Since a word cannot begin with schwa, a word-initial vowel cannot reduce, which is illustrated in (1):

Vowel reduction, however, appears to be such a forceful process that some unstressed, short, word-initial vowels did undergo reduction. The resulting schwa, however, could not endure in word-initial position, for it was replaced by a short full vowel. Some examples of this are given in (2):

Another possibility is that j /j/ was inserted before schwa, see (3):

But schwa could also delete, as in the words in (4):

There is one case, see (5), where the reduced vowel was replaced with the non-native affix eks-:

So, although there is a strong tendency for short, unstressed vowels to reduce − in whichever position − the ban on word-initial schwa appears to be stronger.

/h/ only precedes vowels (see the glottal fricative /h/), but not schwa, so a vowel following /h/ cannot reduce. This means that the highlighted vowels in the words in (6) cannot, and do not, undergo reduction:

Though a vowel is more reduction-prone in an open than in a closed syllable, reduction does not occur here. Neither does it in the words in (7):

Schwa would end up in a closed syllable here, which is a less favoured configuration. However, the coda consonant is /m/ or /r/, both of which belong to the set of unmarked schwa codas (see schwa). This means that the impossibility of vowel reduction here is solely to be ascribed to the presence of /h/.

In this case as well, vowel reduction appears to be such a forceful process that some unstressed, short vowels following /h/ did undergo reduction. But the resulting configuration was repaired, which was achieved in various ways, as illustrated in (8):

If the full vowel of a syllable with a single onset and a single coda undergoes reduction, schwa ends up in a less favoured configuration. The single onset is fine, less so is the fact that schwa is in a closed syllable. There are words in which the latter configuration is repaired by moving the coda consonant to the onset. It is especially with /r/ that this movement took place, and predominantly in loanwords. Examples are given in (9):

Moving /r/ from coda to onset resulted in an open syllable, which is an improvement. At the same time, however, the schwa syllable acquired a complex onset, which is a change for the worse, even though the onset has the ideal shape of obstruent + liquid. There are, so to speak, conflicting interests here. This may be the very reason that moving /r/ was not obligatory, so that forms which do and do not show the traces of it occur side by side, as shown by the doublets in (9). Here as well, vowel reduction manifests itself as a forceful, virtually inescapable process.

If the full vowel of a syllable with a complex onset (obstruent + liquid) undergoes reduction, schwa also ends up in a less favoured configuration, for in the unmarked case a schwa syllable does not begin with a consonant cluster. There are words ‒ only loanwords in this case ‒ in which this was repaired by moving the right-most onset consonant to the coda. Examples are given in (10):

Moving the liquid from onset to coda results in an onset occupied by a single consonant, which is an improvement. At the same time, however, the schwa syllable acquires a coda, a change for the worse. Here as well there appear to be conflicting interests, and here as well this may be the reason that moving the liquid is not obligatory, which results in doublets.

In both cases, vowel reduction creates a schwa syllable. The latter, however, has to accommodate more consonants than it does in the unmarked case. Such a situation is best remedied by the deletion of /r/, especially when /r/ is in the syllable coda, since this results in an open syllable. This is what occurs in a fair number of (loan)words, as illustrated by the examples in (11) (see /r/-deletion for the deletion of /r/ in general):