• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Genitive compounds

Next to endocentric, exocentric and coordinate compounds, Frisian possesses another pattern of NN-compounds which perhaps resembles most the endocentric garden variety, but which also differs in important respects. Phonologically, the stress within such genitive compounds, as they have been dubbed, is not on the first, but rather on the second member of the compound. In addition, in many cases we find a linking element, primarily -s-. Thus, next to the common endocentric NN compound KEAmerdoar we have a genitive compound keamer-s-DOAR. Semantically, there is a part-whole relation between the second and the first member. Moreover, the first member is always definite/specific, and this also renders the whole compound definite/specific. Thus a KEAmerdoar room door may be any door of any room, but in keamersDOAR a specific door of a specific room is meant. In practice, this will be the only door affording entrance to the living room of a house. One effect of this definiteness condition is that pluralizing is difficult. The productivity of genitive compounds is limited.

The pattern is exemplified below (linking element in boldface):

Table 1
Example Gloss Translation
kokensflier kitchen-LK-floor floor of the kitchen
bedsrâne bed-LK-edge edge of the bed
kroadetsjil barrow-wheel wheel of the barrow
jasmouwe coat- sleeve sleeve of the coat
fingerslid finger-LK-lid bone in one's finger, phalanx
tsjerkhôfsstek churchyard-LK-fence fence of the churchyard
spoarbrêge railway-bridge railway-bridge
[+]Input restrictions

Formally, the input of the compound consists of two nouns. The first member may be complex. For instance, this may consist of a garden variety endocentric NN-compound, as in bargehoksDOAR, with the left-hand member BARGEhok < baarch pig + hok pen. The first member may also consist of a genitive compound itself. So from hok shed + doar door we get hoksDOAR. Combined with kaai key, this results in hoksdoarsKAAI. Semantically, both members of the genitive compound should be concrete nouns. They refer to non-animate objects in the direct physical environment of the speaker. See also below.

[+]Phonological properties

An important feature of genitive compounds is that the stress is always on the second member. With complex left-hand members of the compound, the stress is culminating, with reduction of the syllable that received the main stress in the compound that was one phase less complex. So in hoksdoar we have the pattern hoksDOAR. The stress on the part doar is then reduced in hoksdoarsKAAI.

The fact that the second member receives main stress may lead to several reductions in the first member. One is that nouns ending in schwa sometimes lose that final segment, cf. skoalPLEIN school garden (< skoalle school) or skuorREED wagon road to the barn (< skuorre barn). The vowel of the first member may reduce to schwa, as in spoarBRÊGE [spəbrɛ:gə] railway bridge. A combination of final schwa deletion and reduction of the full vowel can be found in molier [mə'li.ər] millrun, from mole [mo:lə] mill plus ier run. In this compound also a resyllabification has taken place, l of mole now being part of the second syllable. We also see resyllabification in the place name Wâldsein wood-s-end [və'zain], which, apart from the regular historical deletion of l, shows intervocalic voicing of de remaining consonant cluster ts > dz, and further simplification to z.


Genitive compounds fairly regularly show a linking element-s-: hoksDOAR door of the shed, skelksBÛSE pocket of the apron, kloksGAT hole of the bell; sound hole, and truisMOUWE sleeve of the pullover are some examples. For phonological reasons, no audible linking phoneme can be observed if the first member itself has a final [s], nor is a linking phoneme spelled out in these cases. So we have jasBÛSE pocket of the coat and húsSIDE side of the house, but an underlying -s might be assumed in these cases as well.

A real category of exceptions, possibly also for phonological reasons, is supplied by compounds with an initial s in the first member, cf. seeDYK sea dike, skoarstienMANTEL mantelpiece or spoarBRÊGE railway bridge. These formations show no linking at all.

Another conspicuous category that is devoid of a linking -s is supplied by those first member nouns that end in a schwa: tsjerkeTOER tower of the church, húskeRÚTSJE window of the toilet, seageBLÊD blade of the saw, etc. Nevertheless, a linking phoneme might be assumed in such cases. This would then be -e, though this schwa would not be audible either, since for phonological reasons the two adjacent schwas clash.

The element -s is probably a residue of an old genitive ending -s. That strong genitive ending had a weak cognate -e. It is reasonable to assume that this ending might have turned into a linking phoneme, too. This is supported by the behaviour of the first member nouns ending in schwa. If such nouns are part of an endocentric NN-compound, so having stress on the first member, we frequently see that the final schwa is truncated. Thus the noun tsjerke church is or may be reduced to tsjerk- in compounds like TSJERKhôf church yard and TSJERK(e)toer church tower. In the genitive compound tsjerk*(e)TOER tower of the church, however, deletion of the schwa is impossible. Also, if the noun lid lid is the second member of a compound, then this is usually a genitive compound. Examples with a linking -s are baksLID lid of a cistern or tsjettelsLID lid of the kettle. Furthermore we have kisteLID lid of the coffin, tsjerneLID lid of the milk churn and kanneLID lid of the milk-jar. The schwa in these compounds may not be omitted. Compare this to PANlid pan lid, an ordinary NN compound, which shows no schwa. Hence, it may be assumed that the schwa occurring at the end of the first member of a genitive compound has a special function and is more then just the schwa ending of the first member noun. It should be noted in this respect that (Hoekstra 2002), the main study on this type of compounds, analyses these elements -s and -e as still being genitive endings synchronically.

[+]Morphological potential

Genitive compounds can only be the input for further word formation if they constitute as the left hand member of a more complex genitive compound. Thus hoksDOAR door of the shed may be combined with kaai key to hoksdoarsKAAI key of the door of the shed. Genitive compounds are not allowed as the left-hand member of endocentric NN-compounds. Hence, ?hoksDOARplanke is odd. This must be due to the definiteness of the genitive compound.

In general, genitive compounds are used in the singular, since they refer to a unique object in the direct physical environment of the speaker. Pluralization is not excluded on principal, however. In a house with a kitchen that has two doors, one could say de kokensDOARREN steane iepen the kitchen doors are open, with the plural doarren of the singular doar. Even more, some genitive compounds only occur in the plural if in daily reality one never encounters the referred object on its own. Examples are foarkeTINEN teeth of the fork, baitsjeKNOPEN buttons of the smock and mantelsBEAMMEN trees of the tree circle around the yard. These examples all show the plural suffix -en. Here, conversely, singularization is odd. So, if one of the teeth of a pitchfork is broken, a farmer will look for a new FOARKEtine, that is, in the form of a normal endocentric NN-compound, and not for a new *foarkeTINE.


Within composition, normal endocentric NN-compounds form a productive category and, moreover, they are not specified for definiteness/specificity. However, the definite use of a normal NN compound is blocked in those cases in which a cognate genitive compound is available. So, if one can say de kokensDOAR is ticht the door of the kitchen is closed, then the comparable NN compound is not available: *de KOKENdoar is ticht.


In accordance with the restrictions with respect to plurality, the property of definiteness or specificity accomplishes that genitive compounds may not occur in indefinite contexts. Therefore, de kokensDOAR the door of the kitchen is fine, but *in kokensDOAR a door of the kitchen is not. Another effect is that adjectives or relative clauses may be added, but only if these do not have a restrictive character. Thus a brown-coloured kitchen door may be referred to as de kokensDOAR, dy't brún is the door of the kitchen, which is brown. On the other hand, *de kokensDOAR dy't brún is is impossible, since this implies that there are more kitchen doors.


Genitive compounds display a restricted productivity. One reason is the limitations resulting from the semantic restriction of definiteness/specificity. One also gets the impression that genitive compounds are connected to a world where everything had its own fixed place and role. Modern concepts are seldom involved in the pattern, although they are not excluded, for instance in garaazjeDOAR the door of the garage. However, a compound like ?tillefoansHOARN the receiver of the telephone sounds odd. Furthermore, the number of members of the category may be diminishing under influence of Dutch, which lacks the pattern of genitive compounds.


Genitive compounds display a part-whole relation: the second member denotes a part of the first member. In many cases, this part is inalienably related to the whole. They always refer to concrete objects in the direct physical environment of the speaker. Here is an overview of the most prominent sub-categories:

a. Parts of the human body
      fingersEINEN tips of the fingers
      fingersLID bone of the finger
      hiersEINEN ends of the hair
b. Parts of (daily) clothes
      broeksBÛSE pocket of the trousers
      jasMOUWE sleeve of the coat
      baitsjeKNOPEN bottons of the smock
      skelksGALGEN strings of the apron
c. Parts of furniture
      taffelsBLÊD top of the table
      bedsRÂNE edge of the bed
      kloksWICHTEN weights of the clock
      kammenetsLAAD drawer of the cabinet
d. Parts of tools and gear
      kroadeTSJIL wheel of the barrow
      loddeHJELT hilt of the shovel
      beitelsBEK edge of the chisel
      amersBOAIEM bottom of the bucket
e. Parts of wel-known buildings, especially one's own house and annexes
      keamersFLIER floor of the living room
      húskeDOAR door of the toilet
      keldersTREP stairs of the cellar
      skuorreNAAL ridge of the barn
      bûthúsSOUDER loft of the cow-house
      skoalleDAK roof of the school
      tsjerkeTOER tower of the church
f. Parts of the immediate surroundings of the house and well-known buildings
      hiemsHIKKE gate of the yard
      tsjerkhôfsSTEK fence of (around) the churchyard
      mantelsBEAMMEN trees of the tree circle around a yard
      steechDOAR door of the narrow passage between two houses
g. Parts of natural objects or construction works
      grêftsWÂL bank of the canal
      dyksFOET foot of the dike
      spoarBRÊGE railway-bridge
      sylsDOAR gate of the lock
      weisKANT side of the road

Sometimes, the relation is primarily locative, as in hurdsHERNE corner where the fire place is situated, mûnePÔLE mound on which the mill is situated or spoarDYK dike on which the railway is situated. Nevertheless, the semantic relation between the two members of a genitive compound is more restricted than it is in normal NN-compounds, in which it is basically free.

Genitive compounds are special in that the first member has referential properties. In koken kitchen in kokensDOAR one or the other arbitrary kitchen or a kitchen in general is not meant, but rather some specific kitchen, notably the kitchen in one's own house. And the whole compound kokensDOAR refers to a specific door in that kitchen, notably the only door that provides entrance to this kitchen. This semantic property has a few consequences: for instance, the indefinite article is not possible:

*Hy seach in keamersDOAR iepen stean
he saw a room-s-door open stand
He saw a door of the room being open

Furthermore, as has also been mentioned above, pluralization of a singular genitive compound is not possible. The feature of definiteness/specificity also obstructs reference to kinds:

*Brune kokensDOARREN mei ik net lije
Brown doors-of-the-kitchen may I not suffer
I do not like brown kitchen doors
[+]Historical note

Genitive compounds probably originated in prenominal genitive constructions, i.e. putative Old Frisian [des kokens]doar; cf. the Old Frisian mention des scholappers hws the shoemaker's house. In early Middle Frisian, inflection of articles was lost. This resulted in putative [de kokens]doar , with phrasal stress on doar; a Middle Frisian record of this pattern is dy bâems virtil the tree's root. At a certain point, this must have been reanalysed to [de[kokensdoar]]. Koken and doar merged into a compound bearing word stress on the part doar. Moreover, the choice of the article no longer depended on the first noun, but rather on the second, as this was now the head of the compound. Thus common gender de koken and neuter it rút result in the genitive compound it kokensRÚT. This historical development may explain the unusual stress pattern (main stress on the second member), the existence of the linking element and, semantically, the definiteness/specificity of the first member. (For the remnants of the genitive in present-day Frisian, see the topic on case).

The pattern must have become productive to a certain extent, since nowadays it is used with respect to concepts that did not exist in Old or Middle Frisian, i.e. garaazjeDOAR ( the door of the garage). Nevertheless, it is under pressure. Apart from the fact that the dominating Dutch language lacks genitive compounds, there are alternatives in Frisian itself as well, both in syntax and morphology. The genitive compound kokensDOAR may equally well be expressed by the syntactic expression the doar fan de koken ( the door of the kitchen) or by the common endocentric NN-compound KOKENdoar ( kitchen door).

Another development is that the genitive compound itself can be involved into a transition towards a normal NN-compound. Firstly, the condition of definiteness/specificity is lost, as in the following example:

Muontsen leine seeDIKEN oan
monks laid seaDIKES on
Monks built sea dikes

A next step is a shift of accent towards the first member, witness formations like daksgoate roof gutter or kielsgat gullet (lit. throat hole), with stress on the first member, but, on the other hand, with an otherwise unusual linking element -s-. Hence, one might infer that such forms acted as genitive compounds at an earlier stage. The final step, then, may be the loss of the linking element itself. An example is the word skouderblêden shoulder blades, which has an older attestation as skoudersbledden.

[+]Geographical note

Apart from West Frisian, genitive compounds have been reported to exist in Frisian varieties in Germany as well. More concretely, instances have been found in the North Frisian dialects of Heligoland and Föhr/Amrum. The now extinct East Frisian dialect of the island of Wangerooge also displayed genitive compounds. Furthermore, the pattern can be found in coastal Dutch dialects with a Frisian substratum, i.e. the dialects of Groningen and so-called West-Friesland. In these dialects a linking element is often lacking or only shows up in combination with a restricted number of second members.

Genitive compounds do not exist in Dutch, although Dutch has some NN-compounds with final stress. However, either these formations act as names (RijksWATERSTAAT Department of Public Works; staatsBOSBEHEER Forestry Commission), or they are not inherently definite (cf. een rijksAMBTENAAR a civil servant).


The main source for genitive compounds is Hoekstra (2002), which also gives a wealth of data. Other examples can be found in Schippers (1967). Synchronically, Hoekstra analyses these compounds as lexical phrases, to be more precise as genitive agreement phrases, the first member being a DP, which would account for the referential properties. A good description of genitive compounds may also be found in (Hoekstra 1998:52-55). Delfitto and Paradisi (2009) propose an analysis along syntactic lines; according to them, this could account for the definite character of the first member of Frisian genitive compounds.

  • Delfitto, Denis & Paradisi, Paola2009Prepositionless genitive and N+N compounding in (Old) French and ItalianTorck, D. & Wetzels, W.L. (eds.)Romance language and linguistic theory 2006Romance language and linguistic theory 2006Amsterdam-Philadelphia53-72
  • Hoekstra, Jarich1998Fryske wurdfoarmingLjouwertFryske Akademy
  • Hoekstra, Jarich2002Genitive Compounds in Frisian as Lexical PhrasesThe Journal of Comparative Germanic Linguistics6227-259
  • Hoekstra, Jarich2002Genitive Compounds in Frisian as Lexical PhrasesThe Journal of Comparative Germanic Linguistics6227-259
  • Schippers, Hendrik K1967De saneamde binings-s - in nuversmaDe Pompeblêdden: tydskrift foar Fryske stúdzje3750-54
printreport errorcite