadou et al. 2005: 6). ‘Lengthen’, for instance, is clearly ergative when it

refers to the lengthening of days, but not with respect to clothes. Or, to use Levin & Rappaport

Hovav’s words: “there are certain things that can be lengthened, but do not lengthen [themselves]”

(Levin & Rappaport Hovav 1995: 86):

13

i.e. canter, drive, fly, gallop, jump, leap, march, race, run, swim, trot and walk (Levin 1993: 31).

34

(51) a. The dressmaker lengthened the skirt. (Levin & Rappaport Hovav 1995: 86, ex.

14a.)

b. *The skirt lengthened. (Ibid., ex. 14b.)

(52) a. The mad scientist lengthened the days. (Ibid., ex. 15a.)

b. The days lengthened. (Ibid., ex. 15b.)

35

1.5 Unaccusativity and unaccusative diagnostics

1.5.1 Unaccusativity

The study of ergativity is closely related to what in formal traditions has been termed

‘unaccusativity’, a phenomenon first described by Perlmutter 1978 as part of the so-called

unaccusative hypothesis. This hypothesis essentially postulated the existence of two types of

intransitive verbs, notably unaccusative and unergative verbs (Alexiadou et al. 2004: 1-2; Levin &

Rappaport Hovav 1994: 37, 58). This is also the reason why unaccusativity has sometimes been

referred to as ‘split intransitivity’ (Levin & Rappaport Hovav 1994: 37). By and large, it is possible to

distinguish between two different approaches to unaccusativity. On the one hand, there are the

proponents of the syntactic approach, claiming that there is a “particular syntactic configuration” all

unaccusative verbs have in common, regardless of their meaning (Levin & Rappaport Hovav 1995: 5).

Here, the distinction between unergatives and unaccusatives hinges on a distinction in their

“underlying syntactic behaviour” (Alexiadou et al. 2004: 2): whereas the former do not experience a

transformation from deep to surface structure and thus realise their initial subject as a final subject,

the latter have a surface-level subject going back to an original deep-level object (Alexiadou et al.

2004: 2; Haspelmath 1987: 6). In other words, unaccusative verbs allow for the promotion of a

prototypical direct object to subject position (Perlmutter 1978: 159, 160). On the other hand, there

are those who believe that unaccusatives should be distinguished on the basis of semantic criteria.

This denial of a syntactic encoding obviously entails “the need to attribute different syntactic

representations” to the verbs believed to belong to this class (Levin & Rappaport Hovav 1995: 10).

Before we move on, it is important to point out that unaccusatives do not show a complete overlap

with what we have been terming ergative/labile verbs. In fact, Alexiadou et al. argue that ergative

verbs are merely a subgroup of the class of unaccusatives (Alexiadou et al. 2004: 2). Davidse makes a

similar observation, perspicuously noting that

Perlmutter’s “unaccusative” verbs include all the “ergative” verbs of the British tradition as well as the

‘superventive’ intransitive verbs from the British tradition such as fall, die, stumble. Perlmutter’s

“unaccusative” verbs are those in which the main thematic role ‘undergoes’ some change of location

or state. (Davidse 1998: 158; not my italics)

Within the formal tradition, there has also been much debate as to whether there are objective tests

to determine if a given verb can be classified as ‘unaccusative’. In the following overview, I will

provide a brief discussion of these so-called ‘unaccusative diagnostics’, an overview which will

essentially be based on the introduction to Alexiadou et al.’s edited volume The Unaccusativity

36

Puzzle. Since unaccusatives encompass more than just ergatives (cf. supra), one could argue that a

discussion of the properties of these verbs is not completely relevant to a study of ergative verbs in

the sense of verbs allowing for a causative-inchoative alternation. Indeed, these tests do not have

the same validity as the sanctioning of from-phrases or by itself-complements, but that does not

mean they cannot serve as additional evidence. In other words, these diagnostics are to be

considered ‘necessary’ (but not ‘sufficient’) conditions for ergativity. Since my paper specifically

focuses on ergative phenomena in English and German, it is evident that I will limit myself to those

diagnostics applying to either or both of these languages alone, and I will not go into detail about

phenomena which have been ascribed diagnostic status in other languages.

1.5.2 Unaccusative diagnostics

1.5.2.1 Possibility to appear in resultative constructions

A phenomenon that has often been considered a reliable test for unaccusativity is the possibility for

unaccusatives, but not unergatives, to appear in a resultative construction. According to Alexiadou et

al. and as illustrated by the following examples, such a phrase “denotes the state achieved by the

referent of the NP it is predicated of as a result of the action denoted by the verb” (Alexiadou et al.

2004: 5):

(53) Unaccusative a. The bottle broke open. (Alexiadou et al. 2004: 5, ex. 15c.)

b. Die Flasche brach offen. (based on Alexiadou et al. 2004: 5, ex.

15c.)

(54) Unergative a. *Dora shouted hoarse. (Alexiadou et al. 2004: 5, ex. 15b.)

b. *Dora rief heiser. (based on Alexiadou et al. 2004: 5, ex. 15b.)

37

1.5.2.2 Prenominal perfect/passive participles

A second test to distinguish unaccusatives from unergatives is the fact that the former’s participles

can be used attributively to modify noun phrases (a feature they share with transitive verbs).

Examples of this diagnostic can be found below:

(55) Transitive a. The cleaned room.

b. Das uneingeleitete Duzen.

(56) Unaccusative a. The withered leaves.

b. Das gefallene Mädchen.

(57) Unergative a. *The laughed king.

b. *Das geweinte Kind.

1.5.2.3 German-exclusive diagnostics

1.5.2.3.1 Auxiliary selection

One of the most widely discussed tests to demonstrate unaccusativity concerns the auxiliary

selection of a given verb. Most Germanic and Romance languages – with English as one of the rare

exceptions – have two distinct auxiliaries at their disposal, more precisely the equivalents of be and

have. Be-selection is traditionally associated with unaccusativity, have-selection with unergativity

(Alexiadou et al. 2004: 5; Davidse 1998: 158; Kalluli 2006: 211):

(58) Peter ist gestern unmittelbar nach Hause gegangen.

(59) Peter hat gestern zwei Äpfel gegessen.

However, Keller & Sorace 2001 argue that auxiliary selection is by no means an either-or story. They

cite a semantically based hierarchy of verb classes (composed by Sorace 2000), which illustrates that

the boundary between have and be-selection tends to fade for classes halfway the continuum (Keller

& Sorace 2001: 60):

38

Change of location selects BE (least variation)

Change of state

Continuation of state

Existence of state

Uncontrolled process

Controlled process (motional)

Controlled process (non-motional) selects HAVE (least variation)

Figure 2: The auxiliary selection hierarchy (source: Sorace 2000, as cited in Keller & Sorace 2001: 60)

This blurred distinction is not only evident between different verbs belonging to the same semantic

class, it can also be observed within one and the same verb. The German verb verweilen ‘to stay’, for

instance, is definitely grammatical in constructions with haben, but it remains unclear to what extent

its appearance with the auxiliary sein would be ungrammatical (Keller & Sorace 2001: 67):

(60) a. Der Wanderer hat kurz verweilt. (Keller & Sorace 2001: 67, ex. 13)

b. ?Der Wanderer ist kurz verweilt. (Keller & Sorace 2001: 67, ex. 13)

1.5.2.3.2 Impersonal passives

Impersonal passives can only be built from unergatives, so they can be used to ‘negatively’ identify

unaccusatives. A construction of this kind is impossible in English.

(61) a. Es wird hier von den jungen Leuten viel getanzt. (translated from Alexiadou et al.

2004, ex. 19a.)

b. Es wird von den Kindern in Amsterdam geblieben. (translated from Alexiadou et

al. 2004, ex. 19b.)

1.5.2.3.3 Split phrases

Another diagnostic exclusively applicable to German is the use of split phrases. Here, the rule seems

to be that split phrases “are not allowed in subject position of transitives [...] and unergatives [...],

but are allowed for unaccusatives [...]:

(62) Transitive *Studenten haben fleißige das Seminar besucht (Alexiadou et al.

2004: 7, ex. 21a.).

39

(63) Unaccusative Widersprüche sind dem Richter mehrere aufgefallen (Ibid., ex.

21d.).

(64) Unergative *Studenten haben fleißige telefoniert (Ibid., ex. 21b.).

1.5.2.3.4 Nominal attributability and agent nominals

Haider 1984 (as cited in Abraham 2001) makes mention of two additional German-exclusive

diagnostics, namely nominal attributability (65) and the formation of agent nominals (66). As can be

deduced from the examples below, the former construction is ungrammatical with unergatives,

whereas the latter leads to an ungrammatical result with unaccusatives:

(65) a. Unergative *Der geschlafene Junge (Abraham 2001: 214, ex. 2a.).

b. Unaccusative Der eingeschlafene Junge (Ibid.).

(66) a. Unergative Der Schläfer (Ibid., ex. 2c.).

b. Unaccusative *Der Einschläfer (Ibid.).

40

1.6 A note on the middle voice

The fact that ergativity and the middle voice are two closely intertwined phenomena can clearly be

deduced from the way in which they are treated in the literature: whereas some sources consider

certain structures fine examples of a middle, others might regard those very same constructions as

ergatives and vice-versa. Kemmer 1993, for instance, treats sentences like The door opened on a par

with constructions like The book sells well, although the former is typically considered an ergative

and the latter a middle (Kemmer 1993: 2). On the other hand, there are authors like Francis et al.

who neutralise the contrast between both structures in favour of the ergative. They argue that

“[m]ost ergative verbs can be used to indicate events that have taken place (actual events), or events

that might take place (potential events)” (Francis et al. 1996), a contrast which is illustrated by the

following (slightly adapted) examples:

(67) a. After you have stepped from a warm bath, you should apply the cream evenly

over your body. (Francis et al. 1996, in-text ex.)

b. This cream smells clean and fresh, and applies easily. (Ibid., in-text ex.)

According to Francis et al., a middle is a construction which is used to refer to potential events, just

like (67)b. (67)a, on the other hand, indicates its actual counterpart. Although the authors of Collins

Cobuild do not elaborate any further on this claim, a link between middles and potential events has

also been suggested by Haspelmath 1987. He proposes the term ‘potential passive’ and states the

following examples from English and German respectively:

(68) Bureaucrats bribe easily. (Haspelmath 1987: 7, ex. 22)

(69) Dieses Buch verkauft sich nicht/gut/nur in Unibuchläden. (Haspelmath 1987: 7, ex. 21)

A typical feature of middles is the fact that they “are often marked like ordinary passives”, either by

inflectional means or – as in German – through reflexivisation (Haspelmath 1987: 7). However, this

does not mean that all reflexive verbs also allow for a middle alternant: middles are a special type of

reflexive, just like squares are a special type of rectangle (Fagan 1992: 2). It is important to realise,

however, that English violates the aforementioned principle of morphological marking, since it

merely relies on its syntax to express a middle construction. Next to that, Haspelmath argues that

middles attain a high degree of lexical generality, contrary to traditional anticausatives, and that they

do not overtly express the actor (Haspelmath 1987: 7). Despite this lack of an explicit agent, Fagan

41

argues that middle constructions do have so-called “understood subjects, just like passive clauses”

(Fagan 1992: 3). But, contrary to passives, middles do not allow for an overt appearance of the

implied agent. The following English and German examples clearly support this claim:

(70) a. passive: Das Buch wird von vielen Lesern gelesen. (Fagan 1992: 3, ex. 8)

The book is read by many readers. (based on Fagan 1992: 3, ex. 8)

b. middle: *Das Buch liest sich leicht von vielen Lesern. (Fagan 1992: 3, ex. 8)

*The book reads easily by many readers. (based on Fagan 1992: 3,

ex. 8)

Note that the presupposition of an implied agent in middles is a clear argument against a unified

analysis of middle and ergative constructions, since it is a generally accepted that ergative

constructions refer to actions or events that seem to happen without the intervention of an external

force or initiator (Francis et al. 1996; Haspelmath 1987: 15; Haspelmath & Müller-Bardey 1991: 5;

Haspelmath 1993: 93).

With respect to the differences between English and German middles, Fagan mentions two

dissimilarities, which are “morphosyntactic rather than semantic” in nature (Fagan 1992: 52). First

and foremost, English does not have reflexive pronouns at its disposal, which obviously entails a

complete absence of reflexive middles. Apart from that, Fagan also argues that English only allows

transitive verbs to appear in middle constructions, whereas German also accepts intransitive middles

(more precisely in impersonal constructions):

(71) a. *It lives well here. (Fagan 1992: 52, ex. 97b.)

b. Es lebt sich gut in dieser Stadt. (based on Fagan 1992: 50, ex. 97b.)

Davidse & Heyvaert 1997 mention that this restriction on English middle forms has been formulated

both in generative (e.g. by Keyser and Roeper 1984 and Fellbaum and Zribi-Hertz 1989) as well as in

functional traditions (e.g. by Halliday 1967 and Fawcett 1980) and is known as the so-called

Transitive Constraint (Davidse & Heyvaert 1997: 43-44). However, the authors explicitly challenge

this principle by arguing that there exists “a subclass of English middles with intransitive verbs whose

subjects are circumstances of either location or instrument” (Davidse & Heyvaert 1997: 50). An

instance of either type can be found below ((72) illustrates the location type, (73) is an instance of a

middle with an instrument subject):

42

(72) A middle for diddle draw on Yarmouth’s straight course means a fancied runner can

switch to either rail depending on which side is riding faster. (Davidse & Heyvaert 1997: 45,

ex. 30)

(73) Narrow tyres manoeuvre more easily. (Davidse & Heyvaert 1997: 49, ex. 38)

With regard to the relation between the middle and the ergative, Davidse & Heyvaert mention that

“diachronically, the ergative intransitive is [possibly] the source from which the middle developed”

(Davidse & Heyvaert 1997: 71), but since that claim is beyond the scope of their study, they do not

further pursue it. However, they do discuss a number of synchronic analogies between ergatives and

middles. First of all, it is argued that, in principle, all ergatives allow for a middle alternant, yet the

reverse is not necessarily true14. Precisely this observation (i.e. the fact that this claim does not have

bidirectional validity) is justly considered problematic by the authors, since it questions the extent to

which ergative and middle are related structures (Davidse & Heyvaert 1997: 72). A second link

between the structures under consideration posited by Davidse & Heyvaert is morphological in

nature. They refer to the observation made by Kemmer that “crosslinguistically, ergative intransitives

[i.e. what we have termed inchoative constructions] and middles tend to have the same formal

marking”, e.g. reflexivisation, as in German (Kemmer 1992 as cited in Davidse & Heyvaert 1997: 72).

Finally, Davidse & Heyvaert argue that the structures under consideration are also related through

what they call a “process of subjectification”, which they define as “a shift from the description of a

verifiable state of affairs to a subjective statement of dynamic modality” (Davidse & Heyvaert 1997:

73). In essence, this claim boils down to the fact that middles always make “conduciveness

judgements” about a certain entity, an inclination towards a certain action rather than an actual

event (Davidse & Heyvaert 1997: 73).

14

The geometrical metaphor I used a few pages back for describing the relation between German middles and reflexives (i.e. the fact that all squares are rectangles, but not the other way around) applies here as well.

43

2 Section II – Analysis

2.1 Aim and Methodology

As already pointed out in the introduction, the aim of this study was to find out what strategies

German uses to express English labile verbs of starting and stopping. This set was taken from Collins

COBUILD Grammar Patterns, one of the few reference works on English syntax providing an extensive

(semantically based) overview of ergative verbs. The main criterion playing a role in the selection of

this set was quantitative in nature, as I aimed for a group of about twenty tokens in order to

safeguard the feasibility of this research project. Therefore, I first determined the number of verbs in

each individual set. The ‘start’ and ‘stop’ group eventually turned out to be an ideal candidate, since

it contained exactly twenty tokens, among which four prepositional verbs.

(74) The ‘start’ and ‘stop’ group

adjourn – begin – break up – commence – continue – convene – end – halt – kick off –

recommence – reconvene – restart – resume – stall – start – still (a sound) – stop – strike up

(a tune) – taper off – terminate

This set was subsequently translated into German with the help of the online Collins Dictionary

(http://www.collinsdictionary.com/) and the English source-words along with their German

equivalents and contexts of use listed in a digital inventory. The acquired verbs then served as the

input to a ‘retranslation operation’ in order to fill any potential anomalies in the original sample. This

strategy proved to be quite successful, as I was able to identify nine additional labile verbs which also

satisfied the semantic criteria of the basic set. The extended ‘start’ and ‘stop’ group thus looked as

follows:

(75) The extended ‘start’ and ‘stop’ group15

abort – adjourn – assemble – begin – break off – break up – carry on – cease – commence –

continue – convene – delay – end – ease – finish – halt – kick off – postpone – recommence –

reconvene – restart – resume – stall – start – still (a sound) – stop –strike up (a tune) – taper

off – terminate

15

New entries are marked in bold.

44

However, retranslating the German data into English also yielded a second, somewhat unexpected

result. Consider the following table:

English source-

word

German translation English retranslation

transitive intransitive transitive intransitive

stop stoppen X stop stop

Table 11: Translation-chain for the verb ‘stop’.

Table 11 above illustrates the translation chain stop – stoppen – stop. As can be observed, the English

entry only generates transitive stoppen as a valid translation, whereas the retranslation of stoppen

clearly seems to indicate that the German verb is actually labile. (rather than merely transitive).

However, dictionaries are in this respect fundamentally unreliable, since they merely associate

cognates with each other. Evidently, these ‘false ergatives’ were left out of the German sample.

Of course, this sample needed to be thinned out in a number of ways. As has been frequently

observed in the past (e.g. Fontenelle 1996; Montemagni 1994), dictionaries often do not accept the

ergative pattern as an independent verb alternation. Instead, they only tend to accept the traditional

dichotomy between the transitive and the intransitive contexts of use. A typical example of this so-

called “splitting strategy” (Fontenelle 1996: 212) is provided below:

(76) to break off

Intransitive verb

1. [branch, piece] abbrechen

a. (= stop) aufhören

b. (= stop speaking) abbrechen

(temporarily) unterbrechen

Transitive verb

1. twig, piece etc. abbrechen

2. a. negotiations, relations abbrechen

b. engagement lösen

45

In other words, a verb can only be considered ergative if it has both a transitive as well as an

intransitive entry. The lexicographic entry above clearly instantiates this splitting strategy for the

verb abbrechen when referring to twigs or branches, since the inchoative alternant is seen as

intransitive and the causative alternant as transitive. Therefore, the verbs in my sample not meeting

this requirement, notably taper off, still (as sound) and postpone, were left out of consideration. Two

other tokens I had to bar, albeit for different reasons, included recommence and reconvene. The

reason was that these verbs simply did not generate any hits at all, neither in the Collins Dictionary,

nor in the Oxford Dictionaries (http://oxforddictionaries.com/), an additional source I consulted in

case the former proved to be incomplete. It is evident that I also tried to preserve the semantics of

the original sample. German translations not pertaining to ‘starting’ and ‘stopping’ were therefore

excluded as well16. Finally, I also left out all too specific terminology, e.g. German legen, meaning ‘to

start’ in the sense of ‘starting a fire’, as well as metaphorical language, like scheitern, ‘to go wrong’.

After having left out all unsuitable German tokens, my sample comprised 112 translations in total, of

which 66 transitive and 46 intransitive verbs or constructions. In principle, every translation was

counted in, even if it concerned a passive, an alternation which can essentially be formed for every

transitive verb, but was only supplied for two of them. Ergative verbs, too, were seen as double

entries. Despite the fact that the causative and inchoative alternants are in their case identical, that

does not mean that they do not distinguish between causative and inchoative strands of meaning.

After having collected and filtered out the data, I grouped them according to way in which they

expressed the ergative alternation. In total, 8 categories could be discerned: (1) ergative verbs, (2)

reflexive verbs, (3) verbs with preverbal elements, (4) passives, (5) Function Verb Phrases

(‘Funktionsverbgefüge’), (6) verbs with a prepositional complement, (7) lassen-periphrases and (8) a

group of remaining verbs containing all tokens not pertaining to any of the categories above.17 It

should be stressed that this study was fundamentally based on ‘dead’ lexicographical data, so in

order to increase their pragmatic value, I included several bilingual examples from the open source

parallel corpus Opus (http://opus.lingfil.uu.se/) to instantiate the translation strategies in question.

16

A good example constitutes the English verb break off, one of the translations of which reading abbrechen with regard to branches (of a tree). 17

Verbs satisfying the criteria for more than one group, e.g. reflexive prefix verbs, like sich vertagen, were obviously counted twice.

46

2.2 Discussion and results

2.2.1 Ergative verbs

Transitive Intransitive

beginnen beginnen

anfangen anfangen

wieder beginnen wieder beginnen

neu beginnen neu beginnen

starten starten

anhalten anhalten

abbrechen abbrechen

unterbrechen unterbrechen

Table 12: Overview of acquired ergative verbs

In absolute figures, my sample displayed a total of eight ergative verbs, which corresponds to

approximately 7% of all translation strategies. With the exception of starten, all but one of the

acquired tokens in this category are verbs with preverbal elements. As will be demonstrated later in

this chapter, this verb class can be subdivided in prefix verbs on the one hand, and particle verbs on

the other. One of the most crucial factors to categorise a specific verb as one but not the other

resides in the preverbal element’s relation to the verb stem: prefixes have to remain bound to the

stem at all times, but particles can occasionally be severed from their base. Thus, ergative particle

verbs include anfangen, anhalten and abbrechen, whereas ergative prefix verbs encompass

unterbrechen as well as beginnen and all of its compounds (i.e. wieder beginnen and neu beginnen).

The meaning of unterbrechen is still largely recoverable from the sum of its parts, but beginnen is an

instance of a so-called lexicalised prefix verbs, meaning that it cannot be meaningfully reduced to its

original base form. Furthermore, the observation that seven out of eight ergative verbs are formed

by means of preverbal elements is completely in line with Letuchij’s claim that crosslinguistically,

there is a strong tendency for derived verbs to become ergative (Letuchij 2004: 3). Of course, the fact

that a derivative is labile does not necessarily imply that its base should be labile too. Letuchij cites

the example of anfangen, a token which also shows up in my sample: its base form (notably fangen)

can only be used transitively (Letuchij 2004: 2).

Because of the fact that all verbs in my sample had to satisfy the semantic criteria of ‘starting and

stopping’, they logically also belong to the class of so-called ‘phase verbs’, a group which according to

Letuchij 2004 shows a natural inclination towards lability. Within this group, those verbs referring to

the initial stage of an action (also termed ‘ingressive’ verbs) “are labile more often than verbs

47

designating other phases of the situation” (Letuchij 2004: 2), a proposition which also seems to be

borne out by my sample, since five out of eight tokens denote the initial stage of a specific action

(Hentschel & Weydt 2003: 41). These verbs include beginnen, anfangen, wieder beginnen, neu

beginnen and starten. In any way, the number of ergative verbs in German was considerably lower

than in English, which seems to confirm the generally accepted thesis that “Modern English differs

markedly from its Germanic neighbours in having more labile verbs” (Van Gelderen 2011: 106; see

also Baron 1973: 305; Garcia Garcia 2012: 138; Haspelmath 1987: 17). An example of the ergative

translation strategy (taken from Opus, the Open Source Parallel Corpus) is provided below:

(77) English transitive use: Mr President, rapporteur Bowis started his explanatory statement

on his report with a quotation from Ludwig Feuerbach , which he also repeated this morning:

“der Mensch ist was er isst” (You are what you eat ). <Opus>

(78) German transitive use: Der Berichterstatter, Herr Bowis, beginnt die Begründung seines

Berichts mit einem Zitat von Ludwig Feuerbach , das er auch heute Vormittag wiederholt hat

: Der Mensch ist , was er isst. <Opus>

(79) English intransitive use: The economic recovery , which started in the second half of

2003 , has continued this year. <Opus>

(80) German intransitive use: Die konjunkturelle Erholung, die in der zweiten Jahreshälfte

2003 begann, setzt sich auch in diesem Jahr fort. <Opus>

2.2.2 Reflexive verbs

Transitive Intransitive

vertagen sich vertagen

auflösen sich auflösen

X sich trennen

entspannen sich entspannen

X sich versammeln

Table 13: Overview of acquired German reflexive verbs along with their transitive counterparts (if

available)

48

Within my sample, reflexivisation was the third most common translation strategy, only exceeded by

labile verbs and function verb phrases. This relatively high number is in line with Garcia Garcia’s

observation that “[in] modern European languages, it is quite common to use the reflexive pronoun

as an anticausative particle” (Garcia Garcia 2012: 125). Since reflexives are characterised by a

convergence of the grammatical categories of subject and object (Haspelmath & Müller-Bardey 1991:

7; Hentschel & Weydt 2003: 67), it is evident that they only show up as intransitive entries. Within

this group, they show up five out of forty-six times, a figure which equates to almost 11% percent of

all intransitive tokens. Three out of five reflexives appeared in a causative-inchoative alternation,

namely vertagen-sich vertagen, auflösen-sich auflösen and entspannen-sich entspannen. Haspelmath

1993 speaks in this respect of a so-called ‘anticausative alternation’, i.e. a verb pair in which “the

causative verb is basic and the inchoative verb is derived” (Haspelmath 1993: 91). Comrie 2006

makes the same observation, stating that Indo-European languages make abundant use of reflexives

to derive intransitives from transitives (Comrie 2006: 313). According to Haspelmath & Müller-

Bardey, the valence patterns of causative base and inchoative derivative can be represented as

follows:

Transitive valence pattern

Referents A B

Roles Agent Patient

Functions Subject Object

Derived valence pattern

Referents A

Roles Agent Patient

Functions Subject

Figure 3: Valence pattern alternations for reflexive verbs in a causative-inchoative alternation

(source: Xrakovskij 1981 as cited in Haspelmath & Müller-Bardey 1991: 7)

Traditionally, the class of German reflexive verbs is broken down into two main groups, notably that

of the ‘pure reflexives’ (‘echt reflexive Verben’) and that of the so-called ‘apparent reflexives’ or

‘reflexively used verbs’ (‘unecht reflexive Verben’). According to Hentschel & Weydt, “echt reflexive

sind solche, bei denen immer ein Reflexivpronomen stehen muss, das durch kein anderes Objekt

ersetzt werden kann“ (Hentschel & Weydt 2003: 67). They encompass verbs like sich schämen, sich

beeilen and sich sorgen, which can only be used in contexts in which they refer back to the

grammatical subject. Consequently, expressions like *Ich schäme euch, *Er beeilt Sie and *Wir sorgen

49

dich are grammatically unacceptable. On the other hand, apparent reflexives can also be used non-

reflexively. A typical example is the German verb sich waschen ‘to wash oneself’, since it is also

possible to wash someone else. It is evident that inchoatively used reflexives belong to the latter

class, as they were all once derived from a transitive base through the addition of the reflexive

morpheme.

Haspelmath argues that ‘anticausative formation’ (i.e. the formation of what we have termed

inchoative verbs) through reflexivisation is a productive process in German, but it seems to be highly

dependent of the morphological composition of the base. With non-derived verbs, the formation of

new inchoatives is “relatively restricted”. Examples include. sich spalten ‘split’, sich teilen ‘divide’ or

sich bilden ‘form’ (Haspelmath 1987: 18). However, inchoative verbs hinging on so-called ‘factitive

derivatives’18 are much more numerous. Examples of such verbs are sich verdunkeln ‘become dark’,

sich erhöhen ‘become higher’ and sich aufklären ‘become clear’ (Haspelmath 1987: 18). Despite the

fact that all but one reflexive inchoatives in my sample were derived from a compound base, none of

them seemed to belong to the class of the so-called factitives.

With respect to their status, Reinhart & Siloni 2004 argue that reflexive pronouns have often been

misinterpreted as unaccusative (e.g. by Bouchard 1984, Grimshaw 1990, Kayne 1988 or Pesetsky

1995). This means that it has been wrongly assumed that “the subject of reflexives is an underlying

object, which has to raise to subject position for Case reasons” (Reinhart & Siloni 1999: 7). Two of the

tests the proponents of the unaccusative analysis regularly appeal to are the choice for the auxiliary

‘be’ in Romance languages and the fact that in certain languages, unaccusatives and reflexives are

marked in the same way (Reinhart & Siloni 1999: 9). The last argument can easily be discarded, since

a correspondence in form does not necessarily entail a correspondence in meaning. With the respect

to auxiliary selection, Reinhart and Siloni argue that that is an “intricate matter” and, hence, not very

reliable as a diagnostic (Reinhart & Siloni 1999: 8). Another argument in favour of Reinhart and

Siloni’s case (but one which they strangely enough seem to overlook) is the observation that German

reflexives always select ‘haben’ in the present perfect, a fact which severely undermines the

crosslinguistic value of the ‘auxiliary selection argument’. An example of a causative-inchoative

translation strategy based on reflexivisation can be found below:

18

Factitives are causatives of the type “X causes Y to Z”, with Z being an adjective (Haspelmath 1987: 33).

50

(81) English transitive use: This was, of course, also done to ease the economic situation and

to support reconstruction in Bosnia and was obviously conditional upon the economy and

the institutions of democracy continuing to be reformed. <Opus>

(82) German transitive use: Dies sollte natürlich auch die wirtschaftliche

Lage entspannen und den Wiederaufbau in Bosnien unterstützen. <Opus>

(83) English intransitive use: We hope that the situation in Albania will then soon ease.

<Opus>

(84) German intransitive use: Dann wird sich die Lage in Albanien hoffentlich bald

entspannen. <Opus>

2.2.3 Verbs with preverbal elements

Containing up to 56% of all German translations, the class of verbs with preverbal elements

constituted by far the most capacious morphological category encountered. This result seems to

confirm Fleischer & Barz’s claim that “prefixation [in the broad sense of the word, referring both to

the attachment of prefixes and particles] is in verbal derivational morphology the most commonly

used word formation process” (Fleischer & Barz 2007: 316). Furthermore, all but one labile verbs in

my sample could be found in this category, suggesting a strong link between preverbal elements and

changes in valence patterns. Indeed, according to Kolehmainen, this seems to be exactly the case:

Dass Valenzänderungen bei den deutschen Präfix- und Partikelverben frequent auftreten, ist eine

altbekannte Tatsache, auf die in den meisten deutschen einschlägigen Darstellungen hingewiesen wird

[…]. (Kolehmainen 2006: 275)

However, despite this common observation, no general account of preverbs and valence patterns

has yet been drawn up (Kolehmainen 2006: 275). Within the field of German linguistics there has

been much debate about the criteria the verbs in this group should satisfy, but their status is still

subject to much debate (Kolehmainen 2006: 25). However, it has been argued repeatedly that verbs

containing preverbal elements can best be divided into two subclasses, viz. prefix verbs

(‘Präfixverben’) and particle verbs (‘Partikelverben’). As Kolehmainen provides a relatively recent and

thorough treatment of the items in question, the following discussion will mainly be based on her

observations.

51

2.2.3.1 Prefix verbs (‘Präfixverben’)

The most fundamental feature of German prefix verbs is that their preverbal element has to remain

bound to the stem at all times. Additional features include the fact that prefixes can never bear

primary stress and that they belong to a closed class of lexical items, meaning that the processes by

which these elements were once generated, are no longer productive (Kolehmainen 2006: 26, 34).

Although the items in this class mostly concern bound morphemes like be-, ver- or ent-, certain

prepositions (e.g. durch-, über- or unter-, as in durchdenken, überqueren or unterbrechen) and

adverbs (e.g. wieder-, as in wiederholen) can also function as prefixes (Kolehmainen 2006: 26-27).

However, Kolehmainen notes that these prepositional and adverbial prefixes are not entirely

unproblematic, as they can also be used as free morphemes or verb particles (elements which can be

severed from the stem in certain contructions and therefore have a higher degree of independence)

(Kolehmainen 2006: 28). About one third of the verbs in my sample containing preverbal elements –

17 out of 58 to be exact – belonged to this subgroup. An overview of the respective prefixes and the

verbs with which they built a meaningful element is provided below.

be- beenden, beginnen

ent- entspannen; sich entspannen

er- erledigen, erleichtern, eröffnen

unter- unterbinden, unterbrechen

ver- verhindern, verringern, verschieben, vertagen, verzögern; sich vertagen,

verstummen

zer- zerstören

2.2.3.1.1 ver-

Transitive ver-verbs Intransitive ver-verbs

verhindern X

verringern X

verschieben X

vertagen sich vertagen

verzögern X

verstummen

Table 14: Overview of acquired ver-verbs according to transitivity along with their respective

counterparts (if available)

52

New High German ver- is a prefix with a long history. According to Leopold, it developed throughthe

merger of the Indo-European forms *pr, *pŕri, *per and *prŏ. Already in Gothic, the oldest attested

Germanic language, this proto-set was reduced to three members: *pŕri and *pr converged into faur-

19, whereas *per and *prŏ developed into fair- and fra- respectively20 (Leopold 1907: 4, 5). To make

things even more complicated, Gothic also had a prefix faura-, the ancestor of modern vor-, which

alternated with the aforementioned faur- form. Going back to two different etyma, modern ver- is

thus by no means to be regarded as a reduced form of vor- and, Leopold adds, neither should faur-

be seen as the apocopated form of faura-. With respect to their semantics, he makes the following

observation:

got. faur I enthält wie lat. porricio, porrigo, portendo, polliceor einen Richtungshinweis. faur II und

faura vertreten die Anschauung der Ruhe, aber verschiedener Art: faura entspricht unserem „vor“ in

mehr sinnlicher Anschauung (auch etymologisch), faur II enthält das abstraktere „für“ (υπέρ, pro).

(Leopold 1907: 9)

However, the direct ancestors of modern ver- need to be traced back to Old High German21, which

had six different preverbal forms to express this prefix, notably far-, fer-, fir(i)-, for- and fur-, as well

as the particle furi-22. The affinity with the aforementioned Gothic forms Leopold reconstructs as

follows (Leopold 1907: 24):

Gothic faur- I OHG fur-, for-

faur- II furi-

fair- fir(i)-

fra- far-, fer-

However, these correspondences are only formal in nature; semantically, the connotational

differences between these particles had largely been levelled out in OHG, making the choice for

either of these elements semantically completely ungrounded (Leopold 1907: 24-25). In my sample,

ver- occurred in approximately 41% of all prefix verbs, thereby greatly overruling the frequency of all

19

In order to simplify the etymological discussion of the elements in question, Leopold makes a distinction between faur- I and faur- II, the former being a reflex of IE *pr, the latter of IE *pŕri (Leopold 1907: 9). 20

The development of the labial plosive /p/ into the labiodental fricative /f/ in the syllable onset is due to the First Germanic Sound Shift. 21

Henceforth: OHG. 22

Remember from the preceding discussion that, contrary to prefixes, particles are elements which can be severed from the stem under certain conditions.

53

other prefixes. The abundance of ver-verbs is also noted by Fleischer & Barz, who establish a direct

link with the etymological convergence of which this prefix is the result:

Aus [der] Etymologie ergeben sich für die Gegenwartssprache eine extreme semantische Vielfalt der

ver-Verben und auch ihre zahlenmässige Dominanz im Vergleich zu anderen Präfixverben. (Fleischer

& Barz 2007: 325)

Apparently, this element shows a strong tendency towards transitivity, as it occurred as a part of an

intransitive verb only twice23: Fleischer & Barz argue that with verbs denoting a resultative aspect, as

in verschieben or verzögern, transitivisation as a rule accompanies ver-prefixation (Fleischer & Barz

2007: 326), but it is unclear to what extent ver- affects verbal valence patterns more generally.

2.2.3.1.2 be-

Transitive be-verbs Intransitive be-verbs

beenden X

beginnen beginnen

Table 15: Overview of acquired be-verbs according to transitivity along with their respective

counterparts (if available)

According to Günther, be-verbs can syntactically be divided into six main categories. Apart from

these, he also distinguishes a seventh category of idiosyncratic forms, as well as a group of highly

lexicalised items, meaning the verbs in question cannot be interpreted anymore as the sum of their

parts (e.g. besitzen ‘to own’). Each of these categories was subsequently divided into a number of

subgroups. Beenden, the only non-lexicalised be-verb in my sample, was included into the seventh

category and constituted a subgroup together with verbs like beschließen, begrenzen and bezahlen.

Two semantic properties shared by these verbs are the fact that their core meanings all boil down to

a ‘restriction’ of a certain kind and, secondly, that their objects are fundamentally barred from

denoting animate referents (Günther 1974: 212). According to Bornschier, the two main functions of

German be-verbs are the transitivisation of intransitives and the derivation of denominal verbs

(Bornschier 1971: 25). In German, be- no longer serves as a productive prefix to form deadjectival

verbs, since “the productive prefix [...] for these predicate conversions [...] are ver- or er-, as in

verkürzen or erleuchten (Los et al. 2012: 181). Los et al. also make mention of its transitivising

23

One of these two was the reflexive sich vertagen. Since this verb is clearly an explicit derivation of German vertagen, one could even dispute the full validity of this token as an instance of an intransitive verb.

54

function and add to this that the object of the clause in question “is invariably fully affected” (Los et

al. 2012: 179). With respect to the semantic effect of be-prefixes, Fleischer and Barz argue that their

presence entails at best the addition of a new connotation to the verbal base, but this is by no means

a constant factor. Thus, be- causes in the first place a change to the quantitative and qualitative

valence of a specific verb, viz. it changes both the number of participants a verb can take, as well as

their function. In the case of beenden, the process of prefixation results in an intransitive base

(namely enden) gaining a direct object governing the accusative case. Other features of be-verbs

include the fact that always display so-called ‘external parasynthesis’ and that be-, contrary to other

preverbal prefixes, never entails the modification of the object it refers to (Bornschier 1971: 25).

Remember that the verb pair enden-beenden is an example of what Haspelmath terms a causative

alternation, i.e. an alternation in which “the inchoative verb is basic and the causative verb is

derived” (Haspelmath 1993: 91). An example of a causative-inchoative translation strategy hinging on

the presence or absence of the prefix be- can be found below:

(85) English transitive use: The present coalition has a job to finish, and the involvement of

the Hungarian minority remains every bit as important in this connection. <Opus>

(84) German transitive use: Die derzeitige Koalition hat eine Arbeit zu beenden, und die

Beteiligung der ungarischen Minderheit daran ist nach wie vor unvermindert wichtig. <Opus>

(86) English intransitive use: I believe I can speak on behalf of all participants in this lengthy

meeting, which did not finish until the small hours, when I say that it was not only a success

from a budgetary point of view, because the general budget for 2002 will now soon be

established, but also politically speaking, for we managed to agree on a number of new

elements which presented themselves to us since our first reading and which need to be

discussed, as well as a number of priorities. <Opus>

(87) German intransitive use: Ich darf, so meine ich, im Namen aller Teilnehmer dieser langen

Sitzung, die erst in den frühen Morgenstunden endete, sprechen, wenn ich sage, dass sie

nicht allein im Hinblick auf den Haushalt ein Erfolg war, weil der Gesamthaushaltsplan für

2002 nun zügig festgesetzt wird, sondern auch in politischer Hinsicht, denn wir haben uns

über die Behandlung ein paar neuer Elemente, die sich seit unserer ersten Lesung aufgetan

haben, sowie über einige Prioritäten verständigen können. <Opus>

55

2.2.3.1.3 er-

Transitive er-verbs Intransitive er-verbs

eröffnen X

erleichtern X

erledigen X

Table 16: Overview of acquired er-verbs according to transitivity

Er-verbs showed up thrice in my sample, thus accounting for approximately 15% of all prefix verbs.

All er-verbs were restricted to transitive contexts of use. According to Fleischer & Barz, the prefix er-

is mainly used to derive verbs from adjectival or verbal base forms24. Verbal base forms are, as a rule,

simplex verbs (Fleischer & Barz 2007: 323). These statements seem to be confirmed by my sample:

eröffnen is clearly derived from the verb öffnen (which, in turn, goes back to the adjective offen),

whereas erleichtern and erledigen are derived from the adjectives leicht and ledig respectively. Er-

also frequently shows up as the second element in double prefixations (e.g. in anerkennen,

auerstehen or umerziehen), but this does not hold for the three verbs discussed here. Bornschier

argues that transitivation only occurs as a side effect of the derivation of resultatives, which have to

be verbal in nature (Bornschier 1971: 29). Eröffnen, the only deverbal item in my sample of er-verbs,

does not seem to falsify this claim. Kolehmainen argues that the valence pattern of er-verbs allows

for some interesting observations. First of all, this prefix often adds to the base a new semantic

component which sometimes results in a change in its valence pattern. An intransitive base

sometimes yields a transitive derivative, a transitive base sometimes allows for ‘object interchange’,

meaning that it can take on a new object “with new semantic properties” (Kolehmainen 2006: 284).

The fact that er- only showed up in transitive verbs in my sample therefore seems to be purely

coincidental.

2.2.3.2 Particle verbs (‘Partikelverben’)

Whereas prefixes need to remain bound to the stem at all times, particles show exactly the opposite

feature, for – under certain conditions – they can be severed from the verbal base and move to

sentence-final position, a process commonly known as tmesis, e.g. Er fing mit der Sitzung an

(Kolehmainen 2006: 31). According to some definitions, combinations of (usually two) infinitives (e.g.

kennen lernen, stehen bleiben) as well as so-called Function Verb Phrases (see infra; e.g. in Angriff

nehmen, zum Stillstand bringen) can also be considered particle verbs, but I will adhere to the more

24

Only one er-verb is denominal in nature, notably ermannen ‘take courage’ (Fleischer & Barz 2007: 323).

56

generally accepted view that conceives of particles as indeclinable items, thereby only excepting

reflexive verbs, since I will discuss these under an independent heading (Kolehmainen 2006: 33-34).

In principle it is also possible to create ever new particles, since they belong to an open

morphological class. With a total of 40 tokens out of 58, particle verbs outranged prefix verbs in my

sample roughly by 2:1.

ab- abstellen, abbrechen, abbestellen, abhalten, abschließen;

abbrechen, abfahren, ablegen, ablaufen

an- anfangen, anhalten, antreten, anlassen; anfangen, anhalten, anlaufen,

andauern, anspringen, anfahren

auf- aufhalten, aufhören mit, auflösen; sich auflösen, wieder aufnehmen25

aus- auslösen

ein- einführen, einberufen, einstellen; eingestellt werden

fort- fortsetzen, fortfahren mit, fortführen; fortgesetzt werden

hin- hinhalten

hinaus- hinauszögern

nach- nachlassen

weiter- weitermachen, weitergehen

zusammen- zusammenrufen, zusammensetzen, zusammentragen, zusammenstellen;

zusammenkommen, zusammentreten

With 22 tokens, transitive particle verbs were slightly more numerous than the 15 intransitive

constructions, but these findings are evidently not very reliable, given the restricted amount of

tokens in general. In what follows, I will take a closer look at three of the more important particles

and the verbs they modify.

2.2.3.2.1 ab-

Transitive ab-verbs Intransitive ab-verbs

abstellen X

abbrechen abbrechen

25

As Luc De Grauwe (personal communication) has pointed out to me, wiederaufnehmen is a special instance of a particle verb, since it consists of the verb aufnehmen together with the element wieder, a so-called ‘Verbzusatz’, which has developed out of an original adverb. The original adverbial nature of this element is still illustrated by the fact that it is severed from the particle in constructions involving tmesis, e.g. Ich nehme mein Deutschstudium wieder // auf.

57

abbestellen X

abhalten X

abschließen abfahren

X ablegen

X ablaufen

Table 17: Overview of acquired ab-verbs according to transitivity along with their respective

counterparts (if available)

In my sample, ab-verbs showed up nine times, a figure which is good for just about 20% of all particle

verbs. According to Fleischer & Barz, the main function of the particle ab- is the “semantic

modification of verbs” as well as the creation of new deadjectival and denominal verbs (Fleischer &

Barz 2007: 328). Their base can be either derived (as in abbestellen, which is a derivative of the be-

prefixed verb bestellen) or underived (as in the rest of my tokens). Transitivisation is not a main

function of ab-verbs, but it sometimes occurs as a side-effect, e.g. when ab- adds to the base a

connotation of ‘distancing’ (Fleischer & Barz 2007: 328). In my sample, this is the case for

abschließen.

2.2.3.2.2 an-

Transitive an-verbs Intransitive an-verbs

anfangen anfangen

anhalten anhalten

antreten X

anlassen X

X anlaufen

X andauern

X anspringen

X anfahren

Table 18: Overview of acquired an-verbs according to transitivity along with their respective

counterparts (if available)

The particle most often encountered is an-, since it modifies ten out of forty-four particle verbs, or

23% of all the tokens in this class. Here, too, transitivisation occurs as a side-effect of semantic

modification, but only when the derivative refers to “seeing or speaking”, a meaning component

which is not expressed in any of transitive tokens I inventorised. An interesting observation is that

58

two out of a total of eight ergative verbs (notably anfangen and anhalten) are members of this class.

It is unclear to what extent both phenomena might be related, but further research is definitely in

order to clear this issue up.

2.2.3.2.3 auf-

Transitive auf-verbs Intransitive auf-verbs

aufhalten X

aufhören mit X

auflösen sich auflösen

X wieder aufnehmen

Table 19: Overview of acquired auf-verbs according to transitivity along with their respective

counterparts (if available)

In terms of syntactic behaviour, auf- is quite an obscure particle, since none of the consulted sources

mentioned anything about changes in valence pattern this element might cause. Its presence mainly

serves semantic purposes, which are of no interest to the present study. Hence, it will not be

discussed any further here.

2.2.4 Passives

Transitive Intransitive

fortsetzen fortgesetzt werden

einstellen eingestellt werden

Table 20: Overview of acquired passives along with their transitive counterparts

Representing only 1,7% of all translation strategies (or two tokens in absolute figures), passives were

an extremely rare phenomenon in my sample. Furthermore, it remains unclear why precisely these

two passives were supplied as intransitive entries. This low outcome is contrary to what might have

been expected, since passives, just like inchoative constructions, are both so-called agent-removing

categories, meaning that “the agent argument [is removed] from subject position, [so that] the

patient argument must take up the subject position instead” (Haspelmath & Müller-Bardey 1991: 4).

However, whereas the inchoative is often considered to present an action as coming about

spontaneously, the passive still implies the presence of a specific actor (Haspelmath & Müller-Bardey

1991: 5). Consider the following examples:

59

(89) The door was opened. (Haspelmath & Bardey-Müller 1991: 5, in-text ex.)

(90) The door opened. (Ibid., in-text ex.)

The logical question example (87) puts forward is evidently Who or what caused the door to open?,

whereas the action in (88) is merely described as happening of its own accord (Haspelmath & Müller-

Bardey 1991: 5; Haspelmath 1993: 90). Haspelmath notes that “that does not mean that there

cannot be an agent in the objective situation” (Haspelmath 1993: 90). Moreover, it is well possible

that the agent was in both cases the same entity, but with respect to the linguistic conceptualisation,

passive and inchoative are crucially different (Haspelmath 1993: 90). Remember that Davidse 1991

does not share this conclusion. She claims that the inchoative ‘syncretises’ the voice distinctions ‘self

instigated’ and ‘externally instigated’ and is consequently characterised by an essential vagueness

which can only be resolved by providing sufficient context (Davidse 1992: 109; see also Section 1.2).

Another difference between passive and inchoative constructions relates to their conceptualisation

of what Kalluli terms ‘eventiveness’. Consider the following examples:

(91) The window broke when we got home. (Kalluli 2006: 209, ex. (i))

(92) The window was broken when we got home. (Ibid., ex. (ii))

The fact that (X) hinges on an inchoative and (Y) on a passive seems to have far-reaching

consequences for their interpretation, since the former conceptualises the two events (i.e. the

‘window-breaking’ and the ‘getting home’) as happening simultaneously, whereas the latter conveys

an essentially asynchronous meaning, namely that “the window-breaking event precede[d] the

getting home event” (Kalluli 2006: 209).

However, Kalluli (2006, 2007) argues that passives and inchoatives are more alike than has thus far

been assumed, since they “only differ with respect to the respective building blocks that enter

syntactic computation, but both arise through the same operation, namely the suppression of a

feature in v”, which is caused by the fact that English displays non-active morphology (Kalluli 2006:

442). For inchoative constructions, this feature is the ontological primitive [+cause], for passives

[+act]. This results in the following syntactic structures:

60

Figure 4: The structure of causative predications (source: Kalluli 2006: 454)

Figure 5: The structure of passive predications (source: Kalluli 2006: 454)

An example of a passive rendering of the causative-inchoative verb alternation in the transcriptions

of the European Union, can be found below:

(93) English transitive use: The part of this agreement which blocks negotiations could mean

that the Union ends up paying producers to stop producing sherry, port and a whole host of

products which are purely and simply copies under international commercial law. <Opus>

(94) German transitive use: Der Teil dieses Abkommens, der die Verhandlungen blockiert,

könnte für die Union die Konsequenz haben, daß sie Erzeuger entschädigen muß, damit sie

die Herstellung von Sherry und Portwein einstellen. <Opus>

(95) English intransitive use: Soon the bombing must stop and the building must begin.

<Opus>

61

(96) German intransitive use: Die Bombardierungen müssen bald eingestellt werden und der

Aufbau muss beginnen. <Opus>

2.2.5 Function Verb Phrases (‘Funktionsverbgefüge’)

Transitive FVPs Intransitive FVPs

ein Ende machen X

ein Ende setzen X

in Angriff nehmen X

in Gang setzen X

in Umlauf setzen in Umlauf kommen

zum Erliegen bringen X

zum Stehen bringen X

zum Stillstand bringen zum Stillstand kommen

X in die Brüche gehen

X zu Ende sein

X zum Ende kommen

Table 21: Overview of acquired FVPs according to transitivity along with their respective counterparts

(if available)

The debate surrounding German Function Verb Phrases (henceforth FVPs) is at least as complicated

as the discussion concerning preverbal elements, since there has been an “almost complete lack of

agreement” on their status. However, Van Pottelberge notes that despite these differences, nearly all

sources agree on the fact that an FVP’s most distinctive feature probably resides in its morphological

composition (Van Pottelberge 2001: 15):

Die unabdingbare Voraussetzung ist, daß [ein Funktionsverbgefüge] aus der Verbindung eines Verbs

mit einem Substantiv besteht, meist in Form eines Akkusativobjekts oder eines Präpositionalobjektes.

Wenn keine Verbindung eines Substantivs mit einem Verb vorliegt, kommt eine Einstufung als

Funktionsverbgefüge offenbar nicht in Frage. (Van Pottelberge 2001: 15)

Van Pottelberge even ascribes this criterion axiomatic status, since no linguist has ever “confirmed

nor questioned” it (Van Pottelberge 2001: 15). A similar observation is made by Von Pollenz (Von

Pollenz 1963 in Persson 1975: 1), but he breaks down Van Pottelberge’s unified Noun + Verb-type

into three subgroups:

62

1. Preposition (+ article) + noun + transitive verb

E.g. in Empfang nehmen, zur Folge haben

2. Preposition (+ article) + noun + intransitive verb

E.g. in Betracht kommen, zur Debatte stehen

3. (Article) + accusative noun + transitive verb

E.g. Anwendung finden, einen Anfang machen

With respect to their meaning, Fischer argues that an FVP’s semantic properties essentially reside in

the noun, as the function verb merely “serves as an auxiliary” (Fischer 1978: Abstract). The table

above provides an overview of all FVPs encountered in my sample. Containing up to 13 tokens, this

category represented a good 12% of all translation strategies. The overview I supplied above allows

for a number of interesting observations. First of all, my sample displays a slight bias towards

transitivity, with eight transitive as opposed to five intransitive FVPs. Furthermore, four FVPs occur in

an causative-inchoative alternation, notably in Umlauf setzen – in Umlauf kommen and zum Stillstand

bringen – zum Stillstand kommen. Strangely enough, the possibility for an FVP to appear in such an

alternation did not seem to be determined by its function verb, since both bringen and setzen also

occurred in FVPs without an intransitive counterpart. However, Persson points out that the

correlation between both types is implicational rather than bidirectional:

Während ein FVG mit kommen ein FVG mit bringen voraussetzt, gilt nicht die umgekehrte Regel, daß

als Entsprechung zu jedem FVG mit bringen ein FVG mit kommen gebildet wird. (Persson 1975: 77)

It is true that this rule remains silent on the position of the function verb setzen, but Persson argues

that bringen and setzen behave as so-called ‘competing forms’ (Konkurrenzformen) in FVPs, meaning

that the causative aspect is inherent to both of them (Persson 1975: 78). In other words: what holds

bringen, also holds for setzen, so Persson’s rule is still valid. A second problem the aforecited

generalisation seems to cause, is the fact that my sample did yield the FVP zum Ende kommen, but

not the corresponding intransitive form with bringen. However, as was confirmed by a look in the

online Duden dictionary, the fact that a translation does not occur in my sample, does not mean that

it does not exist, since the transitive counterpart of zum Ende kommen is zum Ende bringen. The

reason for this lexicographical hiatus in Collins Cobuild is unclear to me.

63

Kaewwipat argues that depending on the number of participants they take, FVPs can clearly be

divided into two groups, namely those hinging on the function verbs kommen, gehen, geraten,

gelangen, sein, stehen and bleiben, which are always intransitive, and those based on the verbs

bringen, stellen, setzen, nehmen and ziehen, which mark transitive FVPs (Kaewwipat 2007: 177).

Despite the fact that this is a very clear-cut division, Starke (1975) states that an FVP’s syntactic (or

semantic) properties can never be deduced from the properties of the respective function verb on

which they are based:

Die Fügungspotenzen der Funktionsverbfügungen weichen in vielen Fällen von der Valenz der darin

enthaltenen „Funktionsverben“ entweder qualitativ, d.h. hinsichtlich der Anzahl der Leerstellen, der

potentiellen Aktanten (Mitspieler), oder qualitativ, d.h. in Bezug auf die grammatische Form und/oder

die semantischen Merkmale der möglichen Aktanten, deutlich ab. (Starke 1975: 159-160)

Furthermore, FVPs also display syntactic deviations from their ‘base verbs’26 (Fischer 1978: 59).

Sometimes both forms correspond (as with zum Abschluss bringen and abschließen, which are both

transitive), but this is definitely not always the case (gehen, for instance, is intransitive, whereas in

Gang setzen clearly takes two participants) (Fischer 1978: 69; Van Pottelberge 2001: 46). With

respect to the number of participants in kommen-FVPs, Fischer observes that they are always

characterised by a reduction in valence compared to their base verb:

(97) a. Die Regierung wendet das Gesetz an. (Fischer 1978: 67, unnumbered ex.)

b. Das Gesetz kommt (durch die Regierung) zur Anwendung. (Ibid., unnumbered

ex.)

This leads Fischer to the conclusion that kommen-FVPs are in fact special instances of passives, since

both of these construction types “display the same semantic structure”, which is caused by a change

in their syntactic framework (Fischer 1978: 68). I believe that the observation Fischer makes is fully

correct, but I strongly disagree with her conclusion to consequently characterise kommen-FVPs as

passives. In my opinion, the alternation exemplified in (95) is in fact causative-inchoative, with the

FVP as inchoative member of the pair. Remember that one of the differences between passives and

inchoatives is the fact that the latter realise their external argument by means of a durch-phrase in

German (i.e. the equivalent of the English from-phrase), a diagnostic which Fischer unknowingly

26

An FVP’s base is a verb with which it can be used interchangeably and from which an FVP’s noun is derived. For example, the base of zu Fall bringen is fallen, that of zur Erkenntnis bringen is erkennen, etc.

64

applied in her example. Passives, on the other hand, denote the cause of an event by using a von-

phrase, a complement which is clearly ungrammatical in this case:

(98) *Das Gesetz kommt von der Regierung zur Anwendung.

Furthermore, as illustrated by the literary example below, the FVP in question also passed the von

selbst-test:

(99) Aus gewonnenen Einsichten werden Einstellungen herausgewachsen, die sich in

Gesinnungen kristallisieren und dann in den alltäglichen Handlungsabläufen wie von selbst

zur Anwendung kommen (taken from: Martin Sieg, Gottes Spuren in unserer Welt, Berlin: LIT

2006, p. 73).

Thus, we can tentatively conclude that kommen-FVPs not only form causative-inchoative pairs with

the corresponding bringen-FVPs, but also with their base forms (with the express condition that they

be transitive).

Examples of the causative-inchoative FVPs in my sample were only found with slightly modified

English source wordse. This means that the FVP zum Stillstand kommen, whose use is exemplified

here, was originally acquired from Collins Cobuild as the translation of the English verb ‘stop’, yet

Opus could not generate any contexts of use in which this translation was literally repeated.

However, it did provide zum Stillstand bringen as a translation of ‘bring to a halt’ and ‘stagger’:

(100) English transitive use: It impedes trade; in fact it could even be said that it will bring it

to a halt, because it is only a question of time before the art market shifts away from Europe

in favour of the USA or Switzerland. <Opus>

(101) German transitive use: Sie erschwert den Handel, ja man könnte sogar sagen, sie bringt

ihn zum Stillstand, denn es ist nur eine Frage der Zeit, bis sich der Kunstmarkt aus Europa

weg in die USA oder die Schweiz verlagert. <Opus>

(102) English intransitive use: Frankly the Convention is staggering and may stall completely

should the Feira Council fail to bring a fresh stimulus. <Opus>

65

(103) German intransitive use: Offen gesagt, der Konvent kommt im Moment nicht so recht

voran und könnte ganz zum Stillstand kommen, wenn vom Europäischen Rat in Feira keine

neuen Impulse ausgehen. <Opus>

2.2.6 Verbs with a prepositional complement

Transitive tokens Intransitive tokens

anfangen mit anfangen

fortfahren mit X

aufhören mit aufhören

Table 22: Overview of acquired prepositional verbs along with their intransitive counterparts (if

available)

Prepositional verbs occurred only thrice, making it a very rare translation strategy. This low number

is in fact quite surprising, since prepositional verbs constitute an excellent way of increasing the

number of participants a verb takes and consequently allow for the creation of transitive

counterparts to intransitives. An example of this translation strategy is provided below:

(104) English transitive use: We must stop the navel-gazing by EU representatives in

international forums and, for the first time, genuinely link multilateral and bilateral

negotiations in the EU so that abusive countries know they will face adverse consequences in

the EU’s diplomatic trade and development policies. <Opus>

(105) German transitive use: Wir müssen damit aufhören, dass sich die EU-Vertreter in

internationalen Gremien nur auf die eigene Nabelschau konzentrieren, und wir müssen

damit beginnen, unsere multilateralen und bilateralen Verhandlungen in der EU wirksam

abzustimmen, damit die Länder, die Menschenrechtsverletzungen begehen, wissen, dass sie

die negativen Konsequenzen eines solchen Verhaltens in den diplomatischen Beziehungen

mit der EU, aber auch in den handels- und entwicklungspolitischen Maßnahmen zu spüren

bekommen werden. <Opus>

(106) English intransitive use: But whatever happens, the human rights abuses must stop.

<Opus>

66

(107) German intransitive use: Doch was auch geschieht, die Mißachtung der

Menschenrechte muß aufhören. <Opus>

2.2.7 Syntactic causatives

Transitive tokens Intransitive tokens

haltmachen lassen haltmachen

anhalten lassen anhalten

Table 23: Acquired syntactic causatives along with their intransitive counterparts

Two instances in my sample expressed the causative variant in the ergative alternation by means of a

lassen-periphrasis, namely haltmachen lassen and anhalten lassen. According to Gunkel’s approach

(see Section 1.1.2), both can be considered plain causatives (as opposed to causative passives), since

the verbs from which they are derived, viz. haltmachen and anhalten, are both intransitive and are

consequently barred from participating in passive constructions:

(108) a. Der Kommandant lässt die Soldaten haltmachen/anhalten.

b. *Von den Soldaten wurde haltgemacht/angehalten (based on Gunkel 1999: 65,

ex. 3b.).

c. *Der Kommandant lässt (von den Soldaten) haltmachen/anhalten (based on

Gunkel 1999: 65, ex. 3c.).

Opus was not able to generate any hit for the constructions haltmachen lassen or anhalten lassen.

The reason for this anomaly, I am convinced, is due to sheer coincidence.

2.2.8 Remaining verbs

Transitive verbs Intransitive verbs

gründen

hemmen

kündigen

lindern

lösen

sperren

67

sprengen

stillen

stoppen

aus sein

fertig sein

nicht mehr laufen

stehen bleiben

vom Neuem [sic] beginnen

warten

Zeit schinden

Table 24: Remaining verbs according to transitivity

The group of remaining verbs comprised sixteen tokens in total, a number which is good for about

14% of all translation strategies and therefore quite high. An interesting observation is that all but

one intransitive verbs are in fact multi-word verbs or periphrastic constructions. An interesting token

from the transitive column is sprengen, since it is an example of a causative relic form which was

once derived from its inchoative base, springen, by means of (palatal) vowel mutation or umlaut

(Garcia Garcia 2012: 126; Van Gelderen 2011: 115). Strangely enough, Collins Cobuild did not make

mention of this intransitive base, since it does not occur in my sample.

68

Conclusion

In this paper, I took a closer look at the causative-inchoative verb alternation in general and the way

in which German represents this alternation more specifically. The analysis I carried out was based

on a semantic set of English ergative verbs pertaining to ‘starting and stopping’ which I translated

into German with the help of Collins Cobuild. The acquired tokens were subsequently re-translated to

bridge the gaps in the original sample. After having filtered out the data, I grouped them into eight

categories, notably (1) ergative verbs, (2) reflexive verbs, (3) verbs with preverbal elements (which

encompassed both prefix and particle verbs), (4) passives, (5) function verb phrases, (6) verbs with a

prepositional complement, (7) syntactic causatives and (8) remaining verbs. As expected, the group

of German ergative/labile verbs was considerably smaller than in English. This observation is in line

with Van Gelderen 2011, who attributed the unusually high number of English ergatives to an

inclination towards ‘morphological opacity’ this language has shown. Another observation was the

relatively high number of so-called ‘function verb phrases’, but by far the largest category was that of

verbs with preverbal elements. More than half of the acquired translations could be attributed to

either of its subgroups (i.e. either prefix or particle verbs), but strangely enough, these verbs proved

to be extremely understudied. Kolehmainen made the same observation, stating that “a general

study of [their] valence patterns has so far not been carried out” (Kolehmainen 2006: 275). It is

evident that this hiatus in the literature opens up new perspectives for future research. Another

peculiar finding was the extremely low number of passives and syntactic causatives in my sample.

Both construction types only occurred twice, whereas it should be theoretically possible to build

them from any type of verb and consequently create as many alternating valence pairs as verbs in

general. Furthermore, their distribution struck me as quite arbitrary, since it remained unclear why

they occurred with exactly those verbs they occurred with (and not just with two others or perhaps

ten more). The same conclusion held for the prepositional verbs, since they, too, could have been

provided for an array of other verbs. Furthermore, they constitute a perfect way to create a

transitive-intransitive verb pair as well.

It is obvious that caution is in order when interpreting the conclusions drawn from my findings, since

the sample I took as my starting-point, even after extension, only comprised twenty-nine tokens. A

study based on a larger number of verbs would evidently yield more reliable results, something

which, for practical reasons, was not possible here. Apart from that, it would also be interesting to

investigate whether another semantic set of verbs (or perhaps a set of morphologically related verbs)

behaves similarly to the verbs I studied.

With respect to the theoretical framework, I have the clear impression that, within the group of West

Germanic languages, German is profoundly understudied for ergativity. An exploration of the

69

potentially transitivising nature of this language (cf. Van Gelderen 2011) or its distinctly higher

number of morphological causatives (‘relic forms’) are therefore definitely within the bounds of

future research.

70

References

Abraham, Werner. 2001. On identifying unaccusativity. GAGL: Groninger Arbeiten zur

germanistischen Linguistik 44: 213-220. Special issue: Gerart van der Meer and Alice G.B. ter

Meulen (eds.). Making sense: from lexeme to discourse.

Alexiadou, Artemis, Elena Anagnostopoulou & Martin Everaert (eds.). 2004. The unaccusativity

puzzle: explorations of the syntax-lexicon interface. Oxford: Oxford University Press.

Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. 2005. The Properties of Anticausatives

Crosslinguistically. In: Frascarelli, Mara (ed.) Phases of Interpretation. Berlin: Mouton. 187-209.

Aronson, Howard. 1977. English as an Active Language. Lingua 41 (3-4). 201-216.

Baron, Naomi. 1974. The Structure of English Causatives. Lingua 33 (4). 299-342.

Bornschier, Marion. 1971. Die Verbalpräfixe im Französischen und Deutschen. Ein Vergleich der

Systeme. Zürich: Juris Druck & Verlag.

Comrie, Bernard. 2006. Transitivity pairs, markedness and diachronic stability. Linguistics 44 (2). 303-

318.

Daunorienè, Justina. 2007. Bedeutungswandel bei den reflexiven Verben des Deutschen. Kalbotyra

57 (3). 62-70.

Davidse, Kristin. 1992. Transitivity/Ergativity: The Janus-Headed Grammar of Actions and Events. In:

Davies, Martin & Louise Ravelli (eds.) Advances in Systemic Linguistics: Recent Theory and

Practice. London: Pinter. 105-135.

Davidse, Kristin & Sara Geyskens. 1998. Have you Walked the Dog Yet? The Ergative Causativization

of Intransitives. Word 49 (2). 155-180.

Davidse, Kristin. 1998. On Transitivity and Ergativity in English, or on the Need for Dialogue between

Schools. In: van der Auwera, Johan, Frank Durieux & Ludo Lejeune (eds.) English as a Human

Language. To Honour Louis Gossens. München: Lincom Europa. 95-108.

Davidse, Kristin & Liesbet Heyvaert. 2007. On the Middle Voice: An Interpersonal Analysis of the

English Middle. Linguistics 45 (1). 37-83.

Dixon, Robert M. W. 1979. Ergativity. Language 55 (1). 59-138.

71

Dixon, Robert. 1994. Ergativity. Cambridge: Cambridge University Press.

Fischer, Mary Wil. 1978. Deutsche und englische Funktionsverbgefüge. Ein Vergleich. London:

University Microfilms.

Fleischer, Wolfgang & Irmhild Barz. 2007. Wortbildung der deutschen Gegenwartssprache. Tübingen:

Max Niemeyer Verlag.

Francis, Gill, Susan Hunston & Elizabeth Manning. 1996. Chapter 7: Ergative verbs. Collins COBUILD

Grammar Patterns. Vol. 1: Verbs. London: HarperCollins. Available from the Centre for Corpus

Research, University of Birmingham: <https://arts-ccr-002.bham.ac.uk/ccr/patgram/>.

Garcia, Luisa Garcia. 2012. Morphological Causatives in Old English: The Quest for a Vanishing

Formation. Transactions of the Philological Society 110 (1). 122-148.

Gese, Helga, Claudia Maienborn & Britta Stolterfoht. Adjectival Conversion of Unaccusatives in

German. Journal of Germanic Linguistics 23 (2). 101-140.

Gunkel, Lutz. 1999. Causatives in German. Theoretical Linguistics 25 (2-3). 65-79.

Hartmut, Günther. 1974. Das System der Verben mit BE- in der deutschen Sprache der Gegenwart. Ein

Beitrag zur Struktur des Lexikons der deutschen Grammatik. Tübingen: Max Niemeyer Verlag.

Haspelmath, Martin. 1987. Transitivity Alternations of the Anticausative Type. Köln: Institut für

Sprachwissenschaft.

Haspelmath, Martin & Thomas Müller-Bardey. 1991 (slightly revised in 2001). Valence Change. In:

Booij, Geert, Christian Lehmann & Joachim Mugdan (eds.) Morphology. A Handbook on

Inflection and Word Formation. Berlin: de Gruyter. 1-23.

Haspelmath, Martin. 1993. More on the Typology of Inchoative/Causative Verb Alternations. In:

Comrie, Bernard & Maria Polinsky (eds.) Causatives and Transitivity. Amsterdam: Benjamins.

87-120.

Hentschel, Elke & Harald Weydt. 2003. Handbuch der deutschen Grammatik. Berlin: Walter de

Gruyter.

Ide, Manshu. 1994. Lassen und lâzen. Eine diachrone Typologie des kausativen Satzbaus. Würzburg:

Königshausen und Neumann.

72

Kaewwipat, Noraseth. 2007. Kontrastive Lesegrammatik Deutsch-Thai für den Unterricht Deutsch als

Fremdsprache in Thailand. Untersuchungen am Beispiel des Nominalstils. Kassel: University

Press.

Kallulli, Dalina. 2006. A unified analysis of passives and anticausatives. In: Bonami, O. & Cabredo

Hofherr (eds.) Empirical issues in Syntax and Semantics 6. 201-225.

Kallulli, Dalina. 2006. Passive as a feature-suppression operation. In: Abraham, Werner & Larisa Leisiö

(eds.) Passivization and typology: Form and function. (Typological Issues in Language, 68.)

Amsterdam: Benjamins. 442-460.

Kalulli, Dalina. 2007. Rethinking the Passive/Anticausative Distinction. Linguistic Inquiry 38 (4). 770-

780.

Keller, Frank & Antonella Sorace. 2001. Gradient auxiliary selection and impersonal passivization in

German: an experimental investigation. Journal of Linguistics 39 (1). 57-108.

Kemmer, Suzanne. 1993. The Middle Voice. Philadelphia: John Benjamins Publishing Company.

Keyser, Samuel Jay & Thomas Roeper. 1984. On the Middle and Ergative Constructions in English.

Linguistic Inquiry 15 (3). 381-416.

Kolehmainen, Leena. 2006. Präfix- und Partikelverben im deutsch-finnischen Kontrast. Frankfurt am

Main: Peter Lang.

Lass, Roger. 1994. Old English. A Historical Linguistic Companion. Cambridge: Cambridge University

Press.

Lemmens, Maarten. 1998. Lexical Perspectives on Transitivity and Ergativity. Causative Constructions

in English. Amsterdam: John Benjamins Publishing Company.

Leopold, Max. 1907. Die Vorsilbe VER- und ihre Geschichte. Hildesheim: Georg Olms Verlag.

Levin, Beth. 1993. English Verb Classes and Alternations: a Preliminary Investigation. Chicago: The

University of Chicago Press.

Levin, Beth & Malka Rappaport Hovav. 1994. A Preliminary Analysis of Causative Verbs in English.

Lingua 92. 35-77.

73

Levin, Beth & Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics

Interface. Massachusetts: MIT Press.

Montemagni, Simonetta. 1994. Non-Alternating Argument Structures: The Causative/Inchoative

Alternation in Dictionaries. In: Martin, Willy, Willem Meijs, Margreet Moerland, Elsemiek ten

Pas, Piet van Sterkenburg & Piek Vossen (eds.). Proceedings of the sixth Euralex international

congress on lexicography. Amsterdam 30-08/03-09. 349-359.

McGregor, William B.. 2009. Typology of Ergativity. Language and Linguistics Compass 3 (1). 480-508.

Ogawa, Akio. 2002. Kausativ in Variationen. Eine typologisch-vergleichende Studie. Koreanische

Zeitschrift für Germanistik 86. 84-96.

Perlmutter, David. 1978. Impersonal Passives and the Unaccusative Hypothesis. Proceedings of the

4th Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society.

Persson, Ingemar. 1975. Das System der kausativen Funktionsverbgefüge. Eine semantisch-

syntaktische Analyse einiger verwandter Konstruktionen. Kristianstad: Kristianstads

Boktryckeri.

Piñón, Christopher. 2001. Modelling the Causative-Inchoative Alternation. Linguistische

Arbeitsberichte 76. 273-293.

Reinhart, Tanya & Tal Siloni. 2004. Against the unaccusative analysis of reflexives. Alexiadou, Artemis,

Elena Anagnostopoulou & Martin Everaert (eds.) The unaccusativity Puzzle. Studies on the

syntax-lexicon interface. Oxford: Oxford University Press. 288-331.

Shibatani, Masayoshi & Prashant Pardeshi. 2002. The Causative Continuum. In: Masayoshi Shibatani

(ed.) The Grammar of Causation and Interpersonal Manipulation. Amsterdam: Benjamins

Publishing Company. 136-177.

Starke, Günter. 1975. Zum Einfluss von Funktionsverbgefügen auf den Satzbau im Deutschen.

Deutsch als Fremdsprache 12. 157-163.

Toyota, Junichi. 2008. Re-evaluating Comparison between English and German: Indo-European

Perspectives. Kalbotyra: Germanų ir romanų studijos 59 (3). 281-289.

Van Pottelberge, Jeroen. 2001. Verbnominale Konstruktionen, Funktionsverbgefüge. Vom Sinn und

Unsinn eines Untersuchungsgegenstandes. Heidelberg: Universitätsverlag C. Winter.

74

Van Gelderen, Elly. 2011. Valency Changes in the History of English. Journal of Historical Linguistics 1

(1). 106-143.