In classic Old Norse, and specifically (Old) Icelandic, there are many homophonous words á. Many of these were probably distinguished in the early 9th century. I thought it might be interesting to examine the different words.
Below are all forms that appears as á on Wiktionary:
http://en.wiktionary.org/wiki/%C3%A1#Icelandic
http://en.wiktionary.org/wiki/%C3%A1#Old_Norse
*ā! interjection
Remained relatively unchanged throughout history, I think.
*āh v ‘own(s), have/has’, 1sg and 3sg prs.ind.act of *æiǥą
1sg: *āh
2sg: *āht
3sg: *āh
1pl: *æiǥum
2pl: *æiǥuð
3pl: *æiǥų
From PG *aih, inf *aiganą
*ą̄ adp ‘on’
From PG *ana.
*ɔ̄ f ‘ewe’, acc.sg of *ɔ̄ʀ
case: sg — pl
nom: ɔ̄ʀ — ɔ̄iʀ
acc: ɔ̄ — ɔ̄iʀ
dat: ɔ̄ — ɔ̄um
gen: ɔ̄aʀ — ɔ̄a
From PG *awiz. OIc has ær for the nom.sg, gen.sg, nom.pl and acc.pl. In other forms, the vowel is á. Early manuscripts have ǫ́, ǫ́um and similar forms, but they still have the æ-vowel before r. Here's my thinking:
- W-umlaut affected the vowel while it was still short.
- Medial *w was lost after short vowels, lengthening the preceding vowel (open stressed stems always have a long vowel).
- The vowels in hiatus were not contracted.
- The OIc vowel æ was caused by ʀ-umlaut and not i-umlaut. This is probably a relatively late change.
case: sg — pl
nom: ɔ̄w — ɔ̄waʀ
acc: ɔ̄w — ɔ̄waʀ
dat: ɔ̄u — ɔ̄um
gen: ɔ̄waʀ — ɔ̄wa
From PG *ahwō > PN *ahwu. Early OIc manuscripts have ǫ́ (before the shift ǫ́ > á). The forms are based on these principles:
- Medial and final *w remained after long vowels and diphthongs, as can be seen in early English loanwords. See E.V. Gordon, An Introduction to Old Norse, p. 328.
- The semivowel *w occured after long vowels in tautosyllabic position for a time after the syncope and apocope, as it did in Old English. I guess it could be said to form overlong diphthongs. Loanwords perhaps indicate that this was still the case in the early 9th century.
- *w was already lost before rounded vowels, however.
- Short *ŭ remained after monosyllabic stems ending in a long vowels, meaning that such stems counted as light.
- There was less contraction of vowels in hiatus.
- There was no contrast between a disyllabic sequence of a long vowel + *ŭ and a monosyllabic sequence of a long vowel + *w. This is mostly speculation but it reduces the number of possible contrasts to uphold. Actually, *w and *ŭ were probably allophones in all positions. It appears that ultrashort *ŭ developed as a vocalization (although the Runic inscriptions are ambiguous) of *w between consonants after vowel loss. Compare Rök <fiaru>, probably *feŭrŭ (although possibly *feŭrw), for OIc fjǫr, from PG *ferwą. Also Rök <karuʀ>, probably *gɔrŭʀ for OIc gǫrr.
- *h was lost in most medial clusters (but not *ht) and between vowels, very much like in Old English.
- *w caused mutation of the previous vowel in both East and West Nordic.
This is perhaps the most uncertain form. I can't find this OIc á in Cleasby/Vigfusson or Zoëga. The entry on Wiktionary appears to be yet another form of the word æ, which in OIc frequently appears as ei or ey. This is from a PG stem *aiwi–, with the interesting *aiw-sequence. According to Kock, OSw had a variant ā although SAOB writes that this form “kan ej anses säkert styrkt”.