FORCE, NORTH, CLOTH, and START

Lexical Sets
Author

Joey Stanley

Published

October 24, 2025

Modified

October 27, 2025

I am doing some research on the cord-card merger in Utah, which means I need to do analyze lexical sets Wells calls force, north, cloth, and start. Like other lexical sets I’ve written about, in the absence of comprehensive list of lexemes belonging to these sets, I have to cobble together what I can from various sources and create the list myself. In this post, I’m primarily basing it off what Wells (1982) says in his definitions of the lexical sets.

1 force

Definition: The force set is defined as words that are /ɔː/ in British English but /oɹ/ in American English. Not all /ɔː/ words in British English are force. Some may be thought, north, and maybe cure and cloth. Since I’m not focused on British English here, I won’t bother with those details.

Historical Roots: Most force words derive from Middle English /ɔː/ before /ɹ/, which is the same source as goat. I take this to also mean that Americans “think” of force words as having the same vowel as goat, and I think that would show up if they had to do a phonological categorization task. However, some force words some derive from Middle English /oː/ or /uː/, although I’m not sure which ones those are or how to tell.

Distribution: Words that are spelled with -ore, -oar, -oor, and -our are force unless we know they’re cure. Only -or is ambiguous with north. force may be word-final (store, boar, door, four), before consonants (fort, sword, torn, portrait, hoarse, court), or before vowels (choral, story, historian, Victorian, uproarious).

Tip

Here are some additional force words. I’ve put them in order that they occur in my Kohler Tapes corpus and their frequency per million words.

more 889.874 floor 177.475 report 43.327 source 14.165
fore 701.567 wore 121.233 shore 41.244 score 11.665
store 641.992 force 118.733 support 37.911 divorce 10.832
four 537.007 chore 92.487 port 37.078 glory 10.415
story 447.853 sport 79.572 pork 26.246 moore 7.082
course 374.114 court 75.406 bore 22.497 core 6.666
door 336.202 tore 62.491 authority 19.997 restore 6.666
board 207.887 fort 59.158 sore 19.164 orient 6.249
ford 195.806 pour 45.410 enforce 15.831 roar 5.416

2 north

Definition: The north set is defined as words that are /ɔː/ in British English but /ɔɹ/ in American English varieties that don’t merge force and north. If the following sound is a vowel (as in horrible, orange), it belongs to cloth.

Distribution: Most north words are spelled with or and are preconsonantal (fork, cord, storm, mortal, orbit, corner). Some are spelled with a but it appears to me that all of them are preceded by /w/ (quart, swarm, ward, warm, wart). North is very rare word-finally, and a potentially exhaustive list is or, for, nor, Thor, and war. There are also very few in open syllables, but they are all spelled with au and are aura, aural, Laura, and Taurus. See below on other words with au.

Tip

Here are some additional north words. I’ve put them in order that they occur in my Kohler Tapes corpus and their frequency per million words.

horse 954.864 corner 108.318 corn 41.661 reward 10.415
war 870.293 warm 106.235 horn 41.661 warn 9.165
morning 345.368 quarter 105.818 fortune 41.244 warden 6.666
north 322.038 forty 102.069 storm 37.078 cork 3.749
ward 220.802 lord 85.405 norm 29.163 board 2.083
important 197.889 mormon 59.991 record 24.163
short 191.223 organ 57.075 resort 18.331
george 166.226 form 46.243 award 12.498
order 124.149 fork 43.744 cord 12.082

3 cloth

Definition: The cloth set is defined as words that are merged with lot in British English but with thought in American English. Non-prerhotic cloth typically has a following voiceless fricative, like in coffee, soft, moss, or often, which is the result of a seventeenth-century lengthening rule. These words are almost always spelled with o, but may have an ou (like trough and cough) or au in proper nouns like Austria and Austin. Cloth may also appear before velar nasals like long and song. Notably, Wells does not list any pre-/ŋ/ words with lot or thought, so it seems like any low vowel followed by a velar nasal belongs to cloth. Pre-/ɡ/ words are not listed in a lexical set; they are discussed in Volume 3 §6.1.3 (which I don’t have).

Relevant to this study, cloth also occurs before intervocalic /ɹ/. These include words like horrible, Florida, and orange which may be variably lowered in some varieties of American English, cheifly in the Northeast, even if those varieties don’t otherwise have variation in north. This small list of prerhotic cloth is this set of words that are listed in the Wikipedia page on English-language vowel changes before historic /ɹ/ as being lowered in “Traditional American” which, in a footnote, is defined as NYC area, Philadelphia are, older Southern, older Northeastern elite, and maybe older people in Eastern New England or coastal southern regions.

Distribution: Most of the time, cloth is spelled with o, but sometimes it is with a like in warrant, quarrel or warrior. Presumably the preceding w is relevant since all the examples spelled with a have it. In a few cases (for reasons also unclear to me), it can also be with au like in Laurence and laurel. While Canadians typically pronounce these pre-rhotic cloth words with [ɔɹ], it is normal for Americans to lower words like sorry and tomorrow, even if they don’t otherwise have variation in cloth.

Tip

Here are all the cloth words in my Kohler Tapes corpus, with their frequency per million words.

orange 70.407
forest 45.827
quarantine 20.830
tomorrow 14.165
horror 2.083

4 start

The start lexical set is defined as words that are /ɑː/ in British English but /ɑɹ/ in American English. Not all British /ɑː/ are start; some are bath and palm, but I won’t dive into those here. Wells says that in American English start patterns with lot phonetically, but in my variety of English, I certainly feel like it patterns more with thought.1

1 I will also add that, unlike the others here, my variety of English has a fair amount of allophony in start. Read more about that here.

Historical Roots: Start derives from Middle English /aɹ/, which was affected by a pre-rhotic lengthening rule. Note that this vowel is the same source as lot.

Distribution: It is usually spelled with -ar but -ear and a few others are possible. It can be found word-finally (star, bar), before consonants (sharp, arch, card, farm, party, heart, hearth, sergeant, aardvark). It is rare in open syllables and all the example listed by Wells are borrowings (safari, aria, tiara).

Tip

Here are some additional start words. I’ve put them in order that they occur in my Kohler Tapes corpus and their frequency per million words.

car 1152.753 charge 87.071 barbara 23.33 chart 9.165
start 1004.024 marble 76.239 marvel 22.913 spark 8.332
farm 720.731 star 71.24 argue 21.247 carton 7.916
hard 613.246 guard 63.741 carnival 20.414 shark 7.916
part 368.281 harbor 63.324 darn 19.581 barge 7.082
far 335.369 mark 60.825 barbecue 19.164 regard 6.666
park 321.204 charles 58.325 martin 19.164 marshmallow 5.416
party 228.718 market 54.576 carpet 17.498 parliament 4.999
charleston 221.635 smart 50.41 bark 17.081 barb 4.583
barn 185.39 article 47.91 partner 16.248 carve 4.583
garden 185.39 department 47.91 sergeant 16.248 charter 4.583
army 163.727 art 44.577 marjorie 15.414 cigar 4.583
yard 156.228 bar 42.911 starve 14.998 sharp 4.166
large 150.812 barber 35.412 scar 14.581 harm 3.749
arm 138.73 apartment 33.745 remark 13.331 garment 2.916
dark 129.982 harvest 30.829 artist 12.915 arlen 2.083
march 108.735 borrow 29.996 arch 12.498 depart 2.083
heart 102.485 cart 26.246 jar 10.832 enlarge 2.083
card 89.571 marshal 24.58 target 9.582

5 Guidelines

With that said, here are some guidelines I’ve come up with, based on my reading of Wells.

  • If it is spelled with an a and has a preceding /w/, it has to be either north or cloth. From there, it’s a matter of whether it’s in an open syllable or not (with open-syllables being cloth).
  • If it’s spelled with ar and is not preceded by /l/ or /w/, it’s start
  • If it’s spelled with oo or ou, check if it’s cure. If Wells doesn’t list it or you can’t otherwise find evidence that it’s cure, code it it as force.
  • Other words spelled with ou, ou, oa are force.

Words spelled with au could be force (aura, Laura), north (Austria, Austin), or cloth (Laurence, laurel). As far as I can tell, there is no pattern.

For words spelled with or, here is what I see:

  • If it’s word-final, it’s force unless it’s one of the rare north words that Wells lists.
  • If it’s in an open syllable, it’s also force unless it has already been tagged as one of the known cloth words.
  • Any remaining words (preconsonantal environment and spelled with or) are unpredictably force or north and you’ll have to check with a native speaker or consult a pronunciation dictionary that maintains the distinction.

If you are a native speaker of a variety that maintains a distinction between north or cloth, I’d love for you to reach out so I can double check my guidelines and classifications.