There are words which are always realized with a nasal vowel, as exemplified in (1):

The Dutch cognates of these words are Wijns, ons, grens, slons, ernst, kans, aanstonds. They all have the consonant /n/ in their underlying and surface (and orthographic) representation. /n/ must have been part of the Frisian words as well; it is still written, but no longer realized.

The question then is whether there are reasons to posit /n/ in the underlying representations of the above words or, put differently, whether there are reasons not to assume that these words have /n/ in their underlying representations.

There are pairs of words which only differ in being realized with an oral versus a nasal vowel, see (2):

Since both the oral and the nasal vowel precede [s] here, there does not seem to be a complementary distribution, so that the difference between these vowels seems to be distinctive, that is, the nasal vowel seems to have phonemic status here. This would imply that these word pairs have the underlying representations in (3):

It might be considered an advantage of underlying representations like the ones in the right-hand column of (3) that the nasal vowel need not be derived, but that it is present as such from the very beginning, so that underlying representation and surface form link up well with each other (in this respect at least).

This advantage, however, is only apparent. Underlying representations with nasal vowels are too 'concrete', too specific, too 'surface-oriented'. Two severe disadvantages of such representations are that they lead to an unnecessarily complex lexicon and that they prevent one from stating generalizations.

It should be kept in mind that vowel nasalization is a fully productive and transparent phonological process in Frisian, which must be accounted for anyway, that is, apart from the words in (1) and those in the right-hand column of (2), which are invariably realized with a nasal vowel. These then should be assigned underlying representations with long, oral vowels + /n/, as in (4):

These underlying representations are more abstract than the ones with underlying nasal vowels in (3), for the nasal consonant /n/ does not show up as such in any of the surface forms of these words.

The underlying contrast is between zero and /n/, as in /vi:z/ (wiis) vs. /vi:nz/ (Wyns). It manifests itself in the nasality of the vowel of Wyns, which contrasts with the orality of the vowel of wiis. The underlying nasal consonant thus is recoverable from the surface nasality of the vowel.

The slightly abstract underlying representations in (4), like /ɡre:nz/ (grins), /u:nz/ (ûns), and /vi:nz/ (Wyns), have many advantages.

In the first place, there is a conceptual advantage. Assuming nasal vowels in underlying representations leads to a considerable increase of the vowel inventory of Frisian. The oral vowels and the alveolar nasal consonant /n/ are needed anyway; therefore, underlying representations with only oral vowels and /n/ pose no problem in this respect.

In the second place, there are descriptive advantages. First, assuming nasal vowels in underlying representations is advantageous only if it would render the statement of the nasalization process in the grammar of Frisian superfluous. But this is not the case. The transparent alternation between oral and nasal vowels, as in base forms vs. inflected forms, must be accounted for, apart from the question as to whether a simplex word like ûns has the underlying representation /ũ:z/ or /u:nz/.

Second, underlying representations are used, among other things, for encoding exceptional and unpredictable phonological properties of morphemes. Therefore, if nasal vowels can be part of underlying representations, they may be expected to occur in contexts which differ from those of the nasal vowels which regularly derive from the sequence oral vowel +/n/. This, however, is not the case. The contexts are exactly the same. Where nasal vowels appear is thus fully transparent and predictable. Not only is it superfluous to include them in underlying representations, this also prevents stating the above generalization.

Third, if nasal vowels derive from the sequence oral vowel +/n/, it is to be expected that all and only the oral vowels which are allowed to precede /n/ within the same morpheme may undergo nasalization. There are three pieces of evidence for this.

The oral vowels which are not allowed to precede /n/ are the ones which do not have a nasalized counterpart, see (5):

So, the ill-formedness of the sequences in the left-hand column matches the ill-formedness of the nasalized vowels in the right-hand column.

Assuming an underlying /n/ in words which are always realized with a nasal vowel allows for a simple statement of some dialectal vowel alternations. First, in the southeastern part of the Frisian speaking area, the sequence /-o:n/ occurs, which is /-un/ elsewhere, see the examples in the table below:

Southeastern	Elsewhere	Translation
groon/ɡro:n/	grûn/ɡrun/	ground
hoon/ho:n/	hûn/hun/	dog, hound
spoons[spõ:s]	spûns[spũ:s]	sponge
doons(je)[dõ:sjə]	dûns(je)[dũ:sjə]	to dance

The vowel contrast in the latter pairs of words nicely fits in with the grûn/hûn - groon/hoon pattern, if spûns/dûns(je) and spoons/doons(je) are assumed to have the underlying representations /spu:nz-spo:nz/ and /du:ns-do:ns/. Underlying representations with nasal vowels, viz. /spũ:z-spõ:z/ and /dũ:s-dõ:s/, would call for a separate explanation, so that a generalization would be missed.

Second, in the northwestern part of the Frisian speaking area, the sequence /-ɔ(:)n/ is /-on/ and /-oən/, as exemplified in the table below:

Northwestern	Elsewhere	Translation
man/mon/	man/mɔn/	man
fan/fon/	fan/fɔn/	of
hân/hoən/	hân/hɔ:n/	hand
lân/loən/	lân/lɔ:n/	land
kâns[kõə̃s]	kâns[kɔ̃:s]	chance
lâns[lõə̃s]	lâns[lɔ̃:s]	along

As is to be expected, words with the sequence /-ɔ:ns/ have northwestern variants with [õə̃s]. Again, slightly abstract underlying representations ‒ /kɔ:ns/ and /koəns/ (kâns) and /lɔ:ns/ and /loəns/ (lâns), so with an oral vowel and the nasal /n/ ‒ allow for a unified and generalizing treatment of this vowel alternation.

In conclusion, all nasal vowels derive from the sequence oral vowel + /n/. In Frisian then nasal vowel and nasalized vowel amount to the same.