The subject form of the personal pronoun of the third person singular feminine and the third person plural used to be hja (hju) /ja/ (/jø/) she; they. In the course of time, this form has acquired a dialectally very limited distribution. In the largest part of the language area it has been replaced by sij/sɛj/ (Klaaifrysk) or sy/si/ (Wâldfrysk). The clitic allomorph of hja/hju and sij/sy is se/sə/. This is also the case with the object form of the above pronouns, viz. har/har/her/ and har(ren)/har(ən)/them. However, it is very common for se to replace these pronouns in non-clitic position, although it cannot be stressed there. The use of se is exemplified in the table below:

a.	As subject	b.	As object
1	hja/sij komt by ússhe comes to us	1	ik haw/ha har(ren)/se dêr troffenI met her/them there
2	wannear komt se?when will she come?	2	har(ren)/*se ha(w) ik dêr troffenit was her/them that I met there
3	se hat ljocht hiershe has fair hair	3	wy ha(wwe) har(ren)/se in boek jûnwe gave her/them a book
4	hat se ljocht hier?does she have fair hair?	4	har(ren)/*se ha(wwe) wy in boek jûnit was her/them that we gave a book
5	se hie ljocht hiershe had fair hair	5	hy groef har(ren)/se ophe dug her/them up
6	hie se ljocht hier?did she have fair hair?	6	hy droech har(ren)/se in kerweike ophe told her/them to do a chore
7	se ha(wwe) allegear ljocht hierthey all have fair hair
8	ha(wwe) se allegear ljocht hier?do they all have fair hair?
9	se hiene(n) allegear ljocht hierthey all had fair hair
10	hiene(n) se allegear ljocht hier?did they all have fair hair?
11	se naam/namen alle boeken meishe/they took all books with her/them
12	naam/namen se alle boeken mei?did she/they take all books with her/them?
13	se falle/foelenthey fall/fell
14	falle/foelen se?do/did they fall?

As a subject, se is realized as [sə] in sentence-initial position, see (a1,3,5,7,9,11,13). As an object, se cannot occur in that position − see (b2,4) −, for it would have to bear a certain degree of prominence there, which is impossible for a word with schwa as its only vowel.

The interesting thing is the realization of se, both as subject and object, when it follows a finite verb, see (a1,2,4,6,8,10,12,14) and (b1,3,5,6). In (a1,2,4), where the verb ends in [-t], se is realized as [sə]. Since obstruents in a sequence must agree in voicing, this is to be expected.

If, on the other hand, the verb ends in a vowel, a sonorant consonant or a voiced fricative − see (a1,6,8,10,12,14) and (11,3,5,6) −, se is realized as [zə], with a voiced fricative. This cannot be ascribed to regular voicing assimilation, for the latter only works regressively in Frisian (see Regressive Voice Assimilation: type 1, Regressive Voice Assimilation: type 2, and Regressive Voice Assimilation of obstruent sequences). Besides, the process would have to affect the word se only, hence it would lack any generality. The most plausible assumption therefore is that se/sə/ has the allomorph /zə/.

It should be borne in mind that se, being a clitic, has to lean on a host, with which it makes up a phonological word. The latter must be in conformity with general conditions on its shape. Since the trochee is the unmarked foot type in Frisian, se is likely to cliticize onto a monosyllabic host to its left. Of the coronal fricatives, /s/ and /z/, the former occurs in word-initial position and in the onset of a stressed word-medial syllable, the latter, among other things, in the onset of an unstressed word-medial syllable (see The obstruents: the fricatives). The realization of se/sə/ as [zə] in (a6,8,10,12,14) and (b1,3,5,6) therefore links up with the independently motivated distribution of the coronal fricatives.

There may, however, be more to the realization of se as [zə] than just this, for there are indications that the phonolological variant [zə] has developed into the independent allomorph ze/zə/ (see Visser (1988:197-199)). The latter is a peculiar form, since it is the only native free morpheme of Frisian which begins with the voiced alveolar fricative /z/. If a finite verb and se are separated by a pause or a parenthetic clause, se can still be realized as [zə], which is exemplified in (3):

The same is found with se in coordinated phrases, see the example in (4):

Embedded in an infinitival verb phrase, [zə] is even allowed in sentence-initial position, of which (6) provides examples:

The element [zə] alone cannot occur in sentence-initial position, neither as a subject nor as an object:

Though ze/zə/ has a more limited distribution than se/sə/, the above is an indication that the former has acquired an independent status vis-à-vis the latter. The relation between se/sə/ and ze/zə/ can be expressed as follows: