• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
Noun incorporation

Frisian has the ability to form complex verbs consisting of a noun and a verb, of the format [NV](V). The noun is an argument of the verb, usually the Patient. Given this, it is natural that there should also be a parallel syntactic construction in which such a noun is the head of an NP figuring as an argument of the verb. Compare

a. Wy wolle de messen slypje
we want the knives sharpen
We want to sharpen the knives
b. Wy wolle messeslypje
We want knife-sharpen
We want to sharpen knives

When we compare these sentences, we see that in the spelling of (1b) the object is concatenated to the verb, without a space separating the two. This reflects the standard feature of Frisian orthography of writing compounds as one word. This, of course, is insufficient to conclude that the object makes up one lexical unit with the verb here: there are more serious differences to be observed. In (1b) the article de is missing, which might be an indication that the object noun is no longer part of a phrasal NP. Another difference can be found in the translation, which is a reflection of a change in meaning.

Most important, however, is the form of the object in (1b): messe- [[mεsə]. This form is different from both the singular, which is mes [mεs], or in northern and western dialects with a long vowel, spelled mês, and from the plural, which is always messen [mεsən], the second syllable often syllabified to [mεsņ]. Significantly, the form messe- only occurs in compounds, for instance in messebak knife-tray or messeheld knife-hero knife fighter.

That the form messe- cannot occur on its own can best be shown when we strand the object in sentences with Verb Second. With its West Germanic relatives Dutch and German, Frisian is an SOV language, having the object canonically in a position to the left of the verb. This can already be seen in (1a) , where the main verb slypje is to the right of the object de messen. The order also shows up in embedded sentences:

Hja sizze dat wy [(NP) de messen] [(V)slypje]
they say that we the knives sharpen
They say that we sharpen the knives

The order is reversed in root sentences: the verb is moved to the second position in the sentence:

a. Wy slypje de messen
we sharpen the knives
b. *Wy de messen slypje
we the knives sharpen

With the object incorporated, this pattern is reversed:

a. *Wy slypje messe
b. Wy messeslypje

Example (4a) shows that the form messe cannot be stranded. Hence it may be concluded that it does not constitute a syntactic phrase by itself, in contrast to de messen in (3a). Example (4b) shows, in addition, that messe must have been moved along with the verb. Since only a single verb may be fronted, it follows that messe must be part of the (now complex) verb. One can thus conclude that the structure of this verb must be something like [[messe](N)[slypje](V)](V). Moreover, it appears that the pattern is very productive.

To introduce this topic, an overview will be provided of morphological, phonological and syntactic evidence, which at the same time strengthens the claim that Frisian possesses noun incorporation. This is followed by sections on the paradigmatic distribution of incorporational forms, on the appearance of linking elements and on semantic matters.

[+]Morphological evidence

Incorporated nouns can appear in a form which can only be found in compounds, not when the noun heads a noun phrase. An example is messe- in, for instance, messeslypje. The form messe- ends in a schwa. This schwa is a linking element, also frequently showing up in nominal compounds. Taking a reverse stand, if it really were the case that the noun was the head of an NP, one would expect that it could be pluralized. But this is not possible. Frisian has two productive plural morphemes, -en and -s. When incorporated, the nouns with a plural -s always show up in their singular form:

Heit jerappel / *jerappels dolt de hiele dei
father potato / potatoes digs the whole day
Our father is digging potatoes all day long

Nouns taking -en as their plural morpheme may appear as singular or with the linking element -e-.The plural morpheme -en itself, however, is prohibited in incorporation:

Heit byt / bite / *biten wjuddet de hiele dei
father beet / beet-LK / beet.PL weeds the whole day
Our father is weeding beets all day long

Additionally, the incorporated noun cannot be accompanied by any determiners:

De buorlju sieten bûten te *de / *dy / *sokke wyndrinken
The neighbours sat outdoors to the / that / such / wine-drink

Similarly, modifiers are out:

* Heit sit te grouwe jerappelskilen
father sits to huge potato-peel
Our father is sitting, peeling huge potatoes
* Heit sit te jerappel mei in soad spruten skilen
Father sits to potato with a lot of sprouts peel
Father is sitting, peeling potatoes with a lot of sprouts

If the noun were the head of a phrase, these restrictions would not have been applicable; the possibility to pluralize or take determiners or modifiers is just what could be expected then. These exclusions are readily explained, however, when we assume that the incorporated noun is a head N which cannot be projected higher up.

If, on the other hand, the incorporated noun were part of a word, one might expect that the complex word obeys Williams' (1981) Right-Hand Head Rule(Williams 1981). This is indeed borne out both categorically and semantically. Jerappelskile potato-peel, for instance, is a verb, as is its right-hand member skile peel, and jerappelskile remains a kind of peeling, although it is only applied to potatoes.

[+]Morphological potential

Another indication that the NV combinations discussed here can be assigned word status is their possibility to be the input of further derivational morphology, though restrictions apply. In the area of composition, for instance, verbs are not involved on a large scale in Frisian. The only productive candidates for compounding with a right-hand complex verb would be either a combination with a left-hand preposition, or a recursive application of NV-compounding itself. The latter would result in structures like [N[NV](V)](V). But it appears that such cases of "double incorporation" simply do not exist. For a possible explanation, see Dyk (1997:154-156).

On the other hand, a combination with a particle, resulting in a particle verb, although it is not very common, is certainly not prohibited, as the following examples may show (sources in Dyk (1997:17)):

Moast mar oanprikkebarne
should.2.SG PTCL on-dry.stick-burn
You should intensify the burning of dry sticks
Lucht, dy't wy ynsykhelje
air, that we in-breath-take
Air, that we breathe
Nou kinne wy mar tabrijite
now can we PTCL to-porridge-eat
Now we have plenty of porridge to eat

Furthermore, complex verbs may appear as the left-hand member of a compound. An example is jerappelskyldersmeske potato-peel-LK-knife.DIM [[[[jerappel](N)[skyl](V)](V)]ers][mes-DIM](N)](N) potato peeler, where the element -ers- is a linking element.

As input for further word formation, NV-formations also encounter serious restrictions in the area of derivation, especially with respect to suffixation. Of the suffixes attachable to a verb, several candidates are excluded beforehand. One reason can be that the potential suffix binds a theme argument. An example is the suffix -sel. So, from the verb bakke to bake one can form the noun baksel, which refers to the output of the baking process. But from an incorporation verb like koekjebakke cookie-bake it is not possible to derive *koekjebaksel, since the theme argument would be expressed twice, once by the incorporated noun, and second by the suffix itself. Another category in which derivation from incorporation verbs is blocked is suffixes which predicate over the theme argument. One example is adjectivizing -ber. So, from steapelje to stack one can form the adjective steapelber stackable. Thus, an object-verb combination like kisten steapelje boxes stack to stack boxes is related to the predicative adjective in dy kisten binne steapelber those boxes are stackable and the attributive adjective in the word group de steapelbere kisten the stackable boxes. However, in an incorporation verb such as kistesteapelje this operation would result in an ungrammatical *kistesteapelber. The theme argument, needed for the subject position or the head of the modified NP, is not available in this case because of incorporation.

Among the suffixes which could take an incorporation verb as input, those referring to Agents and Events, such as the suffix -er and the suffix -erij, are the best candidates. And indeed, kistesteapelder box-stacker or kistesteapelderij box-stacking can easily be derived. For the point at issue, even these are not conclusive examples, since at first sight they could also be analyzed as having a right-branching structure [[N][[V][suffix]](N)](N), i.e. the issue whether or not these formations should be analyzed as synthetic compounds.

So, if we want to find out whether incorporation verbs can be the input for derivation, it is safer to look at prefixes. Here further derivation is certainly not excluded. For instance, we can find formations with the nominalizing prefix ge-, such as gehúshimmel PREF-house-clean, geboatsjefar PREF-boat-sail, gebrieveskriuw PREF-letter-write or gekofjedrink (PREF-coffee-drink. With the prefix fer-, we can find in the comprehensive dictionary WFT (Veen 1984-2011) such examples as ferhânbûtse PREF-hand-beat hurt oneself by beating one's hands about one's body, in order to create some warmth, ferkaartspylje PREF-card-play waste by playing cards or reflexive fersûpedrinke PREF-buttermilk-drink damage oneself by drinking too much buttermilk. Also the prefix be- could be used, for instance in

Wat bepopketekenest de hiele tiid?
what PREF-figurine-draw-2.sg the whole time?
What the hell are you drawing figurines all the time for?

This overview is certainly not meant to be an exhaustive account of all the cases in which incorporation verbs can or cannot function as input for further derivational morphology. But it should make clear that such an operation is not excluded in principle. This in itself could be one of the indications that the NV-formation is a word, and that the incorporated noun is a part of it.

[+]Phonological evidence

From a phonological point of view there are also indications that the incorporated noun builds one word with the verb. One is that Frisian long vowels may undergo shortening. This can occur when a member of a compound is added to the stem. Examples are aai [a:j] egg, which shortens to [aj] in aisykje egg-seek to look for eggs, or pipe [pi:pə] 'pipe which becomes [pip] in pypsmoke pipe-smoke to smoke a pipe. The form pypsmoke also shows another relevant phonological clue since it appears that the final schwa of the noun has been dropped. This truncation may occur when a noun acts as a first member of a compound.

A similar argument can be drawn from breaking. An example is stien [sti.ən] stone, which becomes stiennen [stjInən] in the plural and stientsje [stjIntsjə] in the diminutive, or stienkrobbe [stjIηkrobə] stone-beetle wood louse in a nominal compound. Now, this broken form also surfaces in an incorporation verb like stienbikje [stjImbIkjə] stone-chip to chip stones. Breaking and shortening display many similarities. One essential correspondence is that the instigating factor, loosely to be formulated as a syllable added to the right of the stem, must be contained within the same word as the diphthong or broken or shortened vowel. This condition is met when it is assumed that the noun undergoing the process builds a compound with the triggering verb. When this is not the case, the stem of the noun remains unchanged:

in stien bikje
* [(NP) in stien (= [stjIn])] bikje
a stone chip
to chip a stone

So we see that some phonological processes also prove that the incorporating noun is building one word with the verb.

A difference in pronunciation between the two constructions also manifests itself in the area of stress, although it must be admitted that this needs a subtle ear, caused by the fact that whenever an object noun functions as a head of a phrase or as incorporated, the noun receives the main stress in both cases, not the verb. But there is a difference, though. When we compare the combination of a verb and a (bare) phrasal object as wetter drinke [(VP)[(NP) wetter] [(V) drinke]] water drink to drink water with an incorporational construction wetterdrinke [[wetter](N) [drinke](V)](V) water-drink, it strikes the ear of the attentive hearer that the verbal part seems to be a little more prominent in the case of the phrasal construction. In other words, with incorporation, the accentual contrast between the nominal and the verbal part is more distinct.

[+]Syntactic evidence

Taking a syntactic perspective does not alter this perspective. If the incorporated noun and the verb do build one complex word, it is to be expected that they are subject to the principle of Lexical Integrity, which states that parts of words are invisible to syntactic rules and principles. This appears to be borne out by the facts, for instance with respect to movement rules. The incorporating verbal head itself may not be moved, for instance to a Verb Second position (the line marks the canonical position of the moved constituent):

a. ... dat wy messeslypje
that we knife-sharpen
that we sharpen knifes
b. *Wy slypje messe__
we sharpen knife

So, neither the verbal head nor the incorporated noun may be moved out of the complex verb. Phrasal objects meet no difficulties at this point. This can be shown for instance by topicalization:

a. [De messen] wolle wy hjoed ___ slypje
the knives want we today sharpen
It are the knives that we want to sharpen today
b. *Messe- wolle wy hjoed ___slypje
knife want we today sharpen

We see the same effect with scrambling:

a. Wy wolle de messen hjoed ___ slypje
we want the knives today sharpen
We want to sharpen the knives today
b. *Wy wolle messe hjoed ___slypje
we want knife today sharpen

Furthermore, it is not possible to relativize the incorporated noun:

a. Wy wolle de messen slypje, dy't stomp wurden binne
we want the knives sharpen, which dull became have
We want to sharpen the knives which have been gotten dull
b. *Wy wolle messeslypje, dy't stomp wurden binne

One more conspicuous argument is the shape of negation. In Frisian, the general verbal negative is the adverb net not. Noun phrases may be negated by prenominal gjin no:

a. Ik haw net slipe
I have not sharpened
b. Ik haw net in mês slipe
I have not a knife sharpened
I have not sharpened a knife
c. Ik haw gjin mês slipe
I have no knife sharpened
'I did not sharpen any knife
d. Ik haw gjin messen slipe
I have no knives sharpened
'I did not sharpen any knives

We see that a phrasal object, both in the singular and the plural, may be negated by gjin no. Now, the striking thing is that negation by gjin is excluded when the object is incorporated:

*Ik haw gjin messeslipe
I have no knife-sharpened

The only remaining negator is net not:

Ik haw net messeslipe
I have not knife-sharpened
'I have not been sharpening knives

From this pattern we can conclude that in (20) there is no NP to be negated. The only available nominal element is contained in the complex verb, hence leaving net as the only candidate for negation.

The existence of noun incorporation in Frisian

The precious sections will have made it clear that Frisian features a productive process of noun incorporation. This is remarkable in the light of the fact that in other Germanic languages it is lacking. The question then emerges why this should be the case, and why Frisian has this extra possibility to express a relationship between a verb and for example its object, beyond the usual encoding in syntax. A hypothetic answer is offered in chapter 5 of Dyk (1997). He states that the basis for an explanation should be found in the fact that Frisian possesses two infinitival endings, i.e. -e and -en, also an exclusive property of the language. The latter ending has nominal possibilities, and can therefore easily enter into a nominal compound with another noun. As the ending -en is verbal at the same time, the combination could then easily be reanalyzed as a verbal compound.


As the sections above have made clear, Frisian possesses a construction in which a noun and a verb are concatenated in such a way that together they build a complex verb. The zero option, then, would be that this verb can be inserted in all the positions where verbs usually show up, and by and large this is indeed the case. Nevertheless, it can be observed that in actual usage noun incorporation has certainly a greater preference for one context than for the other. It appears that the preference is related to an inherent semantic feature of noun incorporation, i.e. that it has to fit in durative aspect.

One of the environments where noun incorporation is not uncommon is constituted by the typical Frisian to-infinitives. J. Hoekstra, in Hoekstra (1989) and Hoekstra (1992), divides te-infinitives into four types: verbal, adjectival, prepositional and sentential. The last two appear to be absent in related West Germanic languages like Dutch and German. The most outstanding property of prepositional and sentential te-infinitives is that elements like objects, some adjectives and particles may not occur outside the frame [te ... V]. These types are subsumed under the heading incorporating to-infinitives in the syntactic part of Taalportaal. Noun incorporation in this construction is relatively frequent. Here is an example:

Ruerd siet te biezembinen op 'e bank
Ruerd sat to beson-bind on the bench
Ruerd sat on the bench, binding (a) beson(s)s

At least two reasons can be put forward why noun incorporation has such a prominent place in the context of te-infinitives. The first is that the construction demands a durative aspect. The other is that the verb, as De Haan (1987) discovered, cannot be expanded by for instance phrasal complements or adverbials, at least not between the element te and the verb itself:

* Ruerd siet te [(NP) in biezem] binen
Ruerd sat to a beson bind

Only verbs are allowed to the right of te to Hence, if by any means, an object needs to be expressed after te, one is obliged to incorporate it.

The oan it-construction is reasonably comparable to the to-infinitive. It invokes a durative aspect as well, and again no separate words are allowed between oan it and the verb. But incorporation verbs do not suffer from this restriction:

Hy is oan it biggefangen
he is ON IT piglet-catch
He is catching piglets

Formally, the oan it-construction consists of a preposition oan and a nominalized infinitive. Noun incorporation also occurs in other nominalizations with the suffix -en:

It is op 't lêst gjin botpoeren hwat se dogge
It is at the end no flounder-sniggle what they do
Eventually, they do something else than sniggling flounders

Incorporation can also be observed in participles:

Der waard hânklapt en foetstampt
there was hand-clapped and foot-stamped
there was hand-clapping and foot-stamping

There are, however, also areas where noun incorporation meets more obstacles. These occur in a construction in which the complex verb is finite or appears in an (om) te-infinitive. As to finiteness, it can certainly not be claimed that the process is prohibited, but compared to an ordinary phrasal object it seems that a sentence with an incorporated object is a little strange:

a. Hja bakt mei nocht [(NP) in bôle]
she bakes with pleasure a loaf
'She bakes a loaf with pleasure
b. ? Hja bôlebakt mei nocht
she loaf-bakes with pleasure
She is loaf-baking with pleasure

As such, the question mark in (27b) suggests a too heavy load. Its only function is to indicate that, compared to the (a)-sentence, there is a lesser acceptability, which can certainly not be ascribed to the Verb Second effect:

? ... dat se mei nocht bôlebakt
... that-she with pleasure loaf-bakes

Finite incorporation sentences can be patched up considerably by adding a durative adverbial, however:

Hja bôlebakt al jierren mei nocht
she loaf-bakes already years with pleasure
She has been baking loaves already for many years with pleasure

The probable reason for this behaviour is durativity again. That the formation of sentences with finite forms of incorporation verbs meets some difficulties can also be distilled from their rather low frequency in running texts. Yet, it is certainly not the case that this use is excluded on principle.

Since it looks as if the potential difficulties with finite forms can be overcome, the problems are becoming even more serious in the case of (om) te-infinitives of the verbal or adjectival type, to maintain the distinction of (Hoekstra 1989) and (Hoekstra 1992). These are often (but not always) accompanied by the complementizer om. Curiously enough, incorporation is not allowed in such infinitives. There is a remarkable distinction between (om) te-infinitives and prepositional and sentential te-infinitives, the latter on the contrary being a favourite context for incorporation, as has been shown above. The pattern can be nicely illustrated with the help of the verb helpe to help, which can take both kinds of infinitive complements:

a. Hja helpt my [te bushimmeljen]
she helps me to can-clean
She helps me cleaning cans
b. * Hja helpt my [om te bushimmeljen]

In an (om) te-infinitive, the same message can only be expressed by placing the object in front of the obligatory element te:

Hja helpt my [om bussen te himmeljen]

So, here we have a context in which noun incorporation is not allowed. Note that these facts of noun incorporation verbs resemble those of particle verbs in Frisian and Dutch. Take as an example opblaze blow up, where in (om) te-infinitives the particle is also separated from the verb by the interfering te:

a. ... om it op te blazen
b. * ... om it te opblazen

Therefore, however strange the impossibility of noun incorporation in the context of (om) te-infinitives may seem at first sight, it may be assumed that this restriction can be accounted for on independent grounds. Hence, we conclude that, in principle, noun incorporation can be used in all verbal positions.

[+]Linking elements

When, in Frisian noun incorporation, a noun is combined with a verb to build a new complex verb, the simplest solution, as far as the outer form is concerned, would be to just add the noun stem. This is indeed what happens in a lot of cases, as in rúthimmelje window-clean to clean the windows or apelite apple-eat.

However, the position between the noun and the verb can also be filled by a linking schwa, as in messeslypje knife-sharpen. This linking element -e- also figures in Frisian nominal compounding. There we see another linking phoneme, viz. -s-. This element does not show up in the context of noun incorporation, however.

The linking schwa cannot turn up in every instance of noun incorporation. The major restriction is that the incorporated noun itself should have -en as its plural suffix. The plural of mês knife, for example, is messen. There is one exception, as skiep sheep with its irregular plural form skiep may also invoke a linking -e-, for instance in skieppeskeare sheep-LK-shave to shave sheep. On the other hand, nouns that select the alternative plural suffix -s never show a linking element if they are involved in noun incorporation. Something like *apele-ite does not exist.

Generally speaking, it is not the case that potential -en-plurals obligatorily take a linking element; they can also be incorporated without it, for example as blomsnije flower-cut (next to blommesnije or messlypje knife-sharpen and messeslypje).

Furthermore, there is a large class of nouns for which it cannot be decided whether they show a linking element or not. These nouns, all selecting a plural suffix -en, have a final schwa of their own in the singular. An example is tûkeseagje branch-saw, with the noun tûke branch, which has tûken as plural form.

The restriction that linking -e- only occurs with nouns taking a plural suffix -en suggests that there is some connection with plurality itself. This is confirmed by the fact that mass nouns, which cannot be pluralized in principle, never carry a linking -e-. The linking element -e- induces a plural interpretation. In

Ik sil hjoed stekkefervje
I will today fence-LK-paint
Today, I will paint fences

the minimum number of fences I plan to paint is two. On the other hand, in

Ik sil hjoed stekfervje

the number of fences could also be one (but could also be more).

Apart from -e-, another linking element can be recognized in the field of noun incorporation. This is the diminutive suffix, which may also have such a role in nominal compounding. Examples are beltsjedrukke chime-DIM-push and knibbeltsjefrije knee-DIM-pet. Apart from these it is also possible that real diminutives are incorporated. An example is blomkenaaie flower-DIM-sew in embroidering textile, in which small flowers are referred to.


In comparison to non-incorporated structures, we notice that noun incorporation produces some effects on meaning. These concern the incorporated noun itself and the resulting complex verb.

First the incorporated noun. This becomes non-referential, that is, it cannot refer in a deictic sense; the incorporated noun can only be interpreted as generic. This property has been observed in other incorporating languages as well. It would explain the fact that the incorporated noun cannot be accompanied by determiners and that proper nouns or personal pronouns cannot incorporate. To give an example, in the sentence

Heit sit te jerappelskilen
Father sits to potato-peel
'My father is peeling potatoes

no reference is made to any particular potato(es), and indeed, not a single determiner, such as a definite or indefinite article or a demonstrative pronoun, is allowed in this example:

Heit sit te (*de/*in/*dy) jerappelskilen
Father sits to (*the/*a/*that/*those) potato-peel

Nor is a proper noun allowed:

a. De kapper begjint Oege te knippen
The barber begins Oege to cut
The barber begins to cut Oege's hair
b. * De kapper begjint te Oegeknippen
The barber begins to Oege-hair cut

The same pattern can be observed with respect to personal pronouns:

a. De kapper begjint him te knippen
The barber begins him to cut
The barber begins to cut his hair
b. * De kapper begjint te himknippen
The barber begins to him-cut

This property of non-referentiality follows immediately when we take into account that the incorporated noun is the non-head of a word. Elements within words are referentially opaque, and hence the incorporated noun is deprived of its referential capacity.

With respect to the semantics of the derived complex verb itself, it is obvious that they inherently show durative (or atelic, or imperfective) aspect. The general feature of this time constituency is that the activity the sentence is referring to is presented in such a way that no natural endpoint is implied, in other words, incorporation verbs express an "ongoing activity". One of the ways to accomplish this is by repeating the same activity an infinite number of times. The often observed habituality is thus simply a pragmatically conditioned subcase of iterativity.

There are several tests to detect durativity. The most simple and commonly used test is the one which depends on the selective properties of certain adverbial expressions. In English, the opposition between the prepositions in vs. for is used, where, for instance, for an hour indicates durativity and on the other hand in an hour denotes terminativity. In Frisian, these expressions could be translated as oerenlang and yn in oere , respectively. The aspectual differences can be illustrated by the following examples, where the (a)-example is potentially terminative and the (b)-examples durative:

a. Buorman fervet de doar yn in oere
Neighbour paints the door in an hour
Our neighbour paints the door in an hour
b. ? Buorman fervet de doar oerenlang
Neighbour paints the door for hours
Our neighbour paints the door for hours
a. * Buorman doarfervet yn in oere
Neighbour door-paints in an hour
b. Buorman doarfervet oerenlang
Neighbour door-paints for hours

Translating the result of incorporation into the well-known quadripartition of Vendler (1957), we can simply conclude that the incorporation verbs themselves cannot be an Accomplishment (a standard example being draw a circle), or an Achievement, with win the price as a classical example. These classes describe a definite period of time and hence show terminative (or telic) aspect. A difference between the two is that Achievements are momentary and hence do not show an internal process or development, whereas Accomplishments do. States, exemplified by verbs like know, like or hate, do not show internal development either. This is essentially what differentiates States from incorporation verbs, since it can be said that what they have in common is that both describe an indefinite period of time and hence are inherently durative. In short, incorporation verbs can be subsumed under a fourth Vendlerian class: Activities. As such, they can be put on a par with a verb like walk.

Incorporation verbs are inherently durative or atelic due to the unboundedness of the incorporated object noun. In this respect, unbounded objects like bare plurals or mass nouns behave similarly. There is a difference, however, although far more subtle than between terminativity and durativity itself. But what one could say is that when an activity tends to get institutionalized, then a construction with noun incorporation is the most appropriate way to denote to it. Compare the following sentences, both durative, but the first with incorporation:

a. Wy kofjedrinke om tsien oere
We coffee-drink at ten o'clock
We drink coffee at ten o'clock
b. Wy drinke om tsien oere kofje
We drink at ten o'clock coffee
We drink coffee at ten o'clock

Here, the description in the first example much more evokes a picture of the whole ceremony of drinking coffee, including the pouring of the liquid into the cups, the additional eating of cookies, the collegial chat, to mention a few highlights. The second sentence, on the other hand, is at first hand more restricted to the drinking of coffee proper. This is not to say that a construction with a phrasal bare object is excluded from an institutional or ceremonial flavour. The Dutch translation wij drinken om tien uur koffie , for instance, has it.

What emerges here is that, where a language has two ways to express durativity, the institutionalized (or habitual) reading tends to be reserved for the one with incorporation. No doubt, this follows from the fact that verb and object are tightened together in a compound, that is to say, in one word and concept.

[+]Some further requirements

In the section on syntactic evidence above, some syntactic arguments were provided which pointed at the word status of Frisian incorporation formations. Here, some more syntactic (and sometimes also semantic) properties of Frisian noun incorporation will be presented.

As may be clear from the examples of incorporated nouns thus far, the core of noun incorporation comprises nouns that function as direct object in a syntactic context. On the other hand, subjects are excluded:

a. De frou kuieret
the woman walks
The woman is walking
b. *Froukuieret

This also applies to subjects of unaccusative verbs:

a. De man falt
the man falls
The man is fallling
b. *Manfalt

Nouns that function as adjuncts are also prohibited:

a. Hja sille in middei te riden
they shall an afternoon to skate
They will go skating one afternoon
b. *Hja sille te middeiriden
they shall to afternoon-skate

However, various nominal arguments that syntactically show up in a PP are allowed to incorporate. They may express semantic roles like Location, Goal, Instrument and Patient. Examples are rútsjetikje window-tap (cf. tikje tsjin in rút tap at a window ), snoekfiskje pike-fish (cf. fiskje op snoek fish for pike), angelfiskje rod-fish (cf. fiskje mei in angel fish with a rod) and flaaksride flax-ride (cf. ride mei flaaks ride with flax to transport flax).

On the other hand, if the verb hase two internal arguments, it appears that incorporation is completely prohibited:

a. De direkteur jout de besikers plakplaatsjes
the manager gives the visitors stickers
b. *De direkteur plakplaatsjejout de besikers
The manager sticker-gives the visitors
a. Beppe set de boeken op it rim
grandmother puts the books on the shelf
b. *Beppe boekset op it rim
grandmother book-puts on the shelf
a. De boer laadt de wein mei hea
the farmer loads the wagon with hay
b. *De boer weinlaadt mei hea
the farmer wagon-loads with hay

Even in a simple setup of a verb with a subject and a direct object there are further restrictions. The subject has to be a volitional Agent:

a. Gurbe brekt in faas
Gurbe breaks a vase
a.' Gurbe faasbrekt
Gurbe vase-breaks
b. De bal brekt in faas
the ball breaks a vase
b.' *De bal faasbrekt
the ball vase-breaks

Furthermore, the object needs to be affected, hence it should act as a real Patient:

a. Richt fernimt boumantsjes yn 'e tún
Richt notices wagtails in the garden
b. *Richt boumantsjefernimt yn 'e tún
Richt wagtail-notices in the garden
a. De kealkop hatet negers
The skinhead hates negroes
b. *De kealkop negerhatet
The skinhead negro-hates
a. Oege kriget de hiele dei kadootsjes
Oege receives the whole day presents
Oege receives presents all day long
b. *Oege kadootsjekriget de hiele dei
Oege present-receives the whole day

An in-depth analysis of these and additional facts can be found in the chapters 3 and 4 of Dyk (1997). He rejects a syntactic derivation by way of head-movement in the style of Baker (1988), among others because of ECP-effects that cannot handle the facts. Instead, he proposes a lexical analysis in terms of argument structure in the spirit of Jackendoff (1990) .

The handbook by Hoekstra (1998:57-58) treats Frisian noun incorporation as an instance of univerbation, which, however, is problematic in the light of the various semantic restrictions imposed on noun incorporation.


This topic heavily draws on the doctoral dissertation by Dyk (1997), which is the main study on Frisian noun incorporation.

A short study about the place of Frisian in the typology of noun incorporation in the light of Mithun (1984) is Dyk (1993); see also Dyk (1997), section 2.6 (pp. 37-43). Dyk (1992) is an investigation into the aspectual aspects of Frisian noun incorporation. The question as to why Frisian possesses the phenomenon (which related languages do not) is treated in Dyk (1992) and in chapter 5 of Dyk 1997). These publications also offer older data and data from Frisian varieties in Germany. An Old Frisian instance of noun incorporation has been dealt with in Dyk and Bremmer (2007)

  • Baker, Mark C1988Incorporation. A theory of grammatical function changingChicago/LondonUniversity of Chicago Press
  • Dyk, Siebren1992Aspekt en nomenynkorporaasjeBreuker, Ph.H. & Jansma, L.G. (eds.)Philologia Frisica anno 1990: lezingen fan it tolfte Filologekongres, 17-19 okrober 1990Philologia Frisica anno49-61
  • Dyk, Siebren1992Warum gibt es im Westerlauwersschen und Föhrer Friesischen eine Nomeninkorporation?Faltings, Volkert, Walker, Alastair & Wilts, Ommo (eds.)Friesische Studien 1. Beiträge des Föhrer Symposiums zur Friesischen Philologie vom 10.-11. Oktober 1991NOWELE8Odense143-169
  • Dyk, Siebren1993Typologyen fan nomenynkorporaasje en it FryskTydskrift foar Fryske Taalkunde827-33
  • Dyk, Siebren1997Noun incorporation in FrisianLeeuwardenFryske Akademy
  • Dyk, Siebren1997Noun incorporation in FrisianLeeuwardenFryske Akademy
  • Dyk, Siebren1997Noun incorporation in FrisianLeeuwardenFryske Akademy
  • Dyk, Siebren1997Noun incorporation in FrisianLeeuwardenFryske Akademy
  • Dyk, Siebren1997Noun incorporation in FrisianLeeuwarden