C = any consonant
F = voiceless fricative consonant (/f, θ, s/)
L = liquid consonant (/l, r/)
V = any vowel
Today, children learning English must learn that words like thief exhibit a stem alternation. In the singular, the stem ends with a voiceless fricative (/f, θ, s/), whereas in the plural, the stem-final consonant is the corresponding voiced fricative (/v, ð, z/). The stem alternation is a lexical property. That is, it is an idiosyncratic property of certain words and can't be derived algorithmically (by rule).
This wasn't always the case. In Old English, the forms with /v/ were derivable by rule and these words had only a single stem (ending in a voiceless fricative).
| (1) | Fricative voicing: F → [+voice] / V (L) _ V
In other words, the voiceless fricatives /f, θ, s/ become voiced between vowels, with an optional liquid consonant (/l, r/) immediately preceding the fricative. |
| (2) | Underlying form with /f/: | thief, wolf |
| Add plural suffix: | thief-es, wolf-es | |
| Fricative voicing: | thie[v]-es, wol[v]-es |
The [v] sound was spelled as <f> in Old English. As a result of contact with French, [v] became a phoneme (in other words, a sound able to distinguish minimal pairs like feel and veal). It then came to be spelled with <v>, as in French. Or even with <u>, as we know from the readings.
| Intervocalic voicing is very common in the world's languages. Can you think of examples from languages that you know? |
Further sound changes in Middle English complicated the picture.
| (3) | Final -s voicing: /s/ → [+voice] / unstressed V _
In other words, /s/ became /z/ after unstressed vowels - notably, in the plural ending. |
| (4) | Syncope: unstressed V → / C _ C
In other words, the unstressed vowel in the plural ending, spelled <e>, was generally lost between two consonants. Exception: the vowel was retained if the first consonant was a sibilant - |
| (5) | Underlying form with /f/: | thief, wolf |
| Add plural suffix: | thief-es, wolf-es | |
| Fricative voicing: | thie[v]-es, wol[v]-es | |
| Final -s voicing: | thie[v]-e[z], wol[v]-e[z] | |
| Syncope: | thie[v]-[z], wol[v]-[z] |
So far, so good, the [v] forms are still derivable by rule.
But there are two other rules in Middle English at work.
| (6) | Geminate fricative simplication: FF → F
In other words, double fricatives become indistinguishable from single fricatives. |
| (7) | Regressive assimilation: Final /z/ assimilates in voicing to the immediately preceding segment. |
| (8) | Underlying form with /ff/: | cuff |
| Add plural suffix: | cu[ff]-es | |
| Fricative voicing: | — | |
| Geminate simplification: | cu[f]-es | |
| Final -s voicing: | cu[f]-e[z] | |
| Syncope: | cu[f]-[z] | |
| Regressive assimilation: | cu[f]-[s] |
The sequence above represents the historical development, but it doesn't represent the evidence that children learning Middle English were exposed to.
As a result, all three rules (geminate simplification, final -s voicing, and syncope) were lost as productive rules. We now have the following scenario.
| (9) | New underlying form with single /f/: | cu[f] |
| Add new plural suffix: | cu[f]-[z] | |
| Fricative voicing: | — | |
| Geminate simplification: | no longer productive | |
| Final -s voicing: | no longer productive | |
| Syncope: | no longer productive | |
| Regressive assimilation: | cu[f]-[s] |
In this new situation, there are now two sorts of words with final fricatives: ones like cuff with a plural in [fs] and ones like thief with a plural in [vz]. The thief type can no longer be derived by rule. There are several possible historical developments. The two main ones are:
Which option prevails depends on several factors - chief among them, the frequency of the words in question. All other things being equal, we would expect the words with alternating stems to become increasingly rare.
Use the Google Books
N-gram viewer to track the history of the variation for:
Feel free to explore other f/v words as they occur to you. |
| (10) | Lengthen/shorten vowel in open/closed syllable: In Middle English, vowels became long in open syllables, and short in closed syllables. This change preceded syncope. |
| (11) | Underlying form: | staf |
| Add old plural suffix | sta[v]-es | |
| Lengthen vowel in open syllable: | sta:[v]-e[s] | |
| Final -s voicing: | sta:[v]-e[z] | |
| Syncope: | sta:[v]-[z] | |
| Great Vowel Shift: | stei:[v]-[z] |
This results in the regular (regular from a historical point of view) alternation staff - staves. But once the plural suffix is reanalyzed as -s, the alternation can no long be derived by rule. The case is more complex than thief/v- because of the additional vowel alternation, which becomes increasingly opaque because of the Great Vowel Shift. There were two possible solutions to the learnability problem:
In fact, both of these developments occurred, and over time the two alternatives differentiated semantically in accordance with the Principle of Contrast. (Note the parallel with the gradual differentiation of beef and cow.)
A similar development took place with cloth and clothes vs. cloths, with the additional complication that the <th> is clothes tends not to be pronounced.
In Middle English, the inflectional system of Old English breaks down, and the plural comes to be formed by suffixation.
| (12) | Underlying form: | wi:f |
| Add historically unmotivated old plural suffix | wi:[v]-es | |
| Lengthen vowel in open syllable: | --- | |
| Final -s voicing: | wi:[v]-e[z] | |
| Syncope: | wi:[v]-[z] | |
| Great Vowel Shift: | wai:[v]-[z] |
The stem vowel in the singular form was originally shortened by regular sound change, so that the singular comes to rhyme with cliff. In this case, the alternation between the short and the long vowel is resolved by analogically "undoing" the vowel shortening in the singular. This development was helped along by the fact that already in Middle English, a variant singular wife (with long /i:/ in the now open syllable) had emerged (back-formed from the new plural with -es).
What's your pronunciation for the plural for:
Is your pronunciation documented by Merriam-Webster?