Why Santa is krocking at your door

The short answer as to why Gaelic words that begin with cn- are pronounced [kr], for example, cnoc [krɔ̃xk], is that that's just the way it is.

That's actually not a trivial point to make because you'll find all sorts of folk pontificating about this issue - most of whom have no clue. The Survey of Scottish Gaelic Dialects which recorded people from Lewis to Barra, Strathy to Aran, Dunbeath to East Perthshire (in total, 207 informants when those areas still had local Gaelic speakers) actually recorded several items with cn-, including cnoc. There's not a single incident of this being anything but cnoc [krɔ̃xk] and /kr/ was reflected in the other cn- items cnò, cnap, cnàimh and cnuimh.

There is one variation for one of the three St Kilda speakers, informant 15, who had /kl/ in cnuimheag, but the other two St Kilda speakers had /kr/.

So, to begin with, we can be crystal clear that whatever the pronunciation was at some point in the days gone by, it is now always /kr/, usually with nasalisation on the nearby vowel.

It's also interesting to note that Manx and most of Ireland also have /kr/. Only Kerry Irish retains /kn/. That, along with clues from the way words with cn have been spelled and mis-spelled over the centuries, tells us that once it was /kn/ but that the movement towards /kr/ began a long time ago.

As to why... keeping the explanation simple, phonetically /r/ and /n/ are fairly close to each other. Don't be alarmed, here's a crazy looking linguistic schematic that will demonstrate just how close they are:

Sound	consonantal	syllabic	sonorant	coronal	anterior	high	low	back	nasal	lateral	continuant	tense	voice
[r]	+	-	+	+	+	-	-	-	-	-	+	-	-
[n]	+	-	+	+	+	-	-	-	+	-	-	-	-
[p]	+	-	-	-	+	-	-	-	-	-	-	-	-

You don't need to know what all these features. They're labels that linguists use to categorise the mechanics of a sound. What should be obvious is that except for two features, [r] and [n] are produced exactly the same way. Remember what I said about there usually being nasalisation near /kr/? That's the leftover from the nasal feature. So, if we take that into account, /knɔxg/ and [krɔ̃xk] are actually identical except for one single feature. The more features that overlap, the easier it is for sounds to jump.

In contrast, if you look at [p], it shares very few features with [r] or [n] so the chances of [knɔxg] turning into *[kpɔxg] are very slim indeed.

So, that explains why /kn/ moving to /kr/ is plausible and I'll leave the explanation at that. Sometimes such closely related sounds just flip in languages.

Can you give me a date?

Tricky one. Timing such changes is not easy but not totally impossible either. What linguists usually do in this situation is look at loanwords, typos and authors who did not use the standard orthography but made up their own. At a casual glance, the following are helpful in our case:

Manx orthography which is a heady mix of the phonemic and the insane, was fixed in the 16th century and retains almost no trace of cn- mn- etc spellings e.g. cronk (cnoc), craue (cnàimh), grooish (gnùis)... So it is fairly safe to assume that by the time this orthography was devised, there was little or no trace on the Isle of Man of /kn/.

In Scotland, the Fernaig Manuscript (dated to circa 1688), written in non-standard orthography paints a mixed picture, we have:

• vo ni mrahi : bho na mnàith
• Gin chnodigh : Gun chnòdach
• si chroighk : sa chnoc
• gna gi soyller : ghnàth gu soilleir

Now we have to tread carefully. The presence of <n> spellings is no guarantee it was there in speech. The author may have been trying to adhere to elements of the formal spelling or what may have been considered "correct" speech. On the other hand, at a casual glance, I cannot find traces of hypercorrection i.e. the author taking a word which actually has <r> and writing it with <n>, for example there is no case of *bniven (for britheamhan).

It also appears occasionally in the Dean of Lismore, for example as mrave (for mnáibh).

It tells us this with a fair degree of certainty: by 1688 [kr] etc were sufficiently established in spoken Gaelic to slip into this type of spelling and it was likely more common that the text suggests because people tend to write conservatively in that period. Looking into eDil, there are no Old/Middle Irish incidents of croc type spellings at a casual glance.

Last but not least, the evidence from Irish shows that possible the first evidence of this change goes back to a place name recorded in 1551 as Gryeve na managh (for Gníomh na Manach). Between 1551 and the end of the 17th century evidence is very patchy on the Irish side. For example, the majority of names recorded in English spelling (like the 1603 form Owen atnaght, for Eóghan an tsneachta) show no trace of [r] while on the other hand there is the instance of Cill Mic Crénáin (instead of Cill Mac nÉnáin) in Keating's manuscript which is dated around the end of the 16th and the start of the 17th century.

So putting all this together, i.e. the total absence of cn- etc in Manx by the 16th century and the presence of [kr] etc in the Fernaig manuscripts and the (seeming) absence in Old/Middle Irish, the change probably starts appearing in spoken Irish, Manx and Gaelic somewhere in the Early Modern Irish period (1200-1600) - when a lot of stuff went sideways in the language - but doesn't actually make it into the written record, however faintly, until the 16th/17th century.