I've had requests from Andy J and (long ago) Doug Sundseth to cover this one. Here's an excerpt from Andy's recent email on the topic:
Before I go into the history of the word, I want to do a little bit of "here's how a linguist thinks". Andy's got(ten) us started along the right lines here, in that (a) he talks about variation, rather than deviation, and (b) he looks for broader patterns. It's important to look for the broader patterns because we know that:
We can illustrate those two points with the /r/ after vowels. In my inland northern American accent, I would pronounce the -er in solder with an /r/. In my spouse's London accent, he would pronounce it as an unstressed vowel /ə/ —no /r/. That difference carries on to every word that ends in -er (and every other r that follows a vowel, actually). If we look at that spelling and those pronunciations, we are well justified in thinking that earlier English pronounced the /r/ there, and the English of southern England later stopped pronouncing it. Otherwise, why would all those r's be there in the spelling? And indeed, that's the case.
But in this case, as Andy notes, there is no evidence of a regular sound change. Most Americans don't pronounce an /l/ in solder, but if there had been a sound change that got rid of /l/ after a vowel or before a /d/, then Americans should pronounce folder as "fodder" and soldier as "sodyer", and Americans just don't do that.
In the absence of evidence for a regular sound change, we have two possibilities:
The OED gives the following historical spellings of the word (the numbers indicate the centuries in which you see those spellings):
The Old French word that solder comes from is represented in the OED etymology as:
So what about the two L-ful paths? There are (again) two possibilities (plus the possibility that it is both of these to different degrees/in different places):
But (2) is more likely, judging from the clear history of sentimentality for Latin affecting English spelling. Here's an article by Arika Okrent on weirdly spelt words, and indeed she includes solder in the same category as debt and receipt, as victims of re-Latini{s/z}ing in the 15th and 16th centuries. The L got added into the spelling, and then later, people started pronouncing it as a spelling pronunciation.
We've seen a similar story for herb: the spelling got Latini{s/z}ed, and the English (eventually) went for a spelling pronunciation, but Americans carried on with the old pronunciation.
When did the spelling-pronunciation shift happen? After America had had its English from England (mostly). The OED notes that Smart's 1840 pronouncing dictionary (from England) included only the /l/-less pronunciation, but it looks like this was very much a 'live' problem in the 18th and 19th centuries (when Englsih had been burbling along in America for over 100 years). The 1824 edition of Walker's Critical Pronouncing Dictionary said this:
If you can't read that: the key point is that Johnson's dictionary of 1755 preferred spelling it without the L, but the L spelling was already well established. It acknowledges differences in opinion about the pronunciation among orthoepists [pronunciation prescribers] and that the L-less pronunciation was used by workmen, but "workmen ought to take their pronunciation from scholars and not scholars from workmen". (Ah, social class in England...)
Solder does seem to be exactly the kind of thing whose spelling would revert to older form in AmE, so it's a little surpising we don't spell it sodder. Noah Webster did try to change it. At solder in his 1828 dictionary there is a cross-reference to soder, which reads:
It's no wonder soder didn't catch on, since it looks like it should have the same first syllable as soda. If only Webster had doubled the d.
While I've been known (to myself) to misspell it as sodder, that spelling hasn't had much traction in AmE, and neither has Webster's, as can be seen in numbers from the Corpus of Historical American English. (The Soders in the 2000s here are all someone's name.)
Interestingly, for those who find such things interesting, the addition of L to an L-less French borrowing is also why we have an L in salmon (French saumon, Latin salmo(n)), but there's been no big movements toward(s) pronouncing that L in English. This just goes to show that spelling pronunciation changes are not regular changes.
And I expect someone will have calm on their mind now. That one's pronounced with no /l/ in England but some Americans do have an /l/ in it. The vowels differ in these cases, but then most of our vowels differ, don't they? I believe my own calm varies from pronunciation to pronunciation (and probably did so even before I moved to the UK). Calm differs from solder in that it came into English from French with its L. However, it looks like not everyone was pronouncing it, since there are some caume/cawme spellings in the 1500s and 1600s.
This seems to be a case of the /l/ being lost because it's in a phonetically complicated place—between two other sonorous elements. An /l/ after a vowel/at the end of a syllable is pronounced differently than one at the front, and that back-of-the-syllable "dark /l/" often does strange things, especially in combination with other consonants. You can see (or hear) in Irish and Scottish English the evidence that /l/+consonant combinations often feel a bit unnatural. Those Englishes often sort out /l/+consonant by inserting a vowel between the consonants, which "un-darkens" the /l/. Film→fi-lum, Colm → Co-lum (you can hear that at 1:50 in this Derry Girls clip, and any excuse to watch Derry Girls should not be snubbed). English English (and French before it, it seems) has sorted this out by just not pronouncing the /l/. Whether some Americans have added it back in as a spelling pronunciaiton, or whether the /l/ came over as the original pronunciation and stayed, I'm not sure.
I've been careful to say "England" and not "BrE" in this post, since we're talking about pronunciations and they can vary more than spellings. I've only gone with the pronunciations in the OED, so your mileage may vary. It would be interesting in particular to hear about Scottish and Irish pronunciations in the comments, since they do interesting things with /l/+consonant combinations. But also please let me know if you know of variations within England or elsewhere.
I watch a lot of Youtube videos which feature people who self-describe as makers (part DIYers and part semi-professional craftsmen and women). I have noticed that also without exception those based in the USA and Canada pronounce the word solder as sodder, whereas we BrE speakers would invariably sound the L in both the noun and the verb solder.
The North American variation seems at odds with the similar phonic construction in soldier or for example folder which, to my ear based on film and TV utterances, seem to be pronounced in a largely similar way to BrE, ie the L is sounded.
Before I go into the history of the word, I want to do a little bit of "here's how a linguist thinks". Andy's got(ten) us started along the right lines here, in that (a) he talks about variation, rather than deviation, and (b) he looks for broader patterns. It's important to look for the broader patterns because we know that:
- Where spelling clashes with pronunciation (that is, where spelling is not phonetic), the spelling often gives clues for finding an earlier pronunciation.
- Linguistic sound changes are very often regular. That is to say, they apply across all words that would be susceptible to that change.
We can illustrate those two points with the /r/ after vowels. In my inland northern American accent, I would pronounce the -er in solder with an /r/. In my spouse's London accent, he would pronounce it as an unstressed vowel /ə/ —no /r/. That difference carries on to every word that ends in -er (and every other r that follows a vowel, actually). If we look at that spelling and those pronunciations, we are well justified in thinking that earlier English pronounced the /r/ there, and the English of southern England later stopped pronouncing it. Otherwise, why would all those r's be there in the spelling? And indeed, that's the case.
| Pic from (AmE) Jewelry Making Daily |
In the absence of evidence for a regular sound change, we have two possibilities:
- the /l/ is not an original part of the pronunciation, but people started pronouncing an /l/ because they saw it there in the spelling. This happens often enough that we have a name for this kind of sound change: spelling pronunciation.
- the /l/ is an original part of the pronunciation, but for some idiosyncratic reason, someone started pronouncing it without the /l/ and that caught on. That can happen too.
The OED gives the following historical spellings of the word (the numbers indicate the centuries in which you see those spellings):
α. ME soudur, ME soudure, soudour, sowdur, sowdowre; ME soudre, ME–15 souder, ME–16 (18 dialect) sowder (ME sowdere, 15 soweder); 18 dialect sowther. β. ME sawdur, sawdyr, 15 sawyer; ME sawd(e)re, 15 sawder (16 sawter), 15–16 saudre, 16 sauder. γ. 15–17 soder (16 soader, sodar), 16– sodder; 15 sother, 16 soather. δ. ME souldour, 15–16 soulder (15 sowl-). ε. 16 soldure, 16– solderI've highlighted the five paths that the spelling seems to follow (indicated by the Greek letters). Why five paths? Because language is a moveable, social thing. The word shows up in English in the period when English was getting a lot of vocabulary from France (after the Norman Invasion and all that). But words don't have to just show up once. And once they do show up, they don't stay the same.And when they change, they can change in different ways in different places.
The Old French word that solder comes from is represented in the OED etymology as:
soud-, saud-, soldure (compare Italian saldatura ), souder , etc.,Three of the paths are L-less (and these are the paths for which the OED has more examples—so the L-less spellings were more widespread. That's because it came into English without an /l/ sound because it mostly didn't have one in French. The Italian comparison word that has an L tells us that there's a fair chance that the French came from a Latin word with an /l/, which the French subsequently lost. And that's indeed what we find: the Latin etymon is solidare 'to make solid'. Both French and Italian dropped the Latin word's second syllable, but French did it by losing a consonant and Italian by losing a vowel.
So what about the two L-ful paths? There are (again) two possibilities (plus the possibility that it is both of these to different degrees/in different places):
- Maybe some of the people who brought the word to England did pronounce an /l/ in it, and so the spelling reflected that. Note the soldure spelling that existed in Old French.
- Maybe some scribes started inserting an L because they knew the word came from Latin and they wanted to hono(u)r its Latin roots.
But (2) is more likely, judging from the clear history of sentimentality for Latin affecting English spelling. Here's an article by Arika Okrent on weirdly spelt words, and indeed she includes solder in the same category as debt and receipt, as victims of re-Latini{s/z}ing in the 15th and 16th centuries. The L got added into the spelling, and then later, people started pronouncing it as a spelling pronunciation.
We've seen a similar story for herb: the spelling got Latini{s/z}ed, and the English (eventually) went for a spelling pronunciation, but Americans carried on with the old pronunciation.
When did the spelling-pronunciation shift happen? After America had had its English from England (mostly). The OED notes that Smart's 1840 pronouncing dictionary (from England) included only the /l/-less pronunciation, but it looks like this was very much a 'live' problem in the 18th and 19th centuries (when Englsih had been burbling along in America for over 100 years). The 1824 edition of Walker's Critical Pronouncing Dictionary said this:
| Click to embiggen |
Solder does seem to be exactly the kind of thing whose spelling would revert to older form in AmE, so it's a little surpising we don't spell it sodder. Noah Webster did try to change it. At solder in his 1828 dictionary there is a cross-reference to soder, which reads:
| click to embiggen |
It's no wonder soder didn't catch on, since it looks like it should have the same first syllable as soda. If only Webster had doubled the d.
While I've been known (to myself) to misspell it as sodder, that spelling hasn't had much traction in AmE, and neither has Webster's, as can be seen in numbers from the Corpus of Historical American English. (The Soders in the 2000s here are all someone's name.)
Interestingly, for those who find such things interesting, the addition of L to an L-less French borrowing is also why we have an L in salmon (French saumon, Latin salmo(n)), but there's been no big movements toward(s) pronouncing that L in English. This just goes to show that spelling pronunciation changes are not regular changes.
And I expect someone will have calm on their mind now. That one's pronounced with no /l/ in England but some Americans do have an /l/ in it. The vowels differ in these cases, but then most of our vowels differ, don't they? I believe my own calm varies from pronunciation to pronunciation (and probably did so even before I moved to the UK). Calm differs from solder in that it came into English from French with its L. However, it looks like not everyone was pronouncing it, since there are some caume/cawme spellings in the 1500s and 1600s.
This seems to be a case of the /l/ being lost because it's in a phonetically complicated place—between two other sonorous elements. An /l/ after a vowel/at the end of a syllable is pronounced differently than one at the front, and that back-of-the-syllable "dark /l/" often does strange things, especially in combination with other consonants. You can see (or hear) in Irish and Scottish English the evidence that /l/+consonant combinations often feel a bit unnatural. Those Englishes often sort out /l/+consonant by inserting a vowel between the consonants, which "un-darkens" the /l/. Film→fi-lum, Colm → Co-lum (you can hear that at 1:50 in this Derry Girls clip, and any excuse to watch Derry Girls should not be snubbed). English English (and French before it, it seems) has sorted this out by just not pronouncing the /l/. Whether some Americans have added it back in as a spelling pronunciaiton, or whether the /l/ came over as the original pronunciation and stayed, I'm not sure.
I've been careful to say "England" and not "BrE" in this post, since we're talking about pronunciations and they can vary more than spellings. I've only gone with the pronunciations in the OED, so your mileage may vary. It would be interesting in particular to hear about Scottish and Irish pronunciations in the comments, since they do interesting things with /l/+consonant combinations. But also please let me know if you know of variations within England or elsewhere.
