My last email was an objective list of facts. This one is my opinion. I'm using English Hunspell, with continuous spell checking turned on.
> From: Jim Oldfield <jim_...@yahoo.co.uk> > > here is what Word does in English > > (1) Words with an Initial Capital are considered as normal, except that > proper nouns are allowed. For instance Cat and Jim are correct, whereas Jmi > and jim are wrong. Even the grammar checker does not complain about Cat in > the middle of a sentence. Two issues have come up on the list: (1a) (original poster) Should we ignore capitalised words altogether, since the dictionary doesn't contain them? The reply was "that's your dictionary's fault", which I agree with. When replying to such users in future, it's worth reassuring them that adding "Jim" to their personal dictionary won't make "jim" count as correct. (1b) (raised by devs) Should we make non-proper nouns in the middle of the sentence (like "a big Cat") wrong? My opinion is no! If even Word's grammar checker doesn't try to figure this out, I think it's very ambitious for LyX's spell checker to try (admittedly LyX has a better chance of figuring out what counts as a heading). I've never been in a situation where I've accidentally typed a capital and wished the spell checker had told me. Plus, since there's no "ignore once" in LyX, once it's marked incorrectly you're stuck with it. In summary, I like LyX as it currently is WRT (1). > (2) Words in ALL CAPS are by default ignored completely. If the option is > turned off then proper nouns are allowed i.e. JIM and CAT are correct. > Either way, as with Initial Capital there is no complaint about using all > caps, even from the grammar checker. Personally I dislike the default in Word, and I've been caught out by it before. LyX behaves like Word with that option turned off. Again, I like LyX as it currently is WRT (2). > (3) Words separated by *hyphens* are spell-checked as separate words. The > problem that triggered this discussion (socio-technique, and similar ones > like pseudo-differential, ex-wife etc) is avoided because all common > prefixes (including these ones) are considered correct, even when not before > a hyphen. This is how LyX behaves for me. Indeed the original "socio-technique" works fine for me. Again, perhaps the original user just has a bad dictionary. Again, I like LyX as it currently is WRT (3). There is one exception: leading and trailing hyphens and non-breaking hyphens are not ignored. This is a problem because something like $n$-dimensional is marked as incorrect. > (4) Words separated by *full stops* (AKA periods) and similar are > spell-checked as whole words. For instance cat.Jim is marked wrong. > Obviously trailing full stops are allowed, but sometimes they are necessary. > For instance "eg", "eg." and "e.g" are wrong, whereas "e.g." is correct. This is where LyX differs. Full stops are treated like (3), so cat.Jim is marked as correct. This is problem for English, because words like "e.g." aren't checked properly. My dictionary seems to have individual letters in it, so at least "e.g." isn't marked as wrong, but "i.e." is since "i" is wrong (with "I" suggested instead). So here I would prefer LyX to be changed to match Word's behaviour (at least to consider internal full stops, even if optionally matching trailing ones is too hard). Jim
