Taalportaal - the digital language portal

Dutch
Frisian
Saterfrisian
Afrikaans

Show all

Weak verbs

quickinfo

Frisian verbal inflection is dependent on the type of verb involved. The main distinction is between regular inflection (the so-called weak verbs) and irregular inflection. Regular verbs keep their stem intact; the expression of features like tense and agreement is solely done by way of suffixation. With irregular verbs we typically observe stem alternations. For tradionally strong verbs this is the way of marking the preterite, for instance.

Within the Germanic languages, Frisian is rather peculiar in that it has two weak classes at its disposal. What is traditionally called class I has an infinitive ending -e, where the infinitives of class II show the suffix -je. The paradigm of class I resembles the weak verbs of related languages. For instance, it has a preterite suffix with a dental consonant, in the Frisian case -de or -te. The paradigm of class II is rather different. The distribution of the two classes seems mainly a matter of accident. However, at least one phonological and morphological dependency may be observed. Verbs with a stem ending in a schwa plus sonorant are always je-verbs, e.g. timmer-je to hammer. Also, verbs that are built by way of conversion heavily tend to be a member of class II. Thus from the noun keal calf we can derive keal-je to calve, and the adjective koel cool may thus result in kuol-je to cool.

Membership of a class is not always stable over time. There is a natural pressure to give up irregularity. Transition into the opposite direction is rare, but does occur. Influence of Dutch also causes a switch of the weak class II to class I.

Frisian not only has two classes of weak verbs at its disposition, but in addition there is a great deal of variability within them. This is due to dialectical differences, residues of older forms and effects of phonological processes. We also see interparadigmatic crossovers, in that certain suffixes may show up in another class. Special sections have been devoted to remarkable forms for the second person singular and the imperative.

Two classes of weak verbs
Distribution of the classes
Paradigm of class I
Paradigm of class II
Assibilation
Stem alternations
Shorter suffixes after schwa syllables in class II
Reinforcement of endings
Second person singular
Imperative
Pressure from Dutch

[+]Two classes of weak verbs

Frisian has two classes of weak verbs at its disposal. Class I is commonly referred to by its infinitive, whichs ends in the suffix -e, hence the name e-verbs. It resembles closely the system that may be found in other Germanic languages like Dutch, German and English. Its main characteristic is the formation of the past thense, which is by way of a dental suffix, i.e. -de or -te in the singular. Also the suffix of the past participle is dental: -d or -t.

The second class of Frisian weak verbs behaves very differently. Conspicuous is the infinitival ending -je, hence this class is commonly referred to as je-verbs. The ending returns in the present tense in first person singular, in the plural and in the imperative. The second and third person singular also show a schwa. This has the effect that the suffixation in the present singular builds an extra syllable, which is not the case in Class II. The suffix of the past tense stands out in that it lacks a dental ( /d/ or /t/). This, however, is a feature of Modern West Frisian, where Old Frisian /d/ was deleted (see Meijering (1980)). As is mentioned in the section on Weak verb classes in East and North Frisian below, the situation in East- and North Frisian dialects can be different.

extra

Weak verb classes in East and North Frisian

The Old Frisian two-class-system of weak verbs with its infinitive endings in -a and -ia, occurs in West Frisian as two classes of -e and -je. In East Frisian and North Frisian, it has been preserved differently throughout the varieties. Bosse (2013:7-25) gives a detailed overview of the different developments. The only variety of East Frisian left today, Saterlandic Frisian, preserved the system just like West Frisian did, with the single exception that the -je class did not lose its dental in past tense and past participle (table according to Bosse (2013)):

Inflection of brûke/bruke to use; to need and freegje/fräigje to ask

Figure 1

[click image to enlarge]

The extinct variety of the island Wangerooge had two weak verb classes as well, but its distribution differs from that of the Old Frisian system and is only phonologically based (see Bosse (2012), Bosse (2013) and Versloot (2001:428-429)).

The North Frisian varieties preserved the system in Insular North Frisian, though phonologically different. Due to schwa apocope in Island North Frisian, Old Frisian -a became Ø (null morpheme), while Old Frisian -ja became [ɪ] (-e or -i, spelled differently) in the variety of the islands Föhr, Amrum and Sylt and [ә] in the variety of the island Heligoland. The dental in past tense and past participle has been preserved, too.

As Bosse (2013:16-24) shows, the Mainland North Frisian varieties can be separated into a (now mostly extinct) southern part (the Südergoesharde and the Mittelgoesharde) which preserved the Old Frisian two-class-system, and a northern one (the Halligen, the Norderdoesharde and everything north of it) which reorganized the distribution on a phonological basis and distinguishes two classes, based on the ending of the verb's stem. If the verb stem ends with a sonorant or a vowel, there won't be any vowel between verb stem and inflectional suffix (class 1*), whereas there will be an -e- in all other cases (class 2*). See Hoekstra (2001:778) for a brief overview. As there are exceptions from this rule (see brüke in the table below which won't fit in the schema), Bosse suggests that there could be some relics of the old distribution of classes Bosse (2013:20).

Inflection of brûke/brüke to use; to need, freegje/frååge to ask and helje/hååle to fetch

Figure 2

[click image to enlarge]

The reason for this system of distribution (Insular North Frisian and southern Mainland vs. northern Mainland North Frisian) is not quite clear, yet. Bosse sees a possible explanation in schwa apocope, which basically also took place in the variety of Mittelgoesharde and might have forced a development similar to that on the islands. The variety of the Südergoesharde might hence have been isolated from the developments in the north. There might as well have been other factors, see Bosse (2013) for a detailed analysis.

The Old Frisian plural ending -ath in present tense has been replaced in all Frisian varieties by an ending -e (or apocope where it took place). Only the varieties of Wangerooge and the Südergoesharde have both kept an ending -(e/i)t:

Inflection of freegje/fräigje to ask

Figure 3

[click image to enlarge]

Hoekstra (2001:778-779) assumes a supportive influence from neighboring Low German varieties, which also have got a plural ending -t in present tense.

[This extra is written by Hauke Heyen (Kiel)]

[+]Distribution of the classes

In general, it is hard to tell beforehand which verbal stem takes which suffix, -e or -je. Both classes already existed in the Old Frisian period, with the corresponding endings -a and -ia. In general, it seems that the membership of a class remains relatively stable over time. Thus Old Frisian sitta to sit is sitte nowadays, and Old Frisian folgia to follow is modern folgje. There are exceptions, however. So libje to live nowadays belongs to class II, but in Old Frisian the verb was libba and hence belonged to class I. The Renaissance poet Gysbert Japicx had relatively often je-verbs where these nowadays belong to class I. Examples are bruwckje to use, nowadays brûke, or rêstje to rest, which is modern rêste.

Both classes can be found in all dialects. Some individual verbs have dialectical variants, but it is hard to discern a pattern in the facts. Possibly, there is a slight tendency that -je can be found more often in the south. Examples are the verb passe to fit, which is pasje in the deep southwest, and northern damme to play checkers vs. southern damje.

Despite these uncertainties, there is a phonological criterion that does not allow exceptions. If a stem ends in a syllable that contains a schwa, then it unequivocally opts for class II. Examples are skarrelje to rummage, hoffenje to clean, timmerje to hammer and itigje to calibrate.

There is also a morphological generalization. Conversions of other lexical categories to verbs heavily tend to result in a verb belonging to class II. Thus the adjective grien green becomes grienje make/become green, the noun ferve paint results in fervje to paint. From the cardinal number ien one we can build ienje to bring at one, i.e. to thin out (of sugar beets) and from the now obsolete interjection moart we still have the verb moartsje to yell. Of course, in many cases this morphological generalization combines with the phonological rule. All such products of conversion probably contribute to the fact that the majority of the Frisian verbs belong to class II.

The existence of the two classes can sometimes bring about semantic differences. For instance, the verb hurde to bear, stand is opposed to hurdzje to harden. And the verb sichte to determine the height of an unknown point on the basis of two others is semantically different from sichtsje to cut down with a sickle.

extra

Literature

Some information about the two classes in older stages of the language may be found in Tamminga (1963:142-145). He also mentions the effect of conversion. The existence of southern je-varieties of the verbs brûze to foam, sûze to rustle and a few others with the same rhyme are mentioned in Hof (1933:20).

[+]Paradigm of class I

Weak verbs of class I obey the following paradigm. First the schema of the finite forms is given:

Table 1

Tense	Number	Person	Ending
present
	SG
		1	-
		2	-st
		2 polite	-e
		3	-t
	PL
		1	-e
		2	-e
		3	-e
past
	SG
		1	-de/-te
		2	-dest/-test
		2 polite	-den/-ten
		3	-de
	PL
		1	-den/-ten
		2	-den/-ten
		3	-den/-ten

Note that the endings of the past tense actually consist of two suffixes. Firstly, we have -de or -te indicating the past tense, and secondly there are the agreement suffixes, zero -∅, and -st and -en. Furthermore, it has to be mentioned that the polite pronoun jo you is semantically singular, but that it takes plural verbal morphology.

The non-finite forms, the past and present participle, both forms of the infinitive, and the imperative, are as follows:

Table 2

Category	Subcategory	Ending
participle
	present	-end(e)
	past	-d/-t
infinitive
	I	-e
	II	-en
imperative		-

To be a bit more concrete, see the following forms of the verb wenne get used to (which will be contrasted below with the class II verb wenje to live). First the present tense:

ik wen

I get used to

do wenst

you.SG use.2SG.PRES

you get used to

jo wenne

you.POL use.PL.PRES

you get used to

hy/hja went

he/she gets used to

wy wenne

we get used to

jimme wenne

you.PL use.PL.PRES

you get used to

hja wenne

they get used to

Here are examples with the past tense:

ik wende

I got used to

do wendest

you.2SG use.2SG.PRET

you got used to

jo wenden

you.POL use.PL.PRET

you get used to

hy/hja wende

he/she got used to

wy wenden

we got used to

jimme wenden

you.PL use.PL.PRET

you got used to

hja wenden

they got used to

Finally, examples of the non-finite forms:

wennend(e) oan 'e omstannichheden

use.PRESP to the circumstances

getting used to the circumstances

wend oan 'e omstannichheden

use.PP to the circumstances

used to the circumstances

wenne oan 'e omstannichheden

use.INF to the circumstances

to get used to the circumstances

om te wennen oan 'e omstannichheden

PRT to get.used.INF to the circumstances

in order to get used to the circumstances

wen oan 'e omstannichheden!

use.IMP to the circumstances

get used to the circumstances!

The paradigm of class I gives two possibilities for the suffix of the preterite, and also for the past participle, viz. a form with /d/ or /t/. The choice fully depends on the voicing of the final segment of the stem. If this is voiceless, i.e. ending in /p/, /t/, /k/, /f/, /s/ or /x/, then the first element of the suffix is also voiceless, hence with a /t/. This results in the suffixes -te(n) for the finite forms and -t for the past participle. If by suffixation two identical segments follow each other, then by degemination only one of them is actually heard. Thus if for example of the verb sette to set the stem set [sɛt] combines with the past tense ending -te, then the result is [sɛtə], pronounced with a single [t].

Some general orthographic conventions also apply to the spelling of verbs. In stems with short vowels, we have a doubling of the following single consonant, for instance wy wenne we get used to in the examples above. Long vowels in the stem are doubled in a closed syllable, and get a single notation if the syllable is open. For instance, we have ik praat I talk vs. wy prate we talk. The suffixes of the past tense are simply added to the stem, without change in the orthography. Stems with a long vowel followed by a single consonant consisting of <d> or <t> ar an exception. In this case, this consonant remains single after adding the suffix. So we have the form wy praten we talk.PRET we talked. This is different from Dutch, which explicitely reflects the addition of the suffix to the stem in the spelling. Hence, in Dutch we have the spelling wij praatten, although the meaning and the pronunciation compared to Frisian wy praten is exactly the same. On this issue, Dutch orthography takes a morphological stand, where Frisian has a phonological position. The Frisian solution is in line with other instances where the last segment of the stem and the initial consonant of the suffix are identical. Hence, of the verb passe to fit we have the second person singular present tense do past, and not *passt. Also, it is not hy *sitt, but hy sit he sits.

extra

Loss of /d/ on the island of Terschelling

Knop (1954:243-247) reports that the first segment of the preterite suffix -de(n) may delete after stems consisting of a long vowel which is followed by a consonant. The deletion is more often found in the singular affix -de than in plural -den. So we get forms like dele (cf. standard Frisian deelde parted), keare (cf. standard kearde turned) and klime (cf. standard kliemde stained).

[+]Paradigm of class II

Weak verbs of class II follow the following paradigm. Again, the finite forms are presented first.

Table 3

Tense	Number	Person	Ending
present
	SG
		1	-je
		2	-est
		2 polite	-je
		3	-et
	PL
		1	-je
		2	-je
		3	-je
past
	SG
		1	-e
		2	-est
		2 polite	-en
		3	-e
	PL
		1	-en
		2	-en
		3	-en

The non-finite forms of the verbs of class II are suffixed as follows:

Table 4

Category	Subcategory	Ending
participle
	present	-jend(e)
	past	-e
infinitive
	I	-je
	II	-jen
imperative		-je

extra

Thematic elements

The difference in suffixation between class I and II would reduce significantly if it was assumed that class II contains a thematic element, consisting of -j(e)- or -e-. For instance, an abridged present tense singular would then be -je- ø, -e-st and -e-t. The present plural would be -j-e . This array of (final) suffixes thus boils down to ø-st-t-e, which is exactly the same as the set displayed in class I. A suggestion by Jarich Hoekstra for such an analysis is mentioned in Werner (1992:183) and slightly elaborated in Hoekstra (2001:90-91). See also Werner (1993).

For concreteness' sake, full forms are presented below, and in order to compare optimally with class I, these will be instances of the verb wenje to live (reside). First, the forms in the present tense:

ik wenje

I live

do wennest

you live

jo wenje

you.POL live.PL

you live

hy/hja wennet

he/she lives

wy wenje

we live

jimme wenje

you.PL live.PL

you live

hja wenje

they live

Likewise the past tense of class II verbs:

ik wenne

I lived

do wennest

you lived

jo wennen

you.POL live.PRET.PL

you lived

hy/hja wenne

he/she lived

wy wennen

wy lived

jimme wennen

you.PL live.PRET.PL

you lived

hja wennen

they lived

Finally, the non-finite forms of class II:

wenjend(e) yn Snits

living in Sneek

Ik haw wenne yn Snits

I have lived in Sneek

wenje yn Snits

to live in Sneek

om te wenjen yn Snits

in order to live in Sneek

wenje yn Snits!

live in Sneek!

Also in this class the usual orthographic rules apply. See the doubling of a consonant after a short vowel, as in wenne. Or the double spelling of a long vowel in a closed syllable, as in traapje to kick, which shifts to a single spelling in an open syllable, as in (ik) trape (I) kicked.

extra

The third person present tense in the dialect of Schiermonnikoog

The Frisian dialect of the island of Schiermonnikoog shows a unique difference from the standard pattern of the paradigm of je-verbs. In the third person of the present tense, the /t/ of the suffix -et is lacking. Thus of the verb winje to live we get he wene he lives. This is probably a feature that had a wider spread once than it has today. Hof (1919) notices that it also occurs in a collection of proverbs dating from the middle of the 16th century. Its author Reyner Bogerman originated from the Dongeradielen, the part of the mainland directly opposite to the island. A comparable phenomenon is the loss of -t in a few irregular verbs.

extra

Residues of older participial forms

In Old Frisian the preterite and the past participle showed a /d/ in the suffix (see Bremmer (2009:80)). For instance, of the classical Old Frisian verb folgia to follow the respective forms were folgade follow.PRET.SG followed and folgad follow.PP followed. Later we see a reduction of the vowel /a/ to schwa, and already at the end of the Middle Ages the /d/ of the preterit tends to get lost. Soon after, the participle followed, but in the peripheral Frisian dialects a residue of the old situation may still be found, although this is restricted to the participle. For the historical process, see Meijering (1980).

The dental is still alive verbally in the grammar of the Frisian dialects of the island of Skylge (Terschelling). According to Knop (1954:242), the suffix of the past participle in these dialects is -ed. Hence forms like kleaged complained and korted shortened. We find the same ending -d in the dialect of Schiermonnikoog, but here it is restricted to past participles used attributively. Among others, Fokkema (1969:24) gives the following examples:

ferjôged as in hieun

driven off like a dog

de grienfarved kist

the greenpainted chest

de straungede schiipe

the stranded-INFL ships

the stranded ships

Likewise as on the island of Schiermonnikoog, in the dialect of Hindeloopen the suffix -ed is only found if the participle is used as an adjective. Examples like in lobbeden hûnd a castrated dog or in wyteden moere a whitened wall have been encountered, but nowadays, the suffix is obsolete. Within the written inheritance of the dialect, its greatest spread is in the form of nominalized pseudo-participles, indicating the various patterns of textile for the tradional costume of the small town. Hence assignments as aig-ed-e eye.PP.NOM pattern with eyes or raa-ryng-ed-e red-ring-PP-NOM pattern with red rings. See for this pattern in the part on synthetic compounds.

Superficially, it seems that old forms with /d/ even may be found in the past tense. In the western dialect of the island of Terschelling forms like maked made, drigede threatened, sâltede salted have been encountered. Jongkind and Van Reenen (2006:120-121) suppose, however, that these are not old but rather younger forms, to be analysed as a reinforcement of the past tense by borrowing the past tense suffix of the paradigm of class I. In an older description of the dialect, Knop (1954), such forms fail altogether.

[+]Assibilation

The verbs of class II with a stem ending in /d/ or /t/, respectively, show the insertion of a /z/ and an /s/, thereby respecting the [+/- voice] of the stem. Note that the sibilant only shows up before -je. We thus have an infinitive wedzje to bet, and subsequently also forms like ik wedzje I bet, hja wedzje they bet and wedzjende bet.PRS.PTCP betting. An example with a stem ending in /t/ is rotsje to rotten, with comparable inflectional forms. Hence, the present tense of a verb like fetsje to grasp is as follows:

ik fetsje

I grasp

do fettest

you grasp

hy fettet

he grasps

wy/jimme/hja fetsje

we/you/they grasp

The original phonological change, the insertion of a sibilant between /d/ or /t/ and /j/, took place in the 18th century, but it did not affect the peripheral dialects of Hylpen (Hindeloopen), Skiermûntseach (Schiermonnikoog) and the eastern dialect of the island of Skylge (Terschelling). There the clusters /djə/ and /tjə/ remained intact. In the western Frisian dialect of Skylge this assibilation did occur, however, with the subsequent loss of /j/ (which can also be observed in diminutives). In Westerschelling we thus have forms like fetse to grasp and wedze to bet, which are fetsje and wedzje in standard Frisian. In contrast, we encounter the forms fatje and wedje on the island of Schiermonnikoog, representing one of the three dialects that were not affected by the assibilation.

However, also in the Frisian heartland this insertion does not occur in the particular phonological context after a stem final consonant cluster -st. At least, superficially this seems to be the case. So we have a verb which is spelled treastje to comfort, and not *treastsje. The stem of this verb is treat. Other examples are ferwoastje to destroy, dwêstje to extinguish, fertoarstje to have a thirst and kostje to cost. In order to see what is really going on here, we have to take as closer look at the pronunciation. The last example kostje is pronounced as [kɔsjə]. This surface form is best explained if we assume that insertion of /s/ really took place. This would yield a form /kɔstsjə/. A rule of t-deletion in the context /sts/ would result in the sequence /ss/, which after degemination becomes the single /s/ of /kɔsjə/. In some verbs, and for some speakers, such a form may be the impetus to reanalyze the stem in its entirety. We can then come across forms like the third person singular present tense kosset /kɔsət/ or the past participle kosse /kɔsə/. Such a reanalysis seems to be restricted to verbs with a relative high frequency. A form like [trIəsət] for treasted does not occur. Moreover, the reanalysis is not reflected in the spelling, which always retains <t> here.

extra

Literature

The phonological exception on the insertion of /s/ is analyzed in Visser (1993). He also points to the effect that speakers try to compensate for the decrease of transparency of the stem. This can be done by taking recourse to the paradigm of the weak verbs of class I. Here Visser mentions the infinitive liste, instead of listje to frame. Another strategy might be the insertion of an augment -(i)g-. We then get verbs like hastigje to hasten (instead of haastje, with shortening) or dwêstgje to extinguish instead of dwêstje.

[+]Stem alternations

In the unmarked case, it could be expected that the stem remains intact throughout the paradigm. This is not true, however. A process as final devoicing applies automatically. For instance, it affects the first person singular of the present tense and the imperative of class I. Thus a form ryd of the verb ride to ride is phonetically [rit]. Assimilation applies in the third person rydt, which is also pronounced as [rit]. With respect to the second person, the final segment even deletes, as a consequence of t-deletion before -st. The form rydst then becomes [rist].

In the dialect of the island of Schiermonnikoog we see an interesting allomorphy on the edge of the stem of class I verbs and the suffixes of the past tense and the past participle. In this dialect, clusters of /r/ plus /d/ turn to /z/ (with a hypothetical historical intermediate stage /dz/). In the past tense, this resulted in a form ik hieze I rented (from ik hierde). In combination with final devoicing, the past participle became hies rented, derived from the form hierd. Later, the full form of the suffix became reintroduced in order to make the paradigm of class I more regular, with /z/ remaining, however. Most speakers nowadays say ik hiesde and hiesd.

extra

Literature

For this change see Dyk (2006).

With respect to vowels, we only rarely find effects of breaking and shortening. Although widely available in the verbal paradigm of class II, we never see these processes effected in their canonical context, i.e. theaddition of a syllable. Quite marginally, we see these changes operating before consonant clusters, typically occurring in the paradigm of class I. The only verb that shows traces of breaking is sliepe to sleep. Thus for several speakers, the three persons of the present tense are sliep sleep.1SG.PRS [sli.əp], sliepst sleep.2SG.PRS [sljIpst] and sliept 3SG.PRS [sljIpt]. Hence, the second and third person are broken before the clusters [pst] and [pt], respectively. In the same way, the forms of the past tense and the past participle are broken; in all these cases, we see a cluster [pt]. Historically, in other dialects this verb has been affected by shortening, see strong and other irregular verbs.