I have some hypotheses for English graphotactics:

  1. 〈w〉 and 〈y〉 are optional positional variants (i.e. allographs) of 〈u〉 and 〈i〉, respectively, in digraphs that correspond with diphthongs or vowels: 〈aw〉 ≈ 〈au〉, 〈ew〉 ≈ 〈eu〉, 〈ow〉 ≈ 〈ou〉; 〈ay〉 ≈ 〈ai〉, 〈ey〉 ≈ 〈ei〉, 〈oy〉 ≈ 〈oi〉, 〈uy〉 ≈ 〈ui〉. They are the preferred allograph at the end of morphemes.
  2. 〈y〉 is a required positional variant of 〈i〉 at the end of native words, but digraph 〈ie〉 may be a possible alternate.
  3. Final 〈y〉 in a stem gets replaced by 〈i〉 when a inflection suffix follows unless it is part of a digraph: fly > flies/*flys/*flis but boy > boys/*boies/*bois.
  4. The apostrophe 〈’〉 is used to visually separate the possessive suffix 〈s〉 from proper names – i.e. words with initial capital – to ensure that #3 does not apply, so names have a constant representation.
  5. #4 is not necessary for pronouns, hence 〈its〉, 〈hers〉, 〈his〉 instead of *〈it’s〉, *〈her’s〉, *〈he’s〉. #4 is applied to other nouns as well, though.
  6. In vowel digraphs, round-top letters 〈a〉, 〈e〉 and 〈o〉 are preferred for first/left position whereas flat-top letters 〈i〉/〈y〉 and 〈u〉/〈w〉 are preferred for second/right position.

Are there any graphemic analyses of English that support these observations, especially #4?

I’m only aware of a bachelor thesis in German by Marlene Franke from 2008 which isn’t available online. It’s likely based on theories and work done by Fuhrhop/Buchmann (e.g. 2011: The length hierarchy and the graphematic syllable DOI: 10.1075/wll.14.2.05fuh) and Primus (e.g. foundational 2004: A featural analysis of the Modern Roman Alphabet), who support #6 at least.

Note that #4 is (usually) not extended to the only other possible suffix which is also an 〈s〉, i.e. the plural marker: all the Jennys and Billys.

  • 2
    +1 for the enjoyable question, but one minor remark. Shouldn't observations and analyses that support your hypotheses be supplied by the person bringing the hypothesis, and would the correct question, from the point of view of falsification, not be to supply evidence against the hypothesis? (how come I cont three +1s in the comments but only one on the question? Any politicians in the room?)
    – oerkelens
    Commented Oct 30, 2014 at 10:43
  • 4
    Wikipedia has a section about the history of it: en.wikipedia.org/wiki/English_possessive#History. It apparently derives from Old English using -es as one of the genitice suffixes, and later the e got contracted away with the apostrophe.
    – Barmar
    Commented Oct 30, 2014 at 17:48
  • 2
    @Barmar I think graphotactics should be analyzed synchronicly.
    – Crissov
    Commented Nov 10, 2014 at 19:19
  • 10
    In an orthographic system like English, which is so extremely conservative and does such a piss-poor job of representing the language it’s meant to represent, there is simply no way of analysing graphotactics synchronically. There are far too many historical convolutions for any of the rules (except perhaps #2 and #5) to be anything near consistent. Especially #1 is full of exceptions: trawl, renown vs. haul, noun; oyster vs. hoi polloi; etc. And what of the vacillation between -’ and -’s as the possessive marker after sibilants (boys’ vs. Jesus’ vs. Jesus’s)? Commented Nov 12, 2014 at 12:49
  • 5
    Then I don't quite understand why the question title deals with the reason for the apostrophe. The reason there is an apostrophe is purely historical and not related to graphotaxis at all; there is no graphotactic constraint to disallow non-pronominal possessives without apostrophes; it's a purely orthographic convention. The possessive is not an inflectional suffix, so an apostrophe is not necessary to prevent changes like -y -> -ie- (which, incidentally, is subject to a complex set of constraints itself). Commented Nov 12, 2014 at 14:26

1 Answer 1


The apostrophe is used to indicate something is missing.

So in don't the letter o is missing.

The usage is still the same for possessives: James' indicates the -es is missing from the end the word — Jameses being the correct pronunciation in this case. Similarly with Fred's except we no longer say Fredes. This happens because old English used the Germanic suffix -es to indicate possession.

The situation gets more complex for plural entities. English uses the French suffix -s to indicate plurality. There is an obvious clash with the -es possessive suffix made even more painful by the fact that this is normally rendered as a plain -s.

  • The boy's home — the boy is home
  • The boy's home — the home belonging to this boy
  • The boys' home — the home belonging to those boys
  • Or even The boys' home account — the account belonging to an institution housing boys
  • 2
    That’s the classic interpretation that relies on phonologic information to explain the apostrophe, and it’s even diachronic. I was showing that you can explain it synchronically without resorting to phonology (but morphology) and I was basically asking if someone has already come up with a more complete set of independent graphotactic rules for English.
    – Crissov
    Commented Apr 24, 2015 at 7:35

Not the answer you're looking for? Browse other questions tagged or ask your own question.