<<

of Possessive Noun-Phrases in Germanic

Synchronic correlation analysis of linguistic strategies concerning the loss of -marking

Master Thesis Final Version

Student: . Davies-Popken Student number: 10581324 Supervisor: A.P. Versloot Master: Linguistics: and Society (Scandinavian Languages) Date: 17 December 2017 Word : 20275

1 Index

Index 2 Introduction 4 1. Background 6 1.1 Deformed genitives ...... 6 1.2 Dative/ ...... 10 1.3 Zero Marking ...... 11 1.4 Compounds ...... 12 1.5 Preposition constructions ...... 13 1.6 Personal ...... 14 1.7 Subordinate clauses ...... 17 1.8 Summary ...... 17 2. Methodology 19 2.1 Collection of data ...... 19 2.2 The languages ...... 20 2.3 The translated texts ...... 21 2.4 Categorizing the data ...... 22 2.5 Analysis and hypotheses ...... 26 3. Results 28 3.1 General overview ...... 28 3.2 Correlations ...... 30 3.3 Languages with original genitive case ...... 31 3.4 Partitives in translation ...... 34 3.5 Languages with deformed genitives ...... 35 3.5.1 English ...... 35 3.5.2 Dutch ...... 37 3.5.3 ...... 37 3.5.4 Faroese ...... 38 3.5.5 Swedish ...... 38 3.5.6 ...... 38 3.5.7 Danish ...... 39 3.6 Languages without genitive case ...... 40 3.6.1 ...... 40

2 3.6.2 Fering ...... 41 4. Discussion 43 4.1 Discussing results ...... 43 4.1.1 German ...... 43 4.1.2 Icelandic ...... 44 4.1.3 English ...... 45 4.1.4 Dutch ...... 47 4.1.5 Luxembourgish ...... 48 4.1.6 Faroese ...... 49 4.1.7 Swedish & Danish ...... 50 4.1.8 Elfdalian ...... 51 4.1.9 Afrikaans ...... 52 4.1.10 Fering ...... 53 4.2 Translational decisions ...... 54 Conclusion 57 Bibliography 60

3 Introduction

There are several ways to express a possessive relationship between two nouns in various languages. One can for example place a preposition between the two nouns to imply their relationship with each other (The book of John), use a certain suffix to one of the nouns to express their relationship (Johns book), or use an even different strategy. It depends per language which strategies are common in expressing a possession between two nouns. This research will focus on the differentiation in strategies between in how they express a possessive noun phrase. The direct motivation for this research topic is the loss of the morphological case in most Germanic languages. In languages like English, Dutch, Swedish, Afrikaans and Danish, a case system is barely in use (except for some archaic forms), while in languages like Elfdalian and Faroese, the morphological cases are at least changing. In all of these languages the genitive case, used to express a possessive relationship between two nouns, has been either lost, or changed into a very minimalistic form. Only in German and Icelandic, this genitive case has maintained its original form and function. With the loss of the genitive case in these languages, their other strategies to express a possessive noun phrase, must have come more in use, leading to the question which strategy/strategies is/are preferred. And even without the loss of the genitive case, it is questionable whether a case-marking is the most preferred strategy in languages like German and Icelandic. Much extant research has tried to explain this loss of the genitive case by means of a diachronic analysis and of its evolution to a new form. This leaves an unexplored possibility for a synchronic research concerning these alternatively used possessive constructions in Germanic languages. Furthermore, a can be made between the languages in order to find out how these languages manage the loss of the genitive case and if languages share preferences to certain linguistic strategies to express a possessive noun phrase. In order to be able to conduct such a research, a basic text or speech is required which is available in all the language intended for research. For this purpose, Le Petit Prince by Antoine de Saint-Exupéry (1943) is an ideal source as it is translated into many (Germanic) languages and is therefore great material for a case-study. Not only is it possible to analyze this text per language on the expression of possessive noun phrases, but it also gives the possibility to make an equal comparison between the languages. In order to investigate the preferred strategies per language (in comparison to the others), possessive noun phrases where acquired from Le Petit Prince and analyzed on how they were translated into the following languages: English, Dutch, Afrikaans, Luxembourgish, Faroese, Fering, Swedish, Elfdalian, Danish, Icelandic and German (a further explanation of the choice of

4 languages will be given in the methodology). The possessive constructions were ordered, analyzed and compared, leading towards a clearer insight in which possessive constructions were preferred per language and if languages acted alike in their translation strategies. Chapter One presents a background overview on the possible possessive constructions and how they are used in the aforementioned languages. In Chapter Two, the methodology is presented, together with some hypotheses that resulted from the background information. The results of the comparative analysis will be given in Chapter Three. An interpretation of the results, together with the information in Chapter One, leads to Chapter Four: Discussion. At the end of this thesis, a conclusion is formulated and the research questions will be answered, as well as possibilities for further research will be discussed.

5 1. Background

In linguistics, a possessive construction indicates an (asymmetrical) relationship between the subject that is possessing (possessor) and that which is possessed (possessee/object). This does not necessarily indicate a Noun-Noun only relationship because according to Bril (2012), it can be used in a much broader semantic context. Examples of these possessive relationships are kinship relations, part-whole relationships, body parts, attributes and partitive relationships (p. 65). A possessive construction or possessive Noun Phrase (NP), which contains some form of a possessive modifier, can be formed either morphologically or syntactically. All of these forms, as will presently be outlined, are functionally equivalent to genitives (Dahl 2015, p. 157). The genitive case, part of a morphological case-system, is used to express a possessive relationship by adding a suffix to a possessive NP. In Germanic languages, this case is mostly found in Icelandic and German, as the use of the genitive case has decreased in most other Germanic languages. Due to the loss of the genitive case, other possessive modifiers came into use, as has been discussed by Dahl (2015, p. 157-200). Dahl (2015) mainly focused on the Scandinavian language area but also showed a clear overview of the possible constructions used to express a possessive NP. Solutions which subsequently presented themselves are deformed genitives (a morphological solution that originated from the genitive case), dative/accusative case, compounds, zero-markings, preposition constructions, personal constructions and subordinate clauses. In this chapter a closer look will be taken into what extent these constructions are important in Germanic languages. First, an explanation will be given of each of the given possibilities of expressing possessive NPs.

1.1 Deformed genitives

In most Germanic languages a form of the genitive case is preserved, though often in a different or minimalistic form. A form like this can be called a "Deformed Genitive". This kind of genitive is still a morphological modifier but has lost most of its former dimensions. An example of this kind of the genitive is the so-called s-genitive. Instead of a complete genitive case, only an -s is added as a suffix to the possessive NP. This can be illustrated by examples of German (DE) (with a genitive case) and English (GB) (with an s-genitive):

DE: Das Buch des Mannes > Genitive Case GB: The man's book > S-suffix / S-genitive

6 The genitive case in German is expressed by the des and the adding of - to the possessor. Originally, English contained a similar morphological case-system to German, meaning that the addition of different article forms and suffixes to nouns could express a possessive relationship. In Modern English only the s-suffix remained, resulting in the genitive case losing its original form in English, hence a deformed genitive. Dahl (2015) refers to these deformed genitives as s-genitives, which he claims can be found in many Germanic languages: Dutch, West-Frisian, English, Danish, Swedish and to a lesser extent in Norwegian (p. 158). Other examples of these genitives can be found in Elfdalian (a spoken in ) and Faroese. In Elfdalian, a possessive NP can be formed with a plus an es-suffix (instead of no case-marking plus an s-suffix in other Germanic varieties) (p. 173). In Faroese, the deformed genitive can be formed with a sa-suffix (Koptjevskaja-Tamm 2003, p. 627). That the deformed genitives originate from the genitive case, does not mean that they share the same domains. The genitive case as shown in the German example modifies the possessor in a possessive NP, which also happens in the English example. Remarkably, in English, it is possible to not only modify the possessor but a full NP. Therefore the English s-genitive has extended its domain into a 'group genitive'; the suffixed -s can be added to a complete NP/'Noun Group' (Allen 2003, p. 2). A sentence such as: 'The federal government's chief', is, therefore, possible in English. Feist (2012) finds that it depends on the semantics of a possessive NP as to whether an s-genitive or another possessive construction can be used (p. 286-287). As he finds the English s-genitive to be rather independent, it is often the (grammatically) free choice of the speaker/writer which construction he/she wishes to use (p. 287). The s-genitive in Dutch turns out to be much less freely used than the English s-genitive, leaving Weerman & de Wit (1999) to think that they even might be a different construction. To mark this difference, they refer to the Dutch s-genitive as an s-construction. They first of all argue that the s-construction in Dutch has no synchronic relation to the genitive, as found in modern German. The following arguments are given: an s-construction can only follow an N in an NP (a genitive could occur on both sides) and the s-construction cannot be used as a complement of verbs or (as a genitive might be) (p. 1165-1166). What does relate the s-construction to the older genitive case is their origin; the s-construction develops diachronically from the genitive (p. 1167). These arguments explain the differences between the Dutch s-construction and the genitive case but do not make clear what the difference is with the English s-genitive. This can be shown in the following example:

7 NL: *[mannen]s boek *[Oude mannen]s boek GB: Mens book Old mens book

Both of the Dutch examples are ungrammatical because the Dutch s-construction is mainly restricted to proper names or words that we can understand as being independent. The s-genitive in English can be attached to other words, as for example in the plural form of the word 'man': men. The other difference lies in the placing of the attachment. In English, the s-genitive can be attached to the end of the NP, while the Dutch s-construction can only be attached to a single noun (p. 1172- 1173). For this thesis on the possessive NPs in Germanic languages, the distinction between the terms s-construction and s-genitive will not be applied, as both derive from the genitive case. Both will be referred to with the name s-genitive. The research of Weerman & de Wit shows that there are differences in usage of the s-genitive in both English and Dutch. The research of Weerman & de Wit (1999) sheds a light on the resemblances of the English, Dutch and Swedish s-genitives. According to their research, the Swedish s-genitive follows the same diachronic development as the Dutch s-construction/s-genitive, as they both derive from the genitive case (p. 1176). They mention that the English s-genitive may have a different source (maybe the pronoun his) (p. 1175) and therefore has possibly followed a different diachronic path (although this is far from certain (Norde 1997, p. 91). When analyzing these s-genitives synchronically, it becomes clear that the Swedish s-genitive shares the same features as the English s-genitive and less so with the Dutch s-construction/s-genitive (p. 1176). The difference between the English and Swedish s-genitive lies in their grammatical rules. As stated by Feist (2012), the usage of the English s-genitive is relatively free and depends on semantics and the choice of the speaker/writer. In Swedish, alienability is an extra factor that plays a role in the choice of possessive constructions. When the relationship between a possessor and possessee is alienable, the usage of the s-genitive is mandatory (for example: ''s bil'), while in English this can be replaced with for example a preposition construction: 'The car of Anna'. (Dahl 2015, p. 188). According to Dahl (2015), the same rules on the Swedish s-genitive apply to the Danish s- genitive. When a relationship between possessor and possessee is alienable, one must use an s- genitive. According to Perridon (2013), the Danish s-genitive is used more freely in the spoken language than the English genitive. Younger speakers tend to not only use it as a group genitive but use it even after an adverb, or sometimes a verb (p. 142-143). This usage of the Danish s-genitive is, strictly speaking, ungrammatical; however, due to its frequency in (amongst other things) social media, it possibly constitutes a functional usage (p. 143). Both Swedish and Danish, as standard languages, show the s-genitive but it is less known in

8 Norwegian. It is sometimes found in some resilient forms but it is rarely present in the language (Dahl 2015, p. 158). In the non-standard varieties of Scandinavian languages, one can find not only the s-genitive but also alternative endings, such as an a-genitive (Dahl 2015 p. 158-159). One of these varieties is Elfdalian, spoken in Sweden ( region). This dialect is as far removed from Swedish as Swedish is from Icelandic and therefore Dahl argues that it can be considered to be a different language (Dahl 2005, p. 40). Elfdalian preserves a case-system, although the genitive case is mainly lost. Instead of this, one can find either a dative case (paragraph 1.2 Dative/Accusative case) or a complex dative construction. According to Dahl, this complex dative construction is a form of a deformed genitive, formed by adding an es-ending to the dative case:

EL: Ita jar ir kullumes saing > Complex Dative Construction 'This here is the girls' bed' (Dahl 2015, p. 170)

The es-ending added to a dative case, is supposed to be the closest equivalent to the s-genitive in Swedish (Dahl 2015, p. 173). The Dalecarlian dialect area (Dalarna region) is mostly known for using Elfdalian but also smaller of this 'language' are still in use. These dialects can also use other forms of the es-ending, like: -sa, -os, etc. (p. 178-182). As in Elfdalian, the case-system in Faroese seems to be mainly intact (Kaptjevskaja-Tamm 2003, p. 627). In this language, the deformed genitives have a different ending: a sa-suffix. This sa- suffix seems to have quite a few similarities with the English s-genitive, as it is possible to add it to the end of an NP. The difference lies in where it is used, namely: kinship terms, proper nouns and names (Petersen 2016, p. 2):

FA: Mammusa bilur 'Mother's car'

The s-genitive as found in English is not standard in Faroese (Thráinsson, Petersen, Jacobson, Hansen 2004, p. 63). The Frisian language, spoken in both the and , can be divided into West-, East- and North-Frisian. Dyk & Heyen (2017) have written about Frisian and have found that West-Frisian has, as well as the s-genitive, preserved an - and ene-ending. The use of the s-genitive is rather restricted (and therefore different than the English s-genitive), and is mostly found in names and kinship terms. The e-ending is even more widely found, and can only be

9 used if the possessor ends in a stressed or a . The ene-ending is only used in (archaic) written language when it is in an NP without a . Heyen (2017) found that in North- Frisian, specifically Fering (FE) (spoken on the island Föhr), the s- and e-ending can only be found in fixed constructions, and are still rather characteristic in their partitive use. This can be seen in the following example:

FE: Wat ruads/wat ruaden > Partitive genitive 'Something red'

It is of particular note that in Fering both an s- and en-ending can be used, whilst in most other Germanic languages only the s-ending (s-genitive) is possible.

1.2 Dative/Accusative case

With the loss of the genitive case, other case-markings came into use in order to express a possessive relationship. An example of this is the dative case, as can be seen in the comparison between German and Elfdalian (EL):

DE: Den Zeichnungen des Verfassers > Genitive Case EL: Tekkninggär författeram > Dative Case 'Drawings of the author'

Where the German example has a preposition marked with a genitive case, (des and an s-suffix to the possessor), a possessive relationship is expressed only with a dative case marking ( -am on the possessor), in Elfdalian. This is not the same as the complex dative construction, which still uses an es-ending as a form of a deformed genitive. In Faroese (FA), it is possible to express possession with an accusative case, for example when expressing family relationships:

DE: Der Vater des Jung en > Genitive Case FA: Papi dreingin > Accusative Case 'The boy's father'

The German example has again a genitive case-marking, whereas the possessive relationship is

10 expressed by an accusative case-marking to the possessor (-in) in Faroese (Thráinsson, Petersen, Jacobson, Hansen 2004, p. 63).

1.3 Zero Marking

In the Elfdalian example of the dative case, the possessive relationship is implied by using a dative case. According to Dahl (2015), a lack of morphological marking can imply such a possessive relationship:

EL: Kalln Smis-Margit > No Case Marking 'Smis-Margit's Husband' (Dahl (2015), p. 162)

Dahl calls this construction a zero-marking construction, meaning that in the cases that one would generally expect to find a case-marking, instead, the words seem to be added together without forming a compound, though they still imply a possessive relationship with each other (p. 162). The zero-marking construction is mentioned in Sigurðssons research on the central traits of NPs in Icelandic. He does this in relationship to partitives, finding that in so-called pseudo-partitive constructions (see Delsing 1993, p. 185) Swedish (SE) and German can use a zero-marking construction:

DE: Drei Flaschen wein > Zero-marking Zwei Schachteln Zigarretten SE: Tre flaskor vin > Zero-marking Två paket cigaretter 'Three of bottles wine, two packets of cigarettes'

This is not possible in either Icelandic or English, where a sentence such as this requires a preposition construction (see paragraph 1.5 Preposition Constructions):

IS: Þrjár flöskur af víni > Preposition construction Tveir pakkar af sigarettum 'Three of bottles wine, two packets of cigarettes' (Sigurðsson 2006, p. 13)

11 According to Sigurðsson, it is barely possible to use a genitive case on pseudo-partitive constructions (in contrast with normal partitive constructions) and zero-marking constructions are completely ungrammatical. In this construction, a preposition is mandatory, in contrast to Swedish and German (p. 13).

1.4 Compounds

The somewhat vague relationship between two nouns in a zero-marking construction, in this instance implied by their lexical meaning, is much clearer when they come together in a compound. The following example shows again a genitive case marking in German, in comparison to a compound in Dutch (NL):

DE: Das Land der Tränen > Genitive Case NL: Het tranenland > Compound 'The land of tears'

This genitive case in German does not contain the typical s-suffix, as this is a plural possessor, giving a genitive preposition der and an -n at the word ending. In Dutch, the possessor and the possessee are put together into one word, which can be described as compounding. A possessive relationship is in this case not implied by a morphological suffix, but by having the possessor and possessee (in that order) in a compound. In some Dutch compounds, remnants of the genitive case can be recognised due to a linking- s, found in compounds such as: 'Bakkersroom' [Bakers cream] and 'Stationsgebouw' [Station building] (Weerman & de Wit 1999, p. 125). The first English translation shows the same linking-s, attached to the first noun of the NP. This construction can still be understood as a compound, even though these words are written separately in English. The second construction, 'station building', can easily be mistaken for a zero-marking construction due to the 'loss' of the linking-s. Whether this is a compound or a zero-marking construction would then depend on the semantical meaning of the whole sentence. In West Frisian some compounds still contain a form of the genitive case. These compounds distinguish themselves from normal compounds (N+N) in several ways, but for the purposes of this research, their morphological characteristics are of particular interest (Hoekstra 2002, p. 228). The first noun of the compound ends either with a zero-marking, an -s or an -e. The s-ending appears after a full or a consonant, while the e-ending was historically added to nouns that originally

12 ended with an -e, like: 'Tsjerketoer' [Church tower/The church's tower] (for further explanations see: Hoekstra 2002, p. 230). What is important to note is that the s- and e-ending should, according to Hoekstra, not be confused with the linking-s found in 'normal' compounds, as they are presented in the Dutch examples above (p. 238). According to Hoekstra, the North Frisian dialect area Fering- Öömrang does have genitive compounds, though not in the same form as West Frisian (p. 241).

1.5 Preposition constructions

A construction that does not (necessarily) use a morphological solution, is the preposition construction. A possessive NP with a preposition usually follows the following pattern: possessee - preposition - possessor. The preposition explains the relationship between these two nouns, which can be anything from kinship terms to part-whole relationships. Examples of such a relationship are:

GB: The book of John > Preposition construction NL: Het boek Jan > Preposition construction

In both the English and the Dutch example, the preposition does not govern a case-marking of the possessor (as these Germanic languages no longer use a case-system in the present ), but in German this is still the case:

DE: Samen von schlechten Pflanzen > Preposition construction 'Seeds from bad plants'

In this example, the German preposition von is the equivalent of the English preposition of. In German, this preposition governs the dative case as a suffix to the possessor. This preposition construction with von (or another preposition) can be used when the genitive case is not marked in an NP. It depends on the meaning of the NP as to whether a genitive case or a preposition construction is used in German (Pittner 2014, p. 46-49). The van-construction in the Dutch example has been researched by Weerman & de Wit (1999), who found that use of the construction is relatively high due to the loss of the genitive case (p. 1164). The English of-construction shared a similar diachronic path following the loss of the genitive case and the rise of the preposition construction (which would have happened simultaneously) (p. 1158). Though the form of the van-construction in Dutch is not similar to that of

13 the genitive case, it has semantic overlap, making it a suitable replacement to the case-marking (Scott 2011, p. 105). It is not certain if the same holds true for the of-construction in English, as there are other possessive markers that act independently of this preposition construction. The English s-genitive, for example, is able to attach to the end of an NP, whilst an of-construction must be placed between the possessor and the possessee as a post-modifier. Afrikaans, which originated from the Dutch of the 17th-century colonists, uses the van- construction, but this is not as common as in either Dutch or English (Roberge 1996, p. 124). This does not have the same usage as in Luxembourgish, where a vun-construction (equivalent to the Dutch van-construction) is often the preferred possessive construction1. For both Elfdalian and Faroese, research has been conducted on the use of possessive preposition constructions and which prepositions can be used. Lundqvist (2011) found that when the Swedish s- genitive was translated into an Elfdalian preposition construction, the following prepositions were possible: að, eter, frå, i, millå, å and yvyr (p. 46). An example of a possessive preposition construction with the preposition i [in] is

EL: Swertą i wattnę > Preposition Construction 'The water's blackness' (Lundqvist 2011, p. 46)

It does not become clear which semantic rules apply when using a specific preposition in Lundqvist's research. What is interesting about these different prepositions is that, similar to German, the prepositions govern a case-marking. With the loss of the genitive case, the preposition sometimes shifts to a different case, meaning that (for example) the preposition að no longer governs the genitive, but the dative case (Eekman 2013, p. 21). In Faroese, the preposition most often used is hjá [of]. This preposition governs the dative case. It depends on the semantics as to whether other prepositions can be used (Thráinsson, Petersen, Jacobson, Hansen 2004, p. 62).

1.6 Personal pronouns

A possessive relationship is often expressed by a form of a , for example: 'My house' or 'His bike'. These examples contain a personal pronoun as possessor and a noun as possessee. This differs when a personal pronoun is placed between the possessor and possesee:

1 This information comes from an online platform called . The article is on the Luxembourgish' his-genitive: https://www.duolingo.com/comment/22324745/Luxembourgish-Lessons-38-His-Genitive/

14 *'My mother her cat'. This form of a possessive NP with a personal pronoun is often found in Afrikaans (AF):

DE: Der Boden des Planeten > Genitive case AF: Die planeet se grond > Personal pronoun construction 'The soil of the planet'

The possessive relationship in the German example is again expressed by the article des and the en- suffix attached to the possessor. The possessive relationship in Afrikaans changes the position of the possessor and possessee, and the possession is expressed by the personal pronoun se. The implication of a possessive relationship now lies with the semantic meaning of the resumptive personal pronoun. The personal pronoun construction is known in the by the term: 'n- construction. This construction is not (diachronically or synchronically) related to the genitive case but shows functional similarities with the English s-genitive (Weerman & de Wit 1999, p. 1174- 1175). As described earlier, the English s-genitive can be attached to the end of an NP and is therefore relatively independent (in comparison to the Dutch s-genitive). The z'n-construction appears to have the same level of dependency:

GB: A friend of mine’s house > S-genitive NL: *Een vriend van mij's huis Een vriend van mij z'n huis > Personal pronoun construction (Weerman & de Wit 1999, p. 1174)

According to Weerman & de Wit, the Dutch z'n-construction shows functional similarities with the s-genitive in both English and and Swedish (p. 1176). The se-construction in Afrikaans shows several similarities with the Dutch z'n-construction. The word se derived from the Old Dutch masculine pronoun 'syn' [his] (Roberge 1996, p. 124). Roberge compares the se-construction with the English s-genitive and the Dutch z'n-construction, to show its high level of independence:

15 AF: Die man wat nog in die hospitaai behandel word, se toestand is kritiek. NL: (*) De man die nog in het ziekenhuis behandeld wordt z'n toestand is kritiek. GB: * The man who is still being treated in the hospital's condition is critical. (Roberge 1996, p. 125)

As well as the Dutch z'n-construction and the English s-genitive, the Afrikaans' se-construction can be used after an NP. What makes the se-construction unique is the ability to be attached to a complete subordinate clause that defines the possessor. The Dutch example feels rather unnatural when using a z'n-construction after a subordinate clause. The English example with the s-genitive is ungrammatical. Roberge proves the similarities between the personal pronoun constructions in Dutch and Afrikaans and the English s-genitives, but also shows the differences in dependency. This is supported by Norde (1997) who claims that these constructions might seem compatible but neither is as free in use as the Afrikaans' se-construction (p. 61). The personal pronoun construction can also be found in Luxembourgish, where the source from Duolingo calls this a 'his-genitive'. A construction like: 'Dem Man säi Buch' [*This man his book] would show a personal pronoun that takes over the form of the genitive ('säi') and therefore implies the possessive relationship between the possessor and possessee. For this research, the term 'his-genitive' will not be used and instead be replaced by a personal pronoun construction, as this also refers to the z'n- and se-construction. In Norwegian, a possessive construction with a personal pronoun is not unknown. This is often referred to as an h-genitive, which is not common in other mainland Scandinavian languages such as Swedish and Danish. The h-genitive has the following structure: possessee - possessive pronoun - possessor, and with that is different to the regular structure of linking pronoun constructions (the possessee and possessor are reversed) (Dahl 2015, p. 183). This becomes clear in the following example:

NO: Huset Per > h-genitive GB: *Per his house > Personal pronoun constructions (Dahl 2015, p. 183)

This construction is also found in Icelandic and in some . There is quite a lot of discussion in the linguistic field about whether this h-genitive is not 'just' a possessive pronoun construction (Koptjevskaja-Tamm 2003, p. 630). This will not play a role in this thesis, as I will refer to both as a personal pronoun construction.

16 1.7 Subordinate clauses

The last grammatical construction named in this chapter lies between free and possessive constructions. When using a subordinate clause to express a possessive relationship, the possessor is paraphrased, as showed in the following example in Elfdalian:

DE: Ein Rosengarten > Compounding EL: Ienum traigard, so full i ruosum > Subordinate clause 'A rose garden'

The German example shows a possessive compound, where the possessor comes before the possessee. The Elfdalian example shows the possessee, where the possessive relationship is explained in the subordinate clause.

1.8 Summary

In this chapter, an overview of possible possessive constructions was established. For constructions such as the s-genitive and preposition constructions, several lines of research were followed, leading to an insight into the use of these constructions per language. In particular, the research into the s-genitive was extensive, and comparisons between some of the languages discussed already started to become established. According to this background information, it can be surmised that the s-genitive in English, Swedish and Danish is similar in use as a 'group genitive'. The following rules can also be established:

• The Dutch s-genitive is more restricted in use and cannot be attached to the end of an NP. The s-genitive has a different form in Elfdalian and Faroese, where a different case or ending is used. • The zero marking construction is used in various different forms in Swedish, German, Icelandic and English, depending on the use of partitives. • The preposition construction, mostly described with the preposition of (or an equivalent), is in use in many of the languages. However, this construction is more popular in English and Dutch than it is in Afrikaans.

17 • Afrikaans uses a personal pronoun construction, which has similarities with the s-genitive in English.

Regarding other constructions, such as subordinate clauses and dative/accusative cases, a comparison has not yet been made between the languages. The amount of data connected with these constructions is limited and shows the gap in the current research field. Even with the more-deeply researched constructions, this research can shed a light on the accuracy of the comparisons between the languages.

18 2. Methodology

The goal of this thesis is to analyze possessive constructions in Germanic languages. In order to be able to do this, the number of variables are kept to a minimum, as the constructions and languages can already vary. This means that there should be a set amount of tokens from a comparable source. Based on this set amount of examples it will then be possible to create a numeric overview of data, as opposed to an otherwise one-by-one review of individual examples. In this chapter, the sources for the tokens will be explained, as well as the basis overview and research planning. In the last paragraph of this chapter, various hypotheses will be given and it will be explained how these will be researched and checked.

2.1 Collection of data

The possessive constructions are acquired from the book Le Petit Prince of Antoine de Saint- Exupéry (1943) as this work meets the conditions set in the previous paragraph. Because this book has been translated into nearly all the languages that are of interest for this research, it is also possible to use comparable data for each language. The other reason this book is suitable for this research is the fact that it has been written from the point of view of a child. This, in the first instance removes the necessity (and obstacle) of fluency required in the languages covered in this thesis, due to the outwardly simple nature of the text in question. In the second instance, this is also a text written for children, but to be understood by adults (in a similar vein to the work of .S.Lewis and Oscar Wilde), which means that while the text may be simple, the meaning remains sophisticated, and the necessity of a quality translation remains high. The simple yet layered nature of this text therefore makes this book an ideal subject for a translation comparison. There is a certain advantage in using a French original text for acquiring possessive constructions. This is that a possessive construction is easily recognized in French, as it can be expressed in four different ways: with adjectives, pronouns and two different prepositions. The first two constructions mainly contain personal pronoun constructions (with for example: votre, vôtre, son and sien) and will therefore not be taken into account, as the translation of personal pronoun constructions is not of interest for this research. Out of the two different preposition constructions (á after the verb être and the preposition de), the possessive constructions formed with de (or du/des depending on the noun following) are of special interest. Due to the fact that this possessive marker is an equivalent of the preposition of/van in Germanic languages, it is used as a compound marker and as a partitive marker, and will therefore show possessive constructions between two nouns

19 without personal pronoun constructions. Apart from these features, the preposition constructions are easily recognizable as they appear only in three varieties of the construction (de/du/des). The preposition construction with á after the verb être will not be taken into account, as the scope of this research does not encompass possessive verb phrases. The possessive constructions formed with the possessive preposition de are written down per token in an Excel sheet, omitting any repetition (e.. an example such as 'la planète du Petit Prince' [the Little Prince's planet] (p.27) is only given once). The examples are marked by number in order to keep an overview and are marked as to whether it is a partitive construction. These tokens are then similarly noted in the translations in the other languages.

2.2 The Languages

The languages used for this research are German, Icelandic, Danish, Swedish, Elfdalian, Faroese, Fering, Dutch, English, Luxembourgish, and Afrikaans. As shown in the background overview, German and Icelandic maintained the original form of the genitive case. Icelandic is the most conservative language in the Germanic language group and is therefore of special interest for this research. As German does use the genitive case but is not closely related to Icelandic (there is no pattern of descendance), this language is also taken into account. Danish, Swedish, Elfdalian, Faroese, and Icelandic all descend from . Faroese, being the other insular Scandinavian language (next to Icelandic), is therefore taken into account, since it is different from other Scandinavian languages and is also relatively distant from Icelandic. This can be explained by the difference in use of the morphological case system. Danish and Swedish, being two of the biggest Scandinavian languages, may be closely related, but are known to be less closely related than, for example, Danish and Norwegian (especially syntactically). That is why Norwegian (Bokmål) is not taken into account, but Danish is. could be of interest as an example of a counter-language to Danish, but a translation of Le Petit Prince was not accessible in Nynorsk and could therefore not be taken into account in this research. Swedish is, therefore, the second most significant Scandinavian language used (as it has the largest speech community of all the Scandinavian languages). For both Danish and Swedish it will be interesting to see how the s- genitive is used in comparison to the English s-genitive, or the genitive case in German and Icelandic. This is due to their assumed similarities, as presented in the background information. Elfdalian is not an officially recognized language, but is known to be as different from Swedish as Icelandic. The Elfdalian dialect is also known to have a case-system, although the

20 genitive case is now barely used. Because Elfdalian used the dative case (amongst other constructions) to replace the genitive, this language is taken into account as it can be compared to the construction strategies of other Germanic languages. The Dutch language is spoken by at least 16 million people in the Netherlands alone and is therefore a significant Germanic language. Its noun-based case system is practically defunct but it does contain an s-genitive, which is of interest for this thesis. The same holds for English, which is also the language that is most widely spoken in the Germanic . Both languages use other ways of expressing possession than using a genitive, whether this is an s-genitive or another construction. Additionally, in the North of the Netherlands and Germany, there are many languages and dialects that come together under the name ''. The translation of Le Petit Prince is available in both Westlauwers Fries (Frisian spoken in the Netherlands) and Fering (a dialect of North Frisian, spoken on the island of Föhr), but only the Fering translation is taken up in this research. West-Frisian dialects are more closely related to Dutch than Fering (Simons & 2017) and Dutch has a strong influence on Frisian syntax and morphology (Gorter 1999, p. 115). Fering, having a weaker relationship with its neighbouring languages of Dutch, Danish, and German (Simons and Charles 2017), is therefore, more interesting for this research. Luxembourgish can be of particular interest due to its trilingual state. As there are varieties near to the Belgian border with more French words, and as Luxembourgish is as far removed from as Dutch (Simons & Charles 2017), Luxembourgish is used in this research. The last language chosen is Afrikaans. This language is interesting due to its geographical position and due to the fact that for many years, there was barely any influence stemming from Germanic languages. Since a hundred years or more, English is spoken in , meaning that there is more and therefore a greater likelihood of Anglicisms in the language. As Afrikaans originated from Dutch and now has close connections to English, it is interesting to study the closeness of the bond between these languages when discussing possessive NPs.

2.3 The translated texts

The tokens collected in the original French texts are searched for in the translated texts. The followings translations are used:

 English: The Little Prince 2009 Translated 1943 by K. Woods  Dutch: De kleine prins 2012 Translated 1951 by L. de Beaufort van

21 Hamel  German: Der Kleine Prinz 2015 Translated 2015 by M. Herbert  Danish: Den prins 2013 Translated 1950 by unknown

 Swedish: Den lille prinsen 2015 Translated 2015 by N Gullberg Zetterstrand & H. Petersen  Elfdalian: Lisslprinsen 2007 Translated 2007 by . Westling  Icelandic: Litli prinsinn 1988 Translated 1988 by Þ. Björnsson  Faroese: Tann lítli prinsurin 2015 Translated 2013 by A. Kristiansen  Fering: De letj prens 2016 Translated 2010 by A. Arfsten  Luxembourgish: De klenge Prënz 2009 Translated 2009 by J. Braun  Afrikaans: Die klein prinsie 2015 Translated 1994 by A.P. Brink

The translations of the English, Dutch and Danish texts were all made before 1951, the Icelandic translation is from 1988, the text in Afrikaans is from 1994 and the other translations are all conducted after 2007. Though the translations of the English, Dutch and Danish Le Petit Prince were written more than fifty years ago, they were all reprinted from 2009 onwards. One can assume therefore that the text did not need to be revised in order to be read by a readership of the 21st century. Therefore the texts are suitable for research, when compared to the more recent editions. The Icelandic translation was written in 1988 and is the only extant version, as well as the only one ever published. As Icelandic is the most conservative language of the Germanic language group, and has not changed in recent years, no difficulties of meaning or semantic change are foreseen with using this translation across an extended time period. The Afrikaans version was written in 1994 and reprinted in 2015. The timespan between the first and second edition is only twenty years, which makes the translation valid for this research due to the fact that language changes are minimal during a single generation (which would be longer than twenty years). All the other translations mentioned are more recent. For some of the texts, this was the first translation; it was the first translation in Eldalian, Faroese, Fering and Luxembourgish. The German version is a re-translation, as the previous translation had been conducted as long ago as 1950 and was slightly outdated (Gerling 2015).

22 2.4 Categorizing the data

As mentioned above, the possessive tokens of the French text are collected and put into an Excel spreadsheet. These tokens (now numbered) are searched for in the translated texts ad written down in the spreadsheet. When written down, the tokens are analyzed regarding what occurred with the possessive constructions in the translation, leading to the following categories:

Direct translations:

• Genitive The possessive construction is transformed into an NP with a genitive case. Instead of using a syntactical strategy, a morphological one is applied. This category contains the regular genitive case as used in Icelandic and German, as well as the deformed (s-genitive) genitives that can be found in some other languages. For example:

FR: La vie du petit prince GB: The little prince's life

• Dative The possessive construction is transformed into an NP with a dative case. Instead of using a genitive case, this translation makes use of the dative case to express a possessive relationship. For example:

FR: La planéte du géographe EL: Plajetn landlärdam [The geographer's planet]

• Preposition The possessive construction, being in French already a preposition construction, becomes a preposition construction in the translated text as well. For example:

FR: La copie du dessin GB: A copy of the drawing

23 • Compound The possessive construction is transformed into a compound, resulting in a word where both the possessor and possessee are recognizable. For example:

FR: Un allumeur de réverbères GB: A lamplighter • Zero marking The possessive construction is no longer visible through an extra marker. This means that the possessive relationship is implied. For example:

FR: Un crayon de couleur GB: A coloured pencil

• Personal pronoun construction The possessive construction is translated into an NP with a personal pronoun. For example:

FR: L'intérieur du serpent boa AF: Luislang se binnekant [The inside of a boa constrictor]

• Subordinate clause The possessive construction is translated and expressed through adding a subordinate clause. The possessor is written before the subordinate clause, while the possessee is featured in the additional sentence. This category lies somewhere between direct and indirect translations, due to the rewriting of the possessee in an additional NP. For example:

FR: Un jardin fleuri de roses GB: A garen, all a-bloom with roses

24 Indirect/free translations:

• Verb phrases The possessive construction is translated into an active phrase, making it a VP instead of an NP. The possessive construction is still recognizable. For example:

FR: La planète du petit prince GB: The planet where the little prince lived

• Non possessives The possessive construction in French does not result in a possessive construction in the translated text. However, the construction is nevertheless translated. An example can be a possessive construction in French that is expressed through one noun in the translated text, while having a similar meaning. For example:

FR: De courants d'air GB: Draughts

• Not translated The possessive construction is not translated and therefore not recognizable in the translated text.

After collecting and organizing the data from all the translations, a general analysis can be carried out. Firstly the amount of token per category is written down for every language. Then the total amount of tokens per category is summarized. Using these numbers, it will be visible which construction is favoured in all of the Germanic languages concerned, as well as which construction is preferred per language. For the purposes of displaying any constructions over-represented in a certain language, a table based on absolute and relative numbers will be outlined, in order to clarify whether the over- representation is not only based on absolute numbers but on frequent occurrences (median). This is based on a comparison with the other constructions in the language itself and the usage of that construction in other languages. Based on the general table with the total amounts of constructions per language it is also

25 possible to determine which languages hold a stronger relation to one another. A correlation table will be formed in order to understand in which gradation languages are similar to each other, based on possessive NPs.

2.5 Analysis and hypotheses

In order to find whether there are any over-representations in a certain language regarding possessive constructions, the data must be analyzed using both absolute and relative numbers. In order to find these numbers, firstly an average per constructions must be found. The following formula is applied: (total amount of a construction) / (total amount of constructions of all languages) * z (total amount of constructions per language) = a (average). For example: 120 (preposition constructions in all languages) / 2000 * 200 = 12. With a total amount of 10 tested constructions, 10 averages are calculated. With these averages, a complete table with absolute numbers can be formed. This means that for each language an absolute number can show the absolute over-representation per possessive construction. It will be calculated as follows: (real number of a possessive construction in a language) - a (average of that possessive construction) = b (absolute number of a possessive construction in a language). For example: 120 (total amount of preposition constructions in English) - 12 (average of preposition construction) = 108. When conducted for all languages in relation to possessive constructions, the occurrences of over-representation become visible. In order to see if these over-representations still exist with relative numbers, the following formula is applied: b (absolute number of a possessive construction in a language) / a (average of that possessive construction) = c (relative number of that possessive construction in a language). For example: 108 / 12 = 9. With these relative numbers, it becomes clear that there is a relatively large number of occurrences based on a comparison with the other constructions in the language itself, and the usage of that construction in other languages. To understand which languages hold a stronger relationship with each other, the correlation coefficient is calculated between the cell ranges of two matrices. This correlation coefficient will show to what extent the matrix are related. In this research the two matrices will be extracted from the first table with absolute numbers, using the following formula: CORRELATION (matrix 1; matrix 2) (part of mathematical formula performed by Microsoft Excel). For example: CORRELATION (all constructions in English; all constructions in Dutch). The comparison between the use of different constructions in those two languages will give a percentage, that shows how closely they are related. This means that a 1,0 (or 100%) is the highest possible correlation and 0,0

26 (or 0%) the lowest possible correlation. The following relationships are hypothesized:

I. Scandinavian mainland (Swedish and Danish), due to their shared history and the common ground on possessives as explained in the background information (i.e. the specific rules for the use of the s-genitive and preposition constructions). II. Genitive preserving languages (German and Icelandic), due to their morphological similarities. Both languages maintain the original genitive form, though possibly not always under the same conditions (Icelandic has other rules regarding the partitive genitives).

III. Dutch and Afrikaans, due to the fact that Afrikaans partly descends from (colonial) Dutch. Both languages seem to make use of the personal pronoun construction, although this in Dutch is more of a colloquial form; while in Afrikaans, this is also seen in the written language. As this construction is also described in Luxembourgish, one can expect a higher correlation between these languages.

The other possible relationships can only be guessed at, having no strong indication of common ground based on the expression of possessive relationships in these languages. It is indeed possible to find differences and similarities based on, for example, the use of the s-genitive, but this does not imply relationships between the languages and are therefore not treated as such. The second hypothesis is based on the use of the genitive case. As there might be a difference regarding the use of this case, an analysis will be conducted to research this specific possessive construction. First of all, the translation that contains the most genitive cases (from Icelandic and German), will then be used as a 'filter'. This means that if there are, for example, 50 genitives in the German translation and 40 in the Icelandic translation, a number of tokens will be set to 50. Subsequently, a table of real numbers can then be drawn up, indicating how these 50 tokens have been 'translated' into Icelandic and the other Germanic languages. With this table of real numbers, the absolute and relative numbers can be acquired as described above for the overall analysis. After that, a correlation table will be formed (as described above on the general analysis) which should illustrate how the languages correlate when specifically filtered in the genitive case in either German or Icelandic. In the case of other over-representations, a similar analysis can be conducted. This depends on the relative over-representation found in the general analysis of all the data acquired from the translations.

27 3. Results

3.1 General overview

In this research on the translation of possessive NPs in Germanic languages, 191 French possessive NPs with de/du/des were found and analyzed in the translation of Le Petit Prince. The analysis was performed on eleven translations in the following languages: English (GB), Dutch (NL), German (DE), Swedish (SE), Danish (DK), Icelandic (IS), Faroese (FA), Luxembourgish (LU), Afrikaans (AF), Fering (FE) and Elfdalian (EL). Per language, the translated tokens were marked by either a direct translation: preposition construction (preposition), compounding (compound), zero marking (zero marking), adding of personal pronoun (personal pronoun), genitive case (genitive), dative case (dative), or rewriting into a subordinate clause (sub clause); or by indirect translations: rewriting into a verb phrase (VP), no possessive construction (no possessive), or by being not translated (not translated). All tokens were organized in to a table with real numbers, meaning that it shows how often a certain construction is used per language:

Table 1: general results GB AF NL LU FA FE SE EL DK IS DE TOTAL Preposition 127 79 79 71 67 63 63 63 52 43 32 739 Compound 15 27 31 27 32 18 37 40 35 47 37 346 No possessive 8 11 35 28 28 33 14 37 17 20 18 249 VP 14 18 14 18 23 32 27 26 26 21 17 236 Zero marking 19 30 29 26 17 23 23 6 26 7 23 229 Genitive 6 0 1 1 2 0 21 1 29 35 62 158 Sub Clause 2 6 0 8 13 18 3 5 4 12 0 71 Not translated 0 5 0 7 4 3 3 3 2 3 0 30 Personal pronoun 0 15 2 5 1 1 0 2 0 1 2 29 Dative 0 0 0 0 4 0 0 8 0 2 0 14

With a 191 possessive construction per translation, the total amount of tokens is 2101. From all these constructions used, 32.5% (739) were preposition construction, making it the most popular possessive construction. Other percentages were: compound (16.5%), no possessive (11.9%), VP (11.2%), zero marking (10.1%), genitive (7.5%), sub clause (3.8%), not translated (1.4%), personal pronoun (1.4%) and dative (0.7%). For some of these constructions there appears to be an over-representation in a language. These are, for example, the preposition construction in English (127 out of 191), the genitive in German (62 out of a total of 158 genitive constructions), VPs in Fering (32 out of a total of 236

28 VPs), and the dative in Elfdalian (8 out of a total of 14 dative constructions). Some of these constructions may seem over-represented in the total number but are relatively little used in the language itself. When looking at the Elfdalian example, there might be a relatively high proportion of datives in the Elfdalian language but with only 8 out of 191 tokens, it is still an uncommon construction. To understand whether the assumed over-representations can be proved by numbers, an analysis based on absolute and relative numbers is required. First of all, the averages were obtained for all constructions per language with the following formula: x (total amount of a construction) / y (total amount of constructions of all languages) * z (total amount of constructions per language) = a (average). For example the preposition constructions: 739 / 2101 * 191 = 67.2. The other averages are compound (31.5), no possessive (31.5), VP (22.6), zero marking (20.8), genitive (14.4), subordinate clause (6.5), not translated (2.7), personal pronoun (2.6) and dative (1.3). The averages were put into the following formula to acquire the absolute numbers: r (real number of a possessive construction in a language) - a (average of that possessive construction) = b (absolute number of a possessive construction in a language). For example, the preposition construction in English read as: 127 - 62.7 = 59.8. The following table shows an overview of the absolute numbers:

Table 2 shows a high score for preposition constructions in English, the genitive case in German (and to lesser extent in Icelandic) and the dative case in Elfdalian. There are certainly more languages which show a high use of a certain translated construction, although they might not be as evident as those in English, German, Icelandic and Elfdalian. To support or falsify these assumptions, an overview of relative numbers is needed. These are acquired through the following formula: b (absolute number of a possessive construction in a language) / a (average of that

29 possessive construction) = c (relative number of that possessive construction in a language). For the preposition construction in the English translation this led to the following equation: (127 - 67.2) / 67.2 = 0.9. The following table shows the relative results for all languages and possessive constructions:

The absolute numbers may have shown an over-representation of preposition constructions in English (59.8), but Table 3 shows a relatively low over-representation with 'only' 0.9. The genitive case in German leads to a relative over-representation of 3.3, which means that it is more widely over-represented than the preposition construction in English. The same holds for the genitive case in Icelandic and even in Danish. In this table the relative over-representation of the personal pronoun construction in Afrikaans becomes visible with 4.7. Fering reaches a 1.8 on subordinate clauses and the dative case in Elfdalian amounts to the relative number of 5.3.

3.2 Correlation

The overall results of the use of the translated constructions give a general idea of which languages might be more closely related than others, based on the usage of certain possessive constructions. In any case it would be logical to find a close relationship between German and Icelandic, as these languages are highly marked on the genitive case. To understand to which extent these languages (and the others) correlate, an examination based on correlation is conducted. With this, the degree of relationship between languages becomes apparent. The formula to calculate correlations was implemented into Excel, meaning that only the two matrices had to be filled in, in order to compare them. This led to the following table:

30 The table shows that most correlations are between 80% and 100%. It is for these reasons that the less evident relationships between languages become much more apparent. Firstly, German has the lowest correlation with nearly every other language (all below 80%). Icelandic behaves in a similar way, where only a correlation above 80% was found with Swedish and Danish. Because these four languages contain the most genitive cases, their correlation is unsurprising, but will be investigated further in the next paragraph. The highest correlation is between Dutch and Luxembourgish (98.8%), and between Swedish and Danish with 97.5%. The lowest correlation is between German and Fering with 19.1%. Most languages show a declining relationship with other languages, where the lowest is with German and Icelandic. Although this does not hold true for all translations, it shows a similar pattern that can give an insight in the relationships between the Germanic languages.

3.3 Languages with original genitive case

Icelandic and German are two languages that have preserved a morphological case-system with full use of the genitive case. In the first overview of the used constructions (Table 1), it shows that the genitive case is more widely used in German (62) than in Icelandic (35). As described in the methodology, the language that contains the most genitive constructions acts as the filter for an analysis on the 'translation' of the genitive, as well as a comparison between the German and Icelandic genitive. The outcome of such an analysis would establish if the Icelandic genitive occurred in the same instances as the German genitive and if they would be used under the same circumstances. It would also show whether the s-genitive in other languages would occur in the same instances as the German or Icelandic genitive. With a total amount of 62 tokens, the following overview was conducted:

31 The total amount of tokens is now 620. Out of these 620 tokens, 292 were preposition constructions (47.1%). The other percentages are VPs (12.9%), genitive (11.3%), compounds (7.7%), no possessive (7.6%), zero marking (3.9%), personal pronoun (3.7%), sub clause (3.1%), not translated (1.6%) and dative (1.1%). In the general overview German had the highest correlations with Icelandic, Swedish and Danish. To see whether these languages used the genitive case in the same situations as the German genitive, one could make a comparison with the overall results (table 1) and the results filtered on the use of the German genitive (table 5). Out of the 35 genitives in the Icelandic translation, 25 were used when there was a German genitive. The Danish translation contained 29 s-genitives, of which 22 occurred at the same instance as the German genitive. 22 s-genitives were found in the Swedish translation, of which 15 were used when the German translation used a genitive case. Elfdalian and Faroese maintained a case system but nevertheless lost the original genitive case. Still, case-marking can be used to express a possessive relationship. They can also be compared to the use of the German genitive case. Elfdalian used one genitive case in its translation and this was used when the German translation used a genitive. The dative case was used 8 times, and occurred 6 times, when filtered against the German genitive case. The two genitives in Faroese were used at the same time as the German genitive. To understand which grammatical constructions were used when German used a genitive case, an analysis was established on absolute and relative numbers. Firstly the averages were obtained: preposition (29.2), compound (4.8), no possessive (4.7), VP (8.0), zero marking (2.4), genitive (7.0) sub clause (1.9), not translated (1.0), personal pronoun (2.3) and dative (0.7). The results are presented in the following tables:

32 In absolute numbers, Icelandic (18.0) and Danish (15.0) score highly on the genitive case. In relative numbers it becomes clear that this is indeed an over-representation (2.6 and 2.1) but not nearly as high as other constructions. The dative case in Elfdalian has an absolute number of 5.3 and with that a relative over-representation of 7.6. The personal pronoun construction in Afrikaans has a high absolute number of 12.7 and a relative number of 5.5. The Dutch translation shows a relatively large amount of construction which are not translated (relative number 3.0). Fering used relatively many subordinate clauses (2.7). Based on these results, a correlation table was established. Again this was filtered on the German genitive case in order to see which constructions were used in the other languages:

33 Most correlations are again relatively high, although more differentiated than the correlations of the general results. This can be illustrated by comparing the correlations between Icelandic and the Scandinavian languages. In the general results, the correlations between Icelandic and Scandinavian languages is: Swedish (83.2%), Danish (87.6%), Elfdalian (72.0%) and Faroese (68.4%). The numbers shift quite considerably when filtered on the German genitive case: Swedish (67.9%), Danish (86.0), Elfdalian (17.1%) and Faroese (17.1%). Only the correlation with Danish appears to be stable. As Elfdalian and Faroese use the genitive case very little, this influences the correlation when filtered on the German genitives. English and Dutch now show the most similarities with a correlation of 98.0%. This is a small shift from the 89.4% in the overall results. Danish and Swedish show a relationship of 97.5% in the general analysis and that has lowered to 91.8%, when considering the genitive case.

3.4 Partitives in translation

The background information shows that Icelandic has different rules in expressing a partitive possession. Not all of them can be expressed by genitive or by zero marking, which is different to, for example, Swedish and German. Of the 191 French sentences, 25 were recognized to be partitives. Examples are: 'Une feuille de papier' [a piece of paper] (p. 21) and 'Deux milliards de grandes personnes' [two billion grown-ups] (p. 72). These 25 tokens were used as a filter to analyze how these 25 were translated into the Germanic languages, showing the following results:

34 The total amount of partitives were 275. Most of these were translated with a zero marking construction (132). In Afrikaans, Dutch, Luxembourgish, Fering, Danish and Swedish this was more than half of the tokens (between 64% and 56%). Preposition constructions were the second most used translation method (93). An example that stands out for its translation is: 'A piece of paper'. This is translated with a compound in most Scandinavian languages, with a preposition construction in English, and in other languages with a zero marking construction:

Zero marking P Preposition P Compound P NL een blaadje papier 10 GB a sheet of paper 8 EL papirbit 12 DE ein Blatt Papier 9 IS pappírblað 10 DK et lille stykke papir 10 FA pappírslepa 8 AF n vel papier 10 SE pappersbit 6 LU e blat Pabeier 12 FE en stak papiar 12

German, seemingly using more genitives in general, used the genitive case only once in a partitive construction. Therefore the genitive case was, with only 12 occurrences, not a popular construction. In Icelandic, the genitive case was used 6 times and was therefore less significant in translating a partitive construction.

3.5 Languages with deformed genitives

3.5.1 English

The highest proportion of preposition constructions is found in the English translation. With 127 constructions out of 739, this accounts for 17.1% of the total amount of prepositions. As there were 191 tokens per language, 66.5% of the English tokens were created with a preposition construction. In absolute numbers, the over-representation of English preposition constructions was 59.8, and in

35 relative numbers, still 0.9. In comparison with other relative numbers, this was by far not the highest over-representation in the general results but did give rise to the question of which other translations were used when English used a preposition construction. Therefore, the English preposition constructions were used as a filter and these 127 constructions were analyzed in the other Germanic languages. This led to the following overview:

The preposition construction remains the most used construction, due to its overall number. Therefore it can be noted that the Afrikaans translation used the preposition construction most often but that does not lead to new insights. The constructions which replaced a preposition construction were most often non-possessives, VPs, compounds, zero markings and genitives. The amount of tokens per constructions show relatively little differences, but do show varieties per language. For example, the genitive case is mostly used again in German with 52 tokens. Additionally, Icelandic, Danish and Swedish used relatively many genitives. When comparing to the general results, a following comparison can be made:

Of the 62 genitives used in the German translation, 52 were used at the same time as the English preposition construction (83.9%). The Icelandic has less genitives, with 35, of which 24 were used when English used a preposition construction (68.6%). Danish and Swedish have almost the same

36 rate of use, with 72.4% and 71.4%. It is outstanding that the s-genitive was sparsely used in the English translation. The following 6 tokens contained a genitive marking: 'The little prince's life' (p. 23), 'The game's monotony' (p. 40), 'The little prince's arrival' (p. 41), 'That explorer's moral character' (p. 51), 'The little prince's ankle' (p. 56) and 'The switchman's cabin' (p. 70). Out of 191 tokens, these examples count only for 3.1% of all the translated constructions.

3.5.2 Dutch

This translation contained a single NP in the Dutch translation, which used a remnant of the former genitive case: 'Vier uur 's middags' [four o'clock in the afternoon] (p. 48). There were no NPs that used an s-genitive. In Dutch there were a relatively high number of zero marking constructions. This was generally the case when the original possessive construction in French was a partitive. Out of the 25 partitives, 15 were translated into a zero marking construction. An example is: 'Een blaadje papier' [a piece of paper] (p. 10). The zero marking construction was also found often when there was a partitive based on numbers, such as: 'Vijfhonderd miljoen putten' [Five hundred million wells] (p. 60). Dutch had a high correlation with nearly all Germanic languages. With Luxembourgish this was even 98.8%. However, the expected high correlation with Afrikaans is also visible, with 92.5%. Their biggest difference seemed to be the use of the personal pronoun construction. The lowest correlation is with German, with only 34.9%. This is most probably due to their different morphologies.

3.5.3 Luxembourgish

The translation into Luxembourgish contains one genitive case: 'Mueres fréi' [sunrise] (p. 80). With that, it follows the same pattern as most other Germanic languages, which barely use the genitive case. That is supported by the correlation table, which shows high correlations with English (92.0%), Dutch (98.8%), Afrikaans (92.5%), Faroese (97.3%) and Fering (93.5%). Even the correlation with Swedish and Danish, where the s-genitive is rather over-represented, is still above 80%. The correlation with Icelandic and German is much lower, with 59.2% and 28.4%. Most translated constructions are not over-represented in Luxembourgish. It does contains the most no translations (7), which means that the translator did not translate the French possessive

37 construction, or even the complete sentence.

3.5.4 Faroese

The Faroese translation contained 6 tokens with a form of case-marking. Two NPs were conducted with the genitive case: 'Táranna land' [the land of tears] (p. 25) and 'Koningsins fingur' [the finger of a king] (p. 54). The latter is characterised by the double s-genitive, showing the addition of s-in-s to a noun. This is a rendering of the old-genitive case. The other four tokens contained a dative case. The Faroese translator used quite a number of subordinate clauses (13), though this was not as over-represented as with Fering (1.0 in Faroese and 1.8 in Fering). Other constructions did not show a mentionable over-representation. Faroese shows a high correlation with nearly all Germanic languages, except for Icelandic and German. The few similarities with Icelandic can be explained by their differences in case systems. This also explains the low correlation with German.

3.5.5 Swedish

An over-representation was found of the s-genitive in the Swedish translation. This had a relative number of 0.5, which is the highest relative number in the Swedish construction. In comparison to Danish (1.0), Icelandic (1.4) and German (3.3) this is still not a large over-representation but is nevertheless of note, due to the overall lack of genitives in the other Germanic languages. With 21 genitive tokens, it is not the most widely used construction (this being the possessive preposition constructions). As presented in the background information, a preposition construction cannot be used with an alienable possession; in which case, an s-genitive can be used, as can be illustrated by the following example: 'Generalens fel' [the general's fault] (p. 15). In English, the preposition construction is allowed, resulting in a different translation: 'The fault of the general' (p. 35). Swedish has the highest correlation with Danish (97.5%) which means that in terms of possessive constructions they have a very close relationship. This can be explained by their similar rules on preposition constructions and their use of the s-genitive. Other possessive constructions were similarly distributed in the translation of Le Petit Prince. It has only a low correlation with German with 62.1% but it is much higher than what the majority of the other Germanic (and non- Scandinavian) languages have with German.

38 3.5.6 Elfdalian

The Elfdalian translation contained one deformed genitive and eight dative cases. The deformed genitive was formed as described in the background information, with the dative case and the suffix -es: 'Tårumes land' [The land of tears] (p. 30). The other possessive relationships marked by a morphological case were datives: 'Tekkningär författeram' [Drawings of the writer] or 'Planjetn landlärdam' [The planet of the geographer] (p. 53). Even though the amount of dative cases was only 8, it still represented a large over-representation when compared to the other Germanic languages, with a relative number of 5.3. These were not used in partitive constructions (see Table 9), nor is it certain if they were used at the same instances as the genitive case in German. Of the 8 tokens, 6 were used when the German translation had a genitive case. Elfdalian, together with Icelandic, showed an 'under-representation' of the use of zero marking constructions (6). Other languages used at least 17 or more of this construction, regardless if they had an active case-system or not. In terms of correlation to the other Germanic languages, the Elfdalian translation stood closest to Faroese with 95.3. The correlation with Dutch and Luxembourgish also scored highly, with 90.2% and 89.3%, Again the lowest correlation was with German, with only 29.6%. This can probably be explained by their different case-system and the lack of the genitive case in the Elfdalian translation.

3.5.7 Danish

The s-genitive was mostly found in the Danish translation with a total amount of 29 tokens and had a relative over-representation of 1.0. This was rather similar to the original genitive case in Icelandic, with 35 tokens and a relative number of 1.4. The Danish genitive was used often (75.9%) when the German translation had a genitive case marking. When comparing the s-genitive in Danish and Swedish, their correlation lowers a little from 97.5% till 91.8%. This can also be seen in their relative numbers, when being filtered on the German genitive. Danish has an over- representation of 2.1 and Swedish of 1.1. The Danish s-genitive has the same rules as Swedish, meaning that a sentence like: 'Generalens fejl' [the fault of the general] (p. 36), is written with an s- genitive, as this is an alienable relationship. The English translation used a preposition construction with the preposition 'of'. It becomes apparent that the Danish s-genitive is usually not used for partitives, with the exception of a single usage: 'Afrejsens stund' [The moment of departure] (p. 68).

39 Due to the high number of genitives, Danish correlates 76.1% with German. It also correlates for 86.7% with Icelandic, making it the highest correlation for the Icelandic translation. Even compared to the other Germanic languages, Danish has a correlation above 73.3% (with Fering). This is probably due to the fact that Danish still uses many preposition constructions (52) and uses nearly all available possessive constructions.

3.6 Languages without a genitive case-marking

3.6.1 Afrikaans

Afrikaans has no remnants of the genitive case in the language and is therefore one of the languages that had no possessive construction with a genitive in the results. Most of the possessive NPS in the Afrikaans translation were conducted with a preposition construction (79). That amount was the same as the preposition constructions in the Dutch translation. They were also quite similar in their use of zero-marking construction (0.4 in relative numbers). In the case of the Afrikaans translation this was mostly done when writing a partitive construction. For example: ''n hele trop olifante' [a whole heard of elephants] (p. 20). Partitives that contained numbers, were mostly written with a zero marking construction: 'Vyfhonderd miljoen sterre' [five hundred million stars] (p. 45). This was also carried out in the Dutch translation, where most partitives based on numbers received a zero- marking construction. The personal pronoun construction is found by far the most frequently in Afrikaans (15 out of 29 in total). The over-representation of the personal pronoun construction came to a relative number of 12.4. These constructions were formed with se and were mostly used when the possessor was a person (and two times when it was an object). Examples of this construction are: 'Klein prinsie se enkel [The littl prince's ankle] (p. 60), 'My klein beoordelaar se gesig' [The face of my young judge] (p. 8) and 'Planeet se grond' [The soil of the planet] (p. 21). Roberge (1996) made a comparison between the English s-genitive and the personal pronoun construction in Dutch and Afrikaans. As the se-construction and the s-genitive seem to be compatible (Norde 1997, p. 61), a comparison could be made between the following languages: Dutch, English, Danish and German. As the Dutch results did not have the z'n-construction, this comparison was based on their shared history. English was mainly chosen for the same reason and also because of the recent language contact between Afrikaans and English. Danish had the most s- genitives and German contained the most actual genitives. In the following table it becomes clear which constructions these languages used when Afrikaans used a personal pronoun construction:

40 The genitive case is mostly used when Afrikaans uses a personal pronoun construction. German only uses the genitive case, whilst Danish and English also use different possessive constructions. However, the Danish results show that of the 15 tokens, 11 were conducted with an s-genitive. The English translation mostly used a preposition construction as well as 4 s-genitives (of the 6 s- genitives that were found in the English translation). Dutch stands out for using no genitive cases, as the only genitive found in the Dutch translation was a remnant of the old-genitive case. No s- genitives were found and when Afrikaans used a personal pronoun construction, in Dutch this was mostly written using a preposition construction. This table does not explain whether the se-construction has similar properties as the (s-)genitive in the other Germanic languages, but does show that it often occurs when the other languages use a (deformed) genitive. Afrikaans has a high correlation with most other languages. With English, Dutch, Luxembourgish and Faroese this was even above 90%. The lowest correlations are again with Icelandic (52.9%) and German (25.6%) probably due to their differences in morphology.

3.6.2 Fering

The Fering translation of Le Petit Prince has no tokens with a morphological case to express the possessive relationship. In terms of over-representation, Fering contains a high number of subordinate clauses compared to the other languages (with a relative number of 1.8). An example of this relatively free translation would be: 'Eerd, wat ütj granit bestäänt' [This earth made out of granite] (p. 60). It shows that the possessive relationship was explained by adding a subordinate clause, something that was described in the background information when considering Elfdalian. Another free translation strategy that was relatively often use was to rewrite the possessive construction into a VP. An example of this is the following sentence: 'Hü hiart ham sin stem uun?'

41 [What is the sound of his ?] (p. 19). The possessive construction is transcribed by making it into an active phrase. Fering correlates on a high level with Dutch, Faroese, Luxembourgish and to a lesser extent with the other languages. The lowest correlations are again with German (19.1%) and Icelandic (52.2%). The low correlation with German is in fact the lowest in the correlation table of the general results. This can be explained by the different distribution of not only the genitive but also with the other constructions: compounds (DE: 37, FE: 18), Case markings (DE: 62, FE: 0), prepositions (DE: 32, FE: 63), non possessives (DE: 18, FE: 33), VPs (DE: 17, FE: 32) and sub clauses (DE: 0, FE: 18). Comparing these results, it shows that there is a difference in direct and free translations of both languages.

42 4. Discussion

4.1 Discussing results

The chapter on background information gives an overview on the kind of possessive constructions that can be used and to what extent they are used in the Germanic languages. It also gives an insight into the existence of the genitive in a language and what form it takes in existing language forms. In the upcoming paragraphs I will discuss the background information in relation to the results that were found.

4.1.1 German

In the methodology a strong relationship between German and Icelandic was hypothesized due to their morphological case-system and the remaining genitive case. They might be subject to different rules, as the Icelandic genitive is used differently in partitives, but this does not give an insight into the (lower) correlation between the languages. The results showed a high over-representation of the genitive case in the translation of Le Petit Prince. Out of the 158 genitives in the translations, 62 were found in German (39.2%). In the German translation itself this holds for 32.5% of the possessive constructions. The Icelandic translation contained less genitives, with 35 tokens (18.2% of 191). When comparing the two genitives, with the German genitive as a 'filter' (paragraph 3.3), 25 of the Icelandic genitives were used at the same time as the German genitive. This is a total amount of 71.4% of the Icelandic genitives (25 out of 35), and 40.3% of the constructions used by the Icelandic translator where the used a genitive case (62 German genitives (100%) and 25 Icelandic genitives (40.3%)). Other constructions often used when German used a genitive, were the preposition construction (13) and compounds (11). Due to these results, my earlier hypothesis that German and Icelandic are similar in their use of the genitive case, is now redundant. The German translation contained nearly twice as many genitive cases than the Icelandic translation. Because only 25 Icelandic genitives were used in the translation (when one would have expected 30-35), whereas the German text used 62 cases of genitive, one can conclude that the Icelandic translation shows a different approach to the use of genitive cases. Danish and Swedish showed a relatively high usage of the s-genitive. One could state that the use of the genitive case is rather different in these languages in comparison to the German genitive. 22 of the 29 Danish genitives (75.9%) occurred at the same instance as the German

43 genitive. 15 of the 21 Swedish genitives (71.4%) were used when the German translation had a genitive case-marking. Because only 22/15 Danish/Swedish genitives were used in the translation (when one would have expected 25/18), whereas the German text used 62 cases of the genitive, one can conclude that also Danish and Swedish have a different approach considering the genitive case- marking in comparison to German. In German, the other most used constructions are compounds and preposition constructions. It is not surprising to see that German has the lowest amount of preposition constructions compared to the other Germanic languages, mainly due to the higher amount of genitive cases. This is probably the reason why German correlates the least with other Germanic languages. Only with Icelandic, Danish and Swedish is the correlation higher than 62.1% and that is likely to be due to the use of the genitive case. The Icelandic and German genitives do show overlap, although they are not always used in the same instances. The Danish and Swedish s-genitives overlap with the German genitive, showing that it often still acts like an old genitive (although their usage are not completely the same in comparison to the German genitive). It seems that in German, the loss of the genitive case in Germanic languages, has not (yet) occurred. On the contrary, German contains by far the most genitive cases and a decline of genitives seems unlikely. The German translation did contain other ways to express a possessive relationship (prepositions, compounding, zero marking) or indirect translations (non-possessives, verb phrases, personal pronoun constructions) but these seem to co-exist with the German genitive.

4.1.2 Icelandic

As previously described in the background material, Icelandic maintains the original genitive case. Icelandic is known to be a conservative language and this could potentially lead one to consider that Icelandic can be used as a template of how languages can express a possessive relationship. It transpires that Icelandic uses all possible constructions (as do Elfdalian and Faroese), but it contains half as many genitive cases as the German translation. As the genitive case is only fully used in German and Icelandic, I hypothesize that these languages show similarities in this area. It turns out that the Icelandic translation not only uses fewer genitive cases than the German translation, but that those cases in nearly 30% of their occurrences have a different purpose. The relationship between the two languages therefore exists but can be explained not only by the use of the genitive case but also by the usage of other possessive constructions. Both Icelandic and German show an infrequent use of preposition constructions (43 and 32 tokens), but contain many compounds (47 and 37). The correlation

44 between these two languages is 79.5%, which is lower than the correlation between Icelandic and Danish (86.7%) and Swedish (83.2%). The high correlation between these languages can be explained in the use of the (s-)genitives (35 in Icelandic, 29 in Danish, 22 in Swedish). That all three languages descend from Old Norse, seems to be a sustainable hypothesis in explaining their high correlations. The background information already suggested that the use of zero-marking constructions is more restricted in Icelandic. When it is a pseudo-partitive genitive, zero-marking is, according to Sigurðsson, not allowed (where other Germanic languages (except for English) can use this construction). Instead, a preposition construction is required (see paragraph 1.3 for examples). This could explain the low amount of zero-marking constructions found in the Icelandic translation (7 tokens). The other Germanic languages express their partitives most often with a zero-marking construction. For example:

IS: Tveir milljarðar ibúa > Genitive (p. 51) GB: Two million inhabitants > Zero-marking (p 55)

This partitive was expressed by a genitive case in Icelandic, while in most other Germanic translation it was translated with a zero-marking constructions, as shown in the English example. In terms of similarities in the expression of possessive relationships between Germanic languages, Icelandic correlates mostly with German and the Scandinavian languages.

4.1.3 English

As explained by Feist (2012), the s-genitive is a rather unrestricted way of expressing a possessive relationship, as it can be added to the end of an NP. This is not the case for the s-genitive used in Dutch (see Weerman & de Wit (1999)), and therefore a difference in use is to be expected between these two languages. The s-genitive in Swedish and Danish seems to have more in common with the English s-genitive; although in an alienable relationship, an s-genitive is mandatory in Swedish and Danish. I therefore expected more s-genitives in Swedish and Danish, in comparison to the English translation. The expectation that the s-genitive is used differently in English and Dutch turns out to be difficult to prove. The s-genitive was absent in the Dutch data (the one genitive found was not an s- ) and only found 6 tokens were found in the English translation. A comparison between the two could therefore not be made in terms of the s-genitive. The English and Dutch

45 translation showed a high correlation (84.9%) and this is because of their similarities in their use of personal pronoun constructions (0 in GB, 2 in NL), dative construction (0), non-translated (0), subordinate clauses (2 in GB, 0 in NL), genitives (6 in GB, 1 in NL) and verb phrases (14). Their correlation lowers because of the fewer numbers of preposition constructions in Dutch (79 tokens), twice as many compounds (31 in NL and 15 in GB) and the higher number of non-possessives and zero markings. These differences might be due to the fact that zero-marking constructions are more restricted in English as well as the fact that compounds are less often used (Feist 2012, p. 276). In terms of the loss of the genitive case, this means that Dutch and English still have a rather similar approach (due to the correlation of 84.9%), but do differ in this instance. The second expectation that there would be a discrepancy in s-genitives in English, Danish and Swedish proved to be correct. With only 6 genitives in English, and 29 and 21 in Danish and Swedish, the differences are even higher than anticipated. As there is so much research available on the English s-genitive (of which a few are reviewed in the background material), more was expected. Instead, the preposition construction was by far the most used possessive construction. The higher amount of s-genitives in Swedish and Danish can be explained by the grammatical rule that makes the use of an s-genitive mandatory with an alienable construction (a preposition construction is then not allowed). Nevertheless, I found in the data that this could not be the only explanation, as there were preposition constructions found in alienable constructions:

DE: Den historie med tigerkløerne > Preposition construction (p. 31) EN: This tale of the claws > Preposition construction (p. 30)

According to the rules as presented in the background information, one would expect a preposition construction in English but an s-genitive in Danish: 'Tigerkløerne's historie'. However, it seems that the preposition construction is a more direct and accurate translation of 'Histoire de griffes' (p.42), and therefore an s-genitive was not used in this alienable construction. This was the case in many other alienable constructions both in Danish and in Swedish. The over-representation of s-genitives in Danish and Swedish in comparison to English can therefore not be completely ascribed to this certain grammatical rule. Still, a difference in usage and meaning can be hypothesized and would be an interesting topic for further research. For now, the discrepancy of s-genitives between English, Danish and Swedish turns out to be correct but the underlying reason for this remains unknown. It seems that in the English translation, the preposition construction (often with the preposition 'of') is an alternative strategy in the loss of the genitive case. I feel supported with this claim due to the fact that when the uses a preposition construction, German,

46 Icelandic, Danish and Swedish (all languages with the highest amount of (s-)genitives) all use at least 71.4% of the time a genitive case (see table: 11). When German uses the genitive case (62 in total), English uses the preposition construction (52 in total) with an overlap of 83.9%. Because most languages do use many preposition constructions (except for German and Icelandic), a high correlation between English and the other Germanic languages appears to be a logical outcome.

4.1.4 Dutch

A few general ideas on the similarities between Dutch, English, Afrikaans and Swedish were given in the background information. The s-genitive in Dutch is not a group genitive as in Swedish and English (its use is more restricted); in colloquial Dutch one can expect to find the z'n-construction, similar to the personal pronoun construction with se in Afrikaans, and the van-construction is similar to the of-construction in English. It turns out to be hard to isolate data that would support some of these claims, as the Dutch data lacks the personal pronoun construction with z'n and there were no s-genitives used in the translation. The one genitive found was a relic from the old genitive case-marking: 'Vier uur 's middags' [four o'clock in the afternoon] (p. 48). The original form would have been 'des middags', though in Modern Dutch the article has been shortened to 's. The van-construction was relatively over-represented in the use of preposition constructions. Out of the 79 preposition constructions, 66 (83.5%) were conducted with the preposition van. In comparison to the English data, 127 preposition constructions were used, of which 114 (89.8%) were produced with the preposition of. With this their similarity might not be proven, though their frequency in use is nearly the same. The third hypothesis, stating that a higher correlation between Dutch, Afrikaans and Luxembourgish would be expected, proved correct. Still, the expectation was mainly used on the use of personal pronoun constructions with z'n, which were not found in the Dutch translation. Their correlations are: 92.5 (NL-AF) and 98.8 (NL-LU). The translators in these languages often made a similar choice in how to translate the possessive construction in French, with only a few larger differences. The biggest difference between Dutch and Afrikaans is the use of the personal pronoun constructions (15 in AF, 2 in NL) and non-possessives (11 in AF, 35 in NL). There are barely any large differences between Dutch and Luxembourgish, although Luxembourgish has more subordinate clauses and non-translations (where the Dutch translation has none). The translators' choices therefore explain the high correlation between the languages. An additional explanation for the high correlation between Dutch and Afrikaans is their shared history and the fact that Afrikaans descends from colonial Dutch.

47 4.1.5 Luxembourgish

Little information was found on possessive constructions in Luxembourgish. The background information showed that Luxembourgish holds some remnants of the genitive case, has a preposition construction often conducted with vun and that there were signs of a personal pronoun construction as found in Afrikaans, conducted with säi. Indeed, Luxembourgish showed only one old form of the genitive case: 'Mueres fréi' [sunrise] (p. 80) and most often used a preposition construction (71 tokens). With the high amount of preposition constructions and lack of genitives, Luxembourgish follows a similar pattern as most some other Germanic languages, such as English, Afrikaans, Dutch, Faroese and Fering. The personal pronoun constructions that would support a higher correlation between Luxembourgish, Afrikaans and Dutch were only found five times in the translations, of which three have the personal pronoun säi:

LU: Generol säi Feeler > Personal pronoun (p. 37) AF: Generaal se skuld nie > Personal pronoun (p. 36) 'The fault of the general' LU: Geograph säi Planéit > Personal pronoun (p. 53) AF: Aardrykskundige se planeet > Personal pronoun (p. 53) 'The geographer's planet' LU: Klenge Prënz säi Knéchel > Personal pronoun (p. 60) AF: Kleine prinsie se enkel > Personal pronoun (p. 60) 'The little prince's ankle'

As the number of personal pronoun constructions is too few, no conclusions can be drawn from these results, although the examples show that the personal pronoun constructions in Luxembourgish and Afrikaans look similar. A comparison cannot be made between these constructions and the Dutch examples, as they lack the similar z'n-construction. As described in the previous paragraph, the high correlation with Dutch (98.8%) (and Afrikaans (95.2%)) cannot be explained by the use of personal pronoun constructions, but mainly due to the very similar translation choices made in these texts. It would be interesting to research further similarities between these languages, especially because language contact cannot explain the high correlation due to their geographical distance. Further research could elucidate how these languages correlate on different language properties.

48 The low correlation between Luxembourgish and German (28.4%) is striking due to the trilingual state of . Residents speak French, German and Luxembourgish and the language contact is therefore undeniable. Still, in terms of possessive relationships, Luxembourgish and German have very little in common. It would be interesting to see if there are more similarities in other language areas, including phonology and lexicon. Luxembourgish has a high correlation with both Fering (93.5%) and Faroese (97.3%). With these languages, language contact can be excluded as a possible cause. Paragraph 4.2, translational decisions might shed some light on these unexpected relationships.

4.1.6 Faroese

Together with Elfdalian, Faroese no longer uses the genitive case, but has developed a construction which is not an s-genitive. In the case of Faroese, this is the so-called sa-. It shares quite a number of properties with the s-genitive but is rather restricted in its use, as it can only be used for kinship terms, proper nouns and names. The genitive case is only found in fossilized forms in the written language (Petersen 2016, p 5-6). Other possibilities to express a possessive relationship are found in Faroese , where it is clear that the preposition construction is often used to express a possessive relationship. In comparison to the data of the translation of Le Petit Prince, the following outcomes were found: 1. Preposition constructions were the main elements used for expressing a possessive relationship (67), 2. Only two instances of the genitive case were found, 3. The dative case was also used to express a possessive relationship (as in Elfdalian). Of the 67 preposition constructions, only 8 were constructed with the preposition 'hjá'. Even though the background information suggested that this preposition was most often used in general, this was not found in the Faroese translation. Instead, probably based on the semantics of the possessive relationship, 10 different prepositions were used: hjá, av, eftir, frá, millum, sum, um, við, á and í. With 31 tokens, 'frá' was mostly used to form a possessive NP. One would have expected the s-genitive to be written with the sa-clitic. It turns out that the only two genitives were written with either the double s-genitive ('Koningsins'), or with the actual genitive case: 'Táranna land' (p. 25). Other tokens with a case-marking were, surprisingly, not written with an accusative case, as been described in the background information. Instead, the translator used the dative case. For example: 'Korninum fyri' [The colour of the wheat fields'] (p. 65). Faroese shows a high correlation with all Germanic languages, except for German (29.5%)

49 and to lesser extent, Icelandic (68.4%). The high correlation with other languages can be explained due to the fact that Faroese has many preposition constructions and barely any case-markings as in most other languages. The correlation with Danish (81.0%) cannot be explained based on the genitive but may be due to the fact that Danish is the of the . That the correlation is not higher can again be explained by the use of the s-genitive in Danish.

4.1.7 Swedish and Danish

Both in Swedish and Danish, the rules concerning preposition constructions and s-genitives are different to the other Germanic languages. As the background material shows, the s-genitive is a group genitive (as in English), but is mandatory in certain situations. When a possessive construction is alienable, an s-genitive needs to be used (while English can use a preposition construction). Based on this information, more s-genitives were expected in Danish and Swedish than in the English translation. The distinction based on alienability was not found in the data. Out of the 52 preposition constructions used in Danish, 37 occurred when the relationship was alienable. Further research might show whether other factors play a role in the preferred use of the s-genitive with an alienable relationship. The expected difference between English, Swedish and Danish still comes from the use of the s-genitive and the lesser use of preposition constructions. The discrepancy was even higher than expected, mostly due to the severe lack of s-genitives in the English translation. Danish turns out to have 8 more s-genitives than Swedish; but in comparison with the other Germanic languages, these two languages show more similarities with Icelandic and, to lesser extent, German. The correlation with Icelandic is 83.2% (SE) and 86.7% (DK), and with German 62.1% (SE) and 76.1% (DK). It appears that the high correlation between Swedish and Danish (97.5%) shows that they are syntactically (or at least when concerning possessive NPs) very similar. I expected such a similarity between Danish and Norwegian, which could not be tested, but a larger gap between Swedish and Danish. Still, this high correlation makes it undeniable that the possessive NPs are similarly expressed in Danish and Swedish. In both languages, the loss of the genitive case seems to be partly covered by the deformed genitive (21 in SE, 29 in DK), or by the use of compounds (37 in SE, 35 in DK), preposition constructions (63 in SE, 52 in DK) and zero marking constructions (27 in SE, 26 in DK). The translators also followed more or less the same pattern in their translational choices, when choosing to use a free translation, meaning that there were few non-translations (3 in SE, 2 in DK) and subordinate clauses (3 in SE, 4 in DK).

50 Apart from their high correlation to each other, both languages show a high correlation with all tested languages. The correlations with Danish range as widely as 75.3% (with Elfdalian) to 97.5% (with Swedish). The correlations with Swedish range between 62.1% (with German) to 97.5% (with Danish). All other Germanic languages do at least have lower correlation with German; or in the case of Icelandic, with Fering (52.2%). Danish and Swedish show no extraordinary alternatives with the loss of the genitive case in comparison to the other Germanic languages, meaning that they find themselves somewhere in the 'middle' of the general results.

4.1.8 Elfdalian

Several researches were conducted in order to find possibilities for expressing a possessive relationship in Elfdalian. It showed that the genitive was often replaced with the dative case or a combination of the dative case, plus an es-suffix (deformed genitive). Other possibilities found by Dahl (2016), Eekman (2014), and Lundqvist (2011) were preposition constructions, compounding, subordinate clauses, zero-marking or paraphrasing (either subordinate clauses or changing the sentence). I expected to find occurrences of these constructions in the Elfdalian data. Elfdalian, together with Faroese and Icelandic, displayed all possible constructions in this research, meaning that the expected constructions were found, together with non-possessives, verb phrases, non-translations and personal pronoun constructions. Due to the research done, amongst other researchers, by Dahl, I expected to find more deformed genitives, but only found a single example: 'Tårumes land' [The land of tears] (p. 30). Rather, the dative case was used more often (8 tokens). Still, with only 9 case-markings, it became clear that the case-system was not preferred when expressing a possessive relationship in the Efldalian translation of Le Petit Prince. The translator mostly used preposition constructions (63), compounds (40), non-possessives (37) and verb phrases (26). With this, a relatively high amount of the original constructions were paraphrased (37.1%). The high correlation between Faroese and Elfdalian (95.3%) is striking. Language contact between these two is ruled out due to their geographical positions. Both languages are quite isolated and they originate from the same language, Old Norse. Somehow they must have either followed the same 'path' in dealing with the loss of the genitive case, or had translators who made similar translational decisions. It remains important to note that both languages can replace the genitive case with a different case, whether this is dative or accusative case. The disparity between Swedish and Elfdalian I supposed to be similar to that between Swedish and Icelandic (Dahl 2005, p. 40). Interestingly enough this proved true when comparing

51 their correlations. Between Swedish and Elfdalian there is a correlation of 83.3% and between Swedish and Icelandic 83.2%. This mainly shows that the language distance does not appear to be as big as anticipated, considering the possessive NPs. Even the correlation between Icelandic and Elfdalian is still relatively high with 72.0%. Elfdalian has several alternative strategies which cover the loss of the genitive case. Again, the preposition construction is mostly preferred, followed by compounds. In a certain way, Elfdalian has a rather similar approach to possessives NPs as with most other Germanic languages.

4.1.9 Afrikaans

The background information could not provide a clear vision of the possible possessive constructions in Afrikaans. According to Roberge (1996), Afrikaans has a preposition construction like the van-construction in Dutch and the of-construction in English, but is considered to be less common than the personal pronoun construction with se. This se-construction is supposed to be less restricted in use than the English s-genitive, as it not only behaves as a group genitive but can even be attached to a complete subordinate clause. One would therefore expect that the se-construction would be more common than the s-genitive, and that Afrikaans uses the preposition construction less frequently than English and Dutch. The first expectation proves to be correct. Where only 6 s-genitives are used in English translation, the translation to Afrikaans contains 15 se-constructions. The constructions in Afrikaans were not conducted with complete subordinate clauses. This can be due to the fact that the text is from a children's book, meaning that complex sentences were not preferable or hardly found in the original French text. The differences between the s-genitive in English and the se-construction in Afrikaans, can therefore not be supported by this data. The second expectation, that Afrikaans uses less preposition constructions, turns out to be true when compared to the English translation (127 in GB, 79 in AF). The Dutch translation used as many preposition construction as the Afrikaans translation, which shows that in this text, there seems to be no big discrepancy in the usage of the van-construction. Afrikaans has a high correlation with many languages. The relationship with Dutch (92.5%) and English (94.8%), can be explained by the fact that the language descends from the Dutch language and has a close language contact with English. It is harder to explain the high correlation with Faroese (90.4%) and Luxembourgish (95.2%), as they do not have the same language contact. The only explanation that can be given, is that similar translation decisions were made by the translators of these languages. They also seem to have similar alternative strategies to compensate

52 for the loss of the genitive case, although with both similarities and differences. The se-construction in Afrikaans seems, together with preposition constructions, to be a new strategy to replace the genitive case. The construction shows similarities with the genitive as used in German. 15 se-constructions in Afrikaans were used at the same time as the genitive case in German, as for example:

AF: Kleine prinsie se planeet > Personal Pronoun Construction (p. 20) DE: Dem Planeten des kleinen Prinzen > Genitive case (p. 21) 'The little prince's planet' The se-construction seems to behave relatively the same as the s-genitive in Danish (11 occurred when Danish had a genitive case), the preposition construction in English (9 when English had a preposition construction), and Dutch (12 when Dutch had a preposition construction).

4.1.10 Fering

The research on possessive NPs in Fering seems to be fairly limited. What I did find, showed that the genitive case was lost in Fering and therefore nouns sometimes contain instances of an s- or e- ending. Other possibilities were not discussed in the background information. No signs of a case-system were found in the translation to Fering. It mostly used preposition construction (63), similarly to the other Germanic languages which barely contained case-markings. There was, in comparison to the other languages, a low number of compounds (18). Only English appears to have used the compound strategy less frequently (15) than Fering. With that, and the use of other direct translations, it seems that Fering shows a different tendency through the translation of Le Petit Prince. The number of free translations is 45.0%, and can, for example, be found in the amount of subordinate clauses. Fering used this strategy 18 times and this is the highest amount of subordinate clauses in this research. An attempt to explain why Fering has so many indirect translations will be discussed in the next paragraph. (4.2:Translational decisions.) Fering shows a high correlation with all languages except Danish, Icelandic and German. With Danish, there is still a percentage of 73.3%, but this is much lower already in Icelandic (52.2%). The use of the genitive case can play a role in this, as well as the differences in translational decisions. The lowest correlation of the correlation table is between Fering and German, with only 19.1%. The distance between German and Fering, based on possessive constructions, is outstanding as they are also neighbouring languages. It has to do with the completely different approach that these languages seem to have in expressing possessive NPs. It

53 would be interesting to see if their discrepancy is the same when looking into other languages aspects.

4.2 Translational decisions

The research is based on translations of Le Petit Prince, originally written in French. The book was a useful choice, as it is translated into many of the Germanic languages and into most that were of interest for this research (it was written in Nynorsk but this was no longer accessible). In addition, the book is written for readers of all ages (but is published as a children's book), which means that the language is easier to understood by a non-native speaker with the help of a dictionary. However, as a research tool, other (unexpected) difficulties come into play. First of all, due to the fact of it being a children's book, the language has been adapted in order to make it understandable for young children. That is a translation choice which is specifically popular and/or necessary when translating children's literature (Coillie, van 2005, p. 22-23). This creates the expectation that certain grammatical structures, which are hard for a child to understand, are simplified or freely translated in order to increase the accessibility of a text. Furthermore, when using a free translation, the issue of importance is no longer to stay as close as possible to the original text (foreignisation (p. 22)), but to conduct a translation for the greatest possible benefit of the reading public. I must admit that this can be a point of critique in this research, although it is necessary to note that the original text is also a children's book. The original French text is conceived without subsequent adaption, and translational changes for the purpose of understanding are no longer a factor. The second side note concerning translation choices follows on from the first. As it is a translated text, the translator constantly has to choose whether to make a translation that is faithful to the original text (with the possible outcome that the text is harder to read, as the grammatical constructions might not be as popular in the translated language), or a translation that makes concessions in order for it to be more easily understood in the new language. These choices are recognizable in the translation of possessive NPs in the attested Germanic languages. As a distinction was made between direct (preposition constructions, compounding, zero markings, personal pronoun constructions, genitives and datives) and free translations (VPs, non-possessives and not translated), it is easy to see how many tokens per language were freely translated. Constructions written with subordinate clauses are also taken into account, as they are (nearly) completely rewritten possessive NPs. The following overview shows the number of free translations per Germanic language:

54 Table 13: translational decisions FE FA EL LU IS NL DK SE AF DE GB TOTAL: No possessive 33 28 27 28 20 35 17 14 11 18 8 239 VP 32 23 26 18 21 14 26 27 18 17 14 236 Sub clause 18 13 5 8 12 0 4 3 6 0 2 71 Not translated 3 4 3 7 3 0 2 3 5 0 0 30 TOTAL: 86 68 61 61 56 49 49 47 40 35 24 576

Percentage 45 36 32 32 29 26 26 25 21 18 13 27

The largest group of free translations is non-possessives, followed by verb phrases. Subordinate clauses and not-translated constructions were used less often. It is important to note that with all but the not-translated constructions, the possessive constructions were still recognizable in the translation, but had a different relationship to the text. When, for example, a preposition construction in French was translated, it sometimes led to a suitable translation that was only one noun in the target language (meaning that the translation was accurate, but a possessive construction was no longer possible). In the translations there were nearly as many VPs as non-possessives. In these VPS, the actual possessive relationship was recognizable but was transformed into an active phrase. This translational choice is understandable in the sense that it would make the translation more fluent, specifically in the case that a literal translation would not result in a logical sentence in the translated language. For example, the possessive construction in French: 'Le son de sa voix' [the sound of his voice], is translated into a VP in Swedish: 'Hur låter hans röst?' ['What does his voice sound like?'] (p. 8). There were few not-translated constructions. These constructions are no longer recognizable in the translated text, and can only be identified as such by looking at the original French text. It emerged that of all the used constructions, 27% were free translations. The translations to Fering (45%) and Faroese (36%) contained the most free translations, and the translation to English had the smallest amount with only 13%. Also Elfdalian and Luxembourgish showed a high amount of free translations with 32%. That two of the aforementioned languages are in fact dialects (Fering and Elfdalian), makes this outcome interesting. With the very small amount of speakers (10.000 for Fering and 2.400 for Elfdalian (Simons & Charles 2017)), and the highly endangered nature of the languages, it can be possible that they needed a freer translation in order to express the same meaning from French to Fering/Elfdalian. This might be the case also with Faroese and Luxembourgish, as both languages have a relatively few number of speakers (345.000-415.000 for

55 LU, and 48.000-69.000 for FA). These languages are not as highly developed as the other Germanic languages and therefore need more creativity in their translations. This would not necessarily explain the high percentage found in Icelandic (29%), although this language has also a relatively low number of speakers and is rather isolated as a language. The small amount of free constructions used in the English translation can be explained by the fact that this language has by far the most speakers of the Germanic languages. The language is more highly developed and does not need as much free translation in order to express the possessive construction in the French text. An additional research that could be carried out (in order to omit the translation issues, as well as the limitations of using children's literature), is one which would record spontaneous speech and therefore the use of possessive constructions. As this research would both be time consuming and highly expensive, one would need a large budget and a considerable amount of time to be able to conduct such a study. Most probably this would again result in a research that only takes into account a few languages, though with the use of spontaneous speech it could add valuable results to the current linguistic field.

56 Conclusion

This research was carried out in order to give a general analysis of the expression of possessive noun phrase in Germanic languages. The motivation for this topic was the loss of the genitive case in most of these languages and how possession could be expressed without this morphologic case. Previous research on this topic was mainly conducted from a diachronic perspective where languages like Luxembourgish and Frisian did not have a valid place. In many instances, researchers focussed only on a single language group (e.g. the Scandinavian languages). The only research, which provided the general idea of possible constructions with a synchronic scope which compares two languages, is the research of Lundqvist, who analysed a translation from Swedish to Elfdalian. In order to perform a synchronic research on this scale, many aspects needed to be organized. The data had to be similar for all languages, leading to Le Petit Prince as a case study, as this book has been translated into the required languages and would lead to a set amount of tokens. The choice of this book was not without drawbacks (as explained in 4.2), leading to the adding of free translations to the possible constructions used in translating a possessive NP. By using this book, I was forced to omit Nynorsk from this research, as it was no longer accessible. Before carrying out the research, three hypotheses were constructed concerning the expected relationship between the languages. Other instances, as for example the expectation that the z'n- construction in Dutch would be similar in use to the s-genitive in English, were not taken into account due to the impossibility of predicting an outcome in which languages would prefer a certain possessive NP-construction. The first hypothesis, that Scandinavian mainland languages (Swedish and Danish), would have a close relationship, turned out to be correct. This correlation was nearly the highest of this research, with 97.5%. The number of tokens per possessive NP were similar, with only a few more s-genitives in Danish and a few more preposition constructions in Swedish. The second hypothesis, where the genitive preserving languages (German and Icelandic), would show a high correlation, did not entirely prove correct. It is indeed the case that both languages maintained the genitive case but as it turned out, they did not have a similar usage. The German translation contained the most genitives cases, and the Icelandic genitives were not all used in the same instances as the German genitive. The relationship between these languages still holds for a correlation of 79.5% but is mainly based on the fact that they use this case-system, than on the fact that these case-markings are handled in a similar fashion. The third hypothesis expressed the expectation that Dutch and Afrikaans would correlate on a high level. A predictable reason for this would have been the personal pronoun construction,

57 which was already explored in the background information. The same construction was also described to be used in Luxembourgish. It turned out that the high correlation between Dutch and Afrikaans was visible with 92.5% and also with Luxembourgish 95.2%. The reason for this correlation did not come from the use of the personal pronoun constructions, which was nearly only used in Afrikaans. The translators seemed to have made similar decisions in translating possessive NPs. As explained in the methodology, any over-representations of used possessive constructions were analysed separately. These proved to be the preposition constructions in English, personal pronoun constructions in Afrikaans, the dative case in Elfdalian and the genitive case in German. It became especially clear that the preposition construction in English was used mainly when a genitive case was used in German, showing that English substitute the loss of the genitive case by using preposition constructions. But what are the solutions in other Germanic languages? Most languages showed a preference for the preposition construction. It seems that in the loss of the genitive case, this is by far the most widely used strategy for expressing a possessive relationship. Compounding is used in all of the languages but is not popular in English and Fering. Zero marking, especially popular in Dutch and Afrikaans, turned out to be scarce in both Elfdalian an Icelandic. Many translators used the subordinate clause but it was mostly used in Icelandic, Faroese and Fering. The personal pronoun construction was mainly used in Afrikaans. The (s-)genitive was often used in Danish, Swedish, Icelandic and German, though German had by far the highest over-representation. The following chart shows the preferred strategies for expressing the possessive NPs in French in the translations:

58 The concluding factor would be that nearly all the Germanic languages correlated highly because of the use of preposition constructions to express a possessive relationship. The languages which used fewer of the preposition constructions, correlated on the use of the (s-)genitive (German, Icelandic, Swedish and Danish). All Germanic languages analysed in this research, except for Icelandic and German, show a correlation of at least 75.3% with each other. Icelandic correlates for at least 52.2% with other languages and is much further removed from the mainland languages, than for example Faroese. German shows a singular pattern, with the highest correlation of 79.2% and the lowest of 19.1%. German is the evident exception in conducting possessives in the Germanic languages. As already stated, it would be interesting to conduct a research based on spontaneous speech in all these languages. In this way, translation would no longer be a factor and it would give an idea of the present expression of possessive relationships in a spoken language. For now, this research is based on written language and shows that most Germanic languages, while sometimes being very different, often use the same strategy in the loss of the genitive case and in building their own system in expressing a possessive relationship.

59 Bibliography

Allen, C. H. (2003). Deflexion and the development of the genitive in English. English Language and Linguistics, 7(1), 1-28. Bohn, O. (2004). How to organize a fairly large vowel inventory: The of Fering (North Frisian). Journal of the International Phonetic Association, 34(2), 161-173. Bril, I. (2012). Ownership, part–whole, and other possessive–associative relations in Nêlêmwa (New Caledonia). In A. Aikhenvald, & R. . Dixon (Eds.), Possession and ownership: A crosslinguistic typology [Explorations in Linguistic Typology 6] (pp. 65-89). Oxford: . Dahl, . (2005). Att sätta Älvdalskan på kartan, 31. Dahl, Ö. (2015). In Haspelmath M. (Ed.), Grammaticalization of the north: Noun phrase morphosyntax in Scandinavian (Studies in Diversity Linguistics 6 ed.). Berlin: Language Science Press. Dyk, S., & Heyen, H. (2017). Taalportaal: The linguistics of Dutch, Frisian and Afrikaans online. Retrieved 08/11, 2017, from http://www.taalportaal.org/taalportaal/topic/pid/topic- 14085362002594797 Feist, J. (2012). What constrols the "genitive variation" in present-day English? Studies in Language, 36(2), 261-299. Gerling, V. E. (2015). Sieben auf einen Streich! - Welchen Prinzen hätten sie denn Gerne? Relü, 15/07/2017. Gorter, D. (1991). Het Fries en het Nederlands. enkele aspecten van de taal, de taalsituatie en de taalstudie. Brünner Beiträge Zur Germanistik Und Nordistik, 13, 111-123. Hoekstra, J. (2002). Genitive compounds in Frisian as lexical phrases. Journal of Comparative Germanic Linguistics, 6, 227-259. Koptjevskaja-Tamm, M. (2003). In Plank F. (Ed.), Noun phrase structure in the . Berlin, New : Mouton de Gruyter. Lawless, L. K. (2017). French possession, learn the different ways to express possession in French. Retrieved 07/12, 2017, from https://www.thoughtco.com/g00/french-possession-1368906? i10c.referrer=https%3A%2F%2Fwww.google.nl%2F. Lundqvist, U. (2011). Återgivning av svensk s-genitiv i en översättning till Älvdalska. In G. Nyström, & Y. Sapir (Eds.), Rapport från oðer råðstemną um övdalskų, andra konferensen om älvdalska (pp. 41-51). Uppsala: Institutionen för nordiska språk.

60 Norde, M. (1999). The history of the genitive in Swedish. A case study in degrammaticalization. (Doctoral, Vakgroep Skandinavische taal- en letterkunde, Amsterdam, 1997). Nordic Journal of Linguistics, 22 (1), 77-77-90. Perridon, H. (2012). The emergence of the s-genitive in Danish. Language Sciences, 36, 134-146. Petersen, H. P. The common origin of the phrasal clitic –sa and the forms of personal pronoun hansara ‘his’, hennara ‘her’, okkara ‘our’, tykkara ‘your’ in Faroese. Unpublished manuscript. Petersen, H. P. The spread of the phrasal clitic sa in Faroese. Unpublished manuscript. Pittner, K. (2014). Ist der Dativ dem Genitiv sein Tod? – Funktionen und Konkurrenzformen von Genitiv-NPs im heutigen Deutsch. In C. Reuter, & A. Schlief (Eds.), Linguistische und Sprachdidaktische Aspekte Germanistischer Forschung Chinesisch-Deutsch (pp. 41-56). Frankfurt: Lang. Roberge, P. T. (1996). Multilevel adn the evolution of Afrikaans perphrastic possessives with SE. Papers in Linguistics, 30, 123-153. Saint-Exupéry, d., A. (1945). The little prince [Le Petit Prince] (K. Woods Trans.). (2009th ed.). London: Egmont. Saint-Exupéry, d., A. (1951). De kleine prins [Le Petit Prince] (de Beaufort van Hamel, L. Trans.). (2012th ed.). Rotterdam: A.D. Donker. Saint-Exupéry, d., A. (1988). Litli prinsinn [Le Petit Prince] (Þ. Björnsson Trans.). Reykjavík: Bókaútgáfa Menningarsjóðs. Saint-Exupéry, d., A. (1994). Die klein prinsie [Le Petit Prince] (A. P. Brink Trans.). (2015th ed.). Kaapstad: Tafelberg. Saint-Exupéry, d., A. (2007). Le petit prince. Paris: Gallimard. Saint-Exupéry, d., A. (2009). De klenge prënz [Le Petit Prince] (J. Braun Trans.). (2009th ed.). : Editions Phi. Saint-Exupéry, d., A. (2013). Den lille prins [Le Petit Prince] (1950th ed.). Egmont: Lindhardt og Rinhof Forlag A/S. Saint-Exupéry, d., A. (2015). Der kleine prinz [Le Petit Prince] (M. Herbert Trans.). (2015th ed.). Köln: Anaconda. Saint-Exupéry, d., A. (2015). Den lille prinsen [Le Petit Prince] (N. Gullberg Zetterland, H. Petersen Trans.). (2015th ed.). Stockholm: Modernista. Saint-Exupéry, d., A. (2015). Lisslprinsn [Le Petit Prince] (B. Westling Trans.). (2007th ed.). Neckarsteinach: Edition Tintenfaß.

61 Saint-Exupéry, d., A. (2015). Tann lítli prinsurin [Le Petit Prince] (A. Kristiansen Trans.). (2013th ed.). Fuglafjørður: Egið Forlag. Saint-Exupéry, d., A. (2016). De letj prens [Le Petit Prince] (P. J. Trans.). (2010th ed.). Neckarsteinach: Edition Tintenfaß. Scott, A. K. (2011). The position of the genitive in present-day Dutch. Word Structure, 4(1), 104- 135. Sigurðsson, H. Á. (2006). The Icelandic noun phrase, central traits. Arkiv För Nordisk Filologi, 121, 193-236. Simons, G. F., & Charles, D. F. (Eds.). (2017). : Languages of the (20th ed.). Dallas, Texas: SIL International. Thráinsson, H., Petersen, H. P., Jacobson, J., & Hansen, Z. S. (2004). Faroese, an overview and reference grammar. Tórshavn: Føroya Fróðskaparfelag. Torp, A. (2004). Nordiska språk i forntid och nutid. In I. Stampe Sletten (Ed.), Nordens språk med rötter och fötter (pp. 19). Köpenhamn: Nordiska ministerrådet. Weerman, F. & Wit, de, P. (1991). The decline of the genitive in dutch. Linguistics, 37(6), 1155- 1192.

62