I just kern them all, it is easier than figuring out which ones to leave out. I also make a narrower set of diacritics used only on the "i" glyphs. I make special kerning classes for crashable glyphs to minimize the job..

An analogous problem in Cyrillic: "її" is a fairly common word in Ukrainian, and the combination "ії" also occurs frequently.

In French, conjugations of the verb "faire" include "fît", "fîmes", and "fîtes".

Pinyin for Mandarin with tone marks poses no problems, as the combination "fi" does not occur.

Edited to add: If, as suggested in your image sample, you consider such diacritics next to straight ascenders to be potentially problematic, then to say that you have lots more problem pairs to look at would be an understatement. In French we have "île" and "Thaïlande" just off the top of my head. I wouldn't normally consider them problem pairs, though.

Seems to be a good criterion for the quality of a font.

As it happens, without any kerning:

But thanks very much for the other pairings pointed at, I’ll check my fonts for those too.

Chris, that's great advice. I followed your example, and got away with far lower kerning values than I otherwise would have needed. Here's a quick sketch:

Thanks for that example, Brian. Looks like I won't need any fï kerning at all, but then the f of this typeface is very modest.

Related: http://typophile.com/node/66818

Andreas, the lîb and lïb collisions seldom happens in serifed fonts, as the serifs makes the glyph's body much wider. But the f + diæresis is a nightmare in many serifs. Here's how I solved that problem in one of my typefaces:

Aw, I guess I should've posted my question in that other, highly interesting thread. Thanks for pointing to it, Nina.

I did the very same for fä fö fü in Andron. But I’m not so sure if this is a good thing. The umlauts seem to have a ‘squint’. – Although your solution looks quite successful.

Earlier thread on this very topic of ligated fä, fö, fü:

http://typophile.com/node/40439

I personally don't think such merging is a good idea, for Swedish at least, and I don't think such ligatures have been attempted by Swedes. On the other hand, a number of German-speaking type designers seem to have experimented with them. Note that for Germans, ä and ö are just umlauted versions of a and o while for Swedes ä and ö are separate and independent letters in their own right.

Note that for Germans, ä and ö are just umlauted versions of a and o while for Swedes ä and ö are separate and independent letters in their own right.

That's funny, given that in German you would never use "a o u" as substitute when "ä ö ü" are not available, but I've seen that for Swedish. (In German: ä → ae, ö → oe, ü → ue).

Regarding the f+umlaut question, I repeat what I wrote in the thread Brian has linked to, and what others including Karsten and Ralf have endorsed:
Contextual alternates [a shorter f arm], that’s definitely the way to go.

Brian said:
Note that for Germans, ä and ö are just umlauted versions of a and o while for Swedes ä and ö are separate and independent letters in their own right.
I don’t see the difference, beyond nomenclature and possibly differing dictionary order. I’m curious to learn what you’re implying.

Those who are interested in the compilation of an “endangered glyph pairs” list might continue with this thread.

Jens, your observation is correct. Swedes generally don't use "ae" or "oe" when "ä" and "ö" are not available ("Anja Pärson" being known internationally as "Anja Paerson" is a noteable exception, although there the motivation may have been to avoid the spelling "Parson"). Neither do they use "aa" when "å" is not available, although I understand Danes might do so. Interestingly, Swedes may go pseudo-Norwegian and use "e" instead of "ä".

My interpretation is that the regular and systematic connection between "a o u" and "ä ö ü" is very alive in the minds of German speakers, so that given technical limitations, they find it imperative to at least fall back to indicating them as digraphs "ae oe ue", where the added "e" takes place of the umlaut diacritic. But this sort of connection doesn't exist for Swedish speakers, for whom "å", "ä" and "ö" are indivisible units. They will use a similar-looking or similar-sounding letter to replace each of these if necessary, but not a pair of letters as the Germans do.

Technical explanation about my assertion:

Umlauting refers to a phenomenon affecting vowels in Germanic languages (including German, Swedish, and English) where vowels for certain forms are fronted. In English you have man/men, foot/feet, mouse/mice, etc. In German, the spelling uses the umlaut diacritic to make this linguistic phenomenon very visible. Hence you have Mann/Männer, Fuß/Füße, and Maus/Mäuse; i.e. you have a systematic correspondence between the original and the umlauted vowel in German spelling.

Swedish borrowed "ä" and "ö" from German, I believe, and in some instances the umlauting correspondence holds, e.g. man/män, fot/fötter. But the Swedish variation of the linguistic phenomenon of umlauting being different from that of German, there are countless examples like mus/möss, lång/längre (long/longer), and falla/föll (fall/fell) to demonstrate that there is no systematic correspondence in Swedish spelling as in German. So whatever connection the umlauted vowels had with the original counterparts in German, from which these letters were originally borrowed, is completely lost in the minds of Swedish speakers.

So whatever connection the umlauted vowels had with the original counterparts in German, from which these letters were originally borrowed, is completely lost in the minds of Swedish speakers.

Not at all. The umlauted letters are very much pronounced like their German counterparts in Swedish.

The examples just show that the vowel sometimes changes in words when going from singular to plural. In Danish, some of the examples would be mand/mænd, fod/fødder, lang/længere and I could add barn/børn (child/children).

Elias, of course the pronunciations are similar because the letters were borrowed for their sounds. That's not what I am talking about. By "whatever connection the umlauted vowels had with the original counterparts in German" I mean the connection between "a" and "ä", "o" and "ö", and "u" and "ü" in the minds of German speakers.

In German, there is a systematic correspondence between these pairs. But this relationship is not well preserved in Swedish, whose own omljud has developed rather differently. The past tense of "falla" is not *"fäll" but "föll". In fact, "ö" can alternate with not just "o" but "a" (falla/föll), "u" (ljuga/ljög), and "y" (bryta/bröt), so Swedes are not likely to pair "o" and "ö" in their minds like Germans do.

For Danish, shouldn't the plural of "barn" be *"bærn" according to the strict German-style a/æ, o/ø paradigm? Well, it isn't. This is my point: the linguistic phenomenon of Germanic umlauting and the orthographical representation of it using the umlaut diacritic matches up well only for German.

Sorry, Jongseong, I misunderstood you point if your point being that ö and ä are seen as separate letters in Swedish and as "just" umlauted o and a in German (as a consequence of Swedish having a more irregular grammar, not pronounciation). I suppose you're right on that one, but maybe we can hear some Swedes/Germans about their opinion. Anyone?

I think Jongseong is basically right in his assertion about Swedish "umlaut" vowels. The use of aa, ae, oe instead of å, ä, ö I think was more common back in the early days of computing and internet, when need sometimes arose to actually write rather long texts without the proper letters. In such cases I guess unambiguity takes precedence over aesthetics. Nowadays there is rarely any reed to resort to this except in shorter texts such as e-mail addresses. My Typophile user name is Mans, though I could probably just as well have used the correct Måns. :)