The American Language in 3000 AD

2003+ Justin B Rye


Predicting the future of the English language is rather easy, in the short term.  The odds are, over the next few decades its New World dialects are going to gain increasing global dominance, accelerating the demise of thousands of less fortunate languages but at long last allowing a single advertisement to reach everybody in the world.  Then after a century or two of US dominance some other geopolitical grouping will gain the ascendancy, everyone will learn Chechen or Patagonian or whatever it is, and history will continue as usual.  Ho hum.  But apart from that… what might the language actually look like in a thousand years time?  For comparison, the English spoken at the turn of the last millennium looked like this:

1000 AD: Wé cildra biddaþ þé, éalá láréow, þæt þú tǽce ús sprecan rihte, forþám ungelǽrede wé sindon, and gewæmmodlíce we sprecaþ…
2000 AD: We children beg you, teacher, that you should teach us to speak correctly, because we are ignorant and we speak corruptly…

(From the Colloquy of Ælfric.)  So how far will another thousand years take it?  I've already got pages about time travel and languages in SF, plus a conlang of no very specific origin; this addition, vaguely inspired by the precognitive Darwinism of Dougal Dixon's “After Man: A Zoology of the Future”, should fit in nicely.  It has also now acquired companion pages titled Pleistocenese, Alternese, and Europan.

2013 POSTSCRIPT: for its tenth anniversary (we're one percent of the way there already, folks!) I have finally updated this page to use Unicode for its phonetic symbols instead of seven‐bit ASCII workarounds.  It has taken until now for browsers to support “stacked” diacritics at all reliably, and the results can still be rather ugly!  In the process I have changed my notation slightly to take advantage of some of the more appropriate glyphs now available.

LANGUAGE SF – Futurese Bibliography

Before I start developing a “future history” of my own I'll run through a quick survey of the existing literature.  It's a bit sparse, though, since academic linguists know better than to try, and nobody else has ever shown much interest – except of course the supporters of language‐planning projects like Esperanto or Basic English, which are a bit off‐topic (though they did inspire George Orwell to produce one famous vision of the language of tomorrow).  Most genre Science Fiction ignores linguistic barriers between centuries just as it does all the other kinds – reasonably enough, since they get in the way of the plot – but a handful of stories can be picked out as featuring representations of “Futurese”:

Next Year's Slang
Most works of SF feature at least a few neologisms, slang terms for cyborgs, or the like, but few authors take it to the extreme of writing the whole novel in argot, as Anthony Burgess did for “A Clockwork Orange” (“viddy this, my droogs”) – and even he didn't introduce any grammar or pronunciation shifts to go with the new “Nadsat” vocabulary items.  Heinlein took a more reader‐friendly approach for the Loonie dialect in “The Moon is a Harsh Mistress”; it gives a good impression of being slangy and futuristic, but when you stop and look at it there's nothing there but a few loanwords and telegraphese mannerisms.
Post‐Holocaust Vowel Mutations
Apocalyptic futures and lost‐colony‐world settings are often studded with suspiciously familiar words (forbidden deserts named “Neorksiti” and the like) – which always leaves me uncomfortably aware that vowels have a short halflife compared to radioactive wastelands.  Russell Hoban's “Riddley Walker” deserves a mention as another SF novel written entirely in an imaginary dialect; this time it's a more generally mangled form of post‐nuke English, though it's still closer to the modern standard language than plenty of books written in real UK dialects!
Galactic Empires
Space Opera yarns occasionally mention that everybody is speaking a remote descendant of English, called either (for some reason) “Anglic” or less often something like “Galanglic” or “Galach”.  The name tends to be as much as we learn, unless the footnote is bulked out with a claim that some other present‐day language contributed a lot of vocabulary items – Russian, Spanish, and Japanese being popular choices.
Time Travel
David Masson's short story “The Transfinite Choice” is the only one I can think of where the temporal language barrier is illustrated with a few sentences of vaguely credible future (British!) English – for instance, the displaced hero is referred to as an undrowda (i.e. “intruder”).

2022 ADDENDUM: last year's reread of “classic” science fiction unearthed some further examples worth mentioning.

LANGUAGE CHANGE – Progress and Decay

Let me get one thing clear: there's nothing wrong with languages changing over time.

When looking at a biological “family tree” (such as the evolutionary history of the horse), the general public insists on seeing any movement as intrinsically “progressive”, moving from “primitive” to “advanced” designs.  Yet somehow when looking at the linguistic equivalent (such as the development of the Romance languages from Vulgar Latin) they see exactly the reverse – any change is proof that the language is in decline.  In reality they're just as wrong both times!

The attitude is perfectly understandable; membership of a linguistic community is an important social marker, so people often get neurotic about the way they speak, and cling to the security blanket of vaguely remembered schoolroom mandates, despising those barbarians who split infinitives or mispronounce “shibboleth”.  Ironically, it's this same group‐membership effect that's responsible for many of the changes (see below), but the degeneration the purists warn against is an imaginary danger anyway.  English has gone from being a minor Germanic tongue on Europe's fringe, with a vestigial system of inflections signposting case, mood, gender, and so on, to being a much more weakly inflected language dominating the global landscape.  Every step of the way, old fogeys moaned that it was going to the dogs; but although the noun‐gender system of Old English has crumbled away entirely, it turns out not to have been a structural support in the first place… and the simplifications have been balanced by increased complexity in other places, such as in the sheer size of the vocabulary.

Changes can occur in every aspect of a language:

Words can be lost or shift their meanings, and new ones can be derived from a variety of mechanisms.  I could fill a couple of extra paragraphs with examples of borrowings, acronyms, and the like, but it would be a waste of effort if you already grok the way common parlance can be groovy one year and naff the next.
Or in strict linguistics‐jargon terms, morphology (word‐building) and syntax (word‐arranging), which manage the Escheresque trick of getting simpler and simpler until they end up just as complicated as ever – on the one hand “whom” becomes “who”, but on the other new constructions arise like “y'all ain' gonna‐hafta”.
You'll be aware of individual words that have two alternative pronunciations, one of them “proper” but endangered and the other “wrong” but spreading (such as “pronunciation” vs. “pronounciation”); but in fact the most significant changes are the ones that happen to particular sounds right through the dictionary, like dropping aitches or lengthening stressed syllables.

These different types of language change don't happen in isolation – the blurring of word‐final sounds erodes grammatical features, the development of new ways of stringing syllables together triggers shifts in pronunciation, and so on.  Nevertheless, my futurological efforts will be based purely on projected sound changes, since they tend to be astonishingly regular and thus offer the easiest opportunities for mock‐ups of Futurese.

LANGUAGE TRENDS – Causes and Effects

There's a widespread popular assumption that modern technology (gramophones, cinema, CNN etc.) will stop languages changing in the new millennium, because these days everybody knows what everybody else's accent sounds like.  But accents such as Cockney never did arise because working class Londoners were unaware of how the aristos talked.  They knew perfectly well; but that wasn't the accent they grew up with, and there was no reason to want to imitate it when their own accent was a badge of solidarity with their peer‐group.  Nothing has happened to reduce the allure of a distinctive way of speaking as a badge of in‐group membership; and the more positively people identify with some particular accent, the more likely that high‐status speech variety is to drift, as social climbers refine their vowels while the native speakers react to being imitated by innovating further.  Linguists studying modern “Network English” find that it has several regional subvarieties, which are diverging rather than converging.

That's not to say that technology has no effects.  For a start, when the global media bring linguistic communities into contact with one another, that can have all sorts of unforeseeable results – for instance, we loaned the Japanese the words “walk” and “man”, and got them back compounded.  The opportunities for interactions like that will inevitably increase as the number of non‐native speakers of English continues to rise.

Over the centuries, language change has been affected in various minor ways by innovations such as the printing press (there were no spelling‐based pronunciations such as “almond”‐with‐an‐L until there were misleading standard spellings), and of course Chaucer didn't have a word for “helicopter”.  It's easy to imagine other technological developments that might have further‐reaching effects in the future:

I'm going to have to leave possibilities such as this for another day – not only because they raise questions about the real likelihood of 3000 AD Earth being inhabited by hominids that still bring their young up to speak a traditional wild‐grown language but also because they don't make the language's future form any more predictable.

On the other hand, some factors do show long‐term directional influences.  An obvious one is ease of use: people won't bother saying “omnibus” when “bus” will do, or “environment” when their friends are getting away with “emviromment”.  But another factor is that the language has to work as a language; any change that impedes communication spurs the development of workarounds – so, for instance, people who pronounce “pen” and “pin” indistinguishably soon start talking about “ink pens”.  And a third, less obvious influence is ease of learning.  Children forming their initial mental model of how English works don't want to believe it's a mess of random idioms; any regularities they notice (like “past tenses end in ‑ED”) are extended by analogy as far as their peers will let them (“bended”).  All these consistent “trends” in language change make prediction more feasible, or at any rate, less obviously hopeless.

Nonetheless, futurology is a mug's game, and I don't expect my “predictions” to come true.  My methodology consists of nothing more rigorous than applying some of the kinds of changes that are commonly seen in historical linguistics and seeing what further development patterns they suggest; it's just a bit of fun, intended to dramatise the way things might plausibly end up if things go on the way they always have.  You could come up with something completely different and at least as plausible by extrapolating from the Northern Cities Vowel Shift

2005 ADDENDUM: I've mentioned the two commonest misconceptions about language change – that it's a bad thing, and that it has stopped; but a few other odd assumptions seem to be more widespread than I'd realised, so perhaps I'd better deal with them here so I don't have to carry on doing it in email.

“It's changes in vocabulary that matter.”
Monoglots often seem to think of languages as consisting of wordlists and nothing else!  Slang does serve as one of the most obvious markers of variation; but this is a superficial kind of change, often reversed a decade later, and rarely extending to the core vocabulary.  Meanwhile, shifts in vowel‐sounds or verb‐endings attract less attention, but they're cumulative and systematic; and it's these, not the vocabulary churn, that make 1000 AD English unintelligible.
“All changes can be traced back to the influence of other languages.”
After the Norman Conquest, the eclipse of English as a standard language made it easy for dialectal variant forms to get established, but apart from a transfusion of loanwords, the changes themselves were things that had already been going on before the French‐speakers turned up.  Grammatical “cross‐contamination” between neighbouring languages is the exception rather than the rule.  Indeed, it's rare for changes to have obvious “causes” at all.
“English is a pidgin.”
No, pidgins arise as grammarless codes for rough‐and‐ready communication between people who have no language in common.  If it has native speakers, it isn't a pidgin!  English isn't even a “creole”, the kind of language that's formed when a pidgin becomes a mother‐tongue.  However, the trace of truth in this myth is that being used as an auxiliary language often seems to trigger languages to become more “streamlined”.

I'm hoping not to have to turn this section into a “Language Change Myths FAQ”, since that would be a lifetime's work!

INTERMISSION – Notation and Terminology

I've delayed defining some of this glossary stuff in the hope of suckering people into reading this far, but if you want to follow the next few sections it's important to understand the difference between…

Standard Orthography: sample
Angle‐brackets enclose examples of words spelled in the language's standard orthography.  I imagine by 2200 nobody will be writing anaesthetise the same way as I do, but spelling reform is not the topic here (and nor is the kind of so‐called “bad grammar” that's really just non‐standard spelling and punctuation).
Phonetic transcriptions: ˈsãːmpɫ̩
Square‐brackets mark phoneTic transcriptions, which give a close‐up view of the sounds involved in terms of the International Phonetic Alphabet (if your browser can't handle the Unicode characters, please get a better one).  It shows even the “trivial” articulatory features that aren't used to distinguish words from one another – necessary because languages vary in their opinions about what counts as trivial and what sounds form natural sets.
Phonemic transcriptions: ˈsampl̩
IPA in slant‐brackets is a phoneMic transcription, a subtly but significantly different form of notation which spells things out in terms of the mutually contrastive sound “building‐blocks” of a particular language and ignores the irrelevant details.  Compare my own key to my pronunciation of English (though that accent isn't directly relevant here).

(If you're wondering where the brackets are, sorry: it'll be because your browser is ignoring my CSS…)
Bored already?  If you can't be bothered with all this you can always just take my word for it and skip to the end where I give examples of the final result.  Otherwise here are definitions of a few phonological terms I'll be using to get there:

Pronounced with or without vibration of the vocal cords.  This effect provides the glottal humming component that distinguishes (for instance) voiced “V” from voiceless “F” or voiced “G” from voiceless “K”.
I'm not talking letters here – there are many more than five vowels!  In phonological terms, vowels are the sounds made up of little‐modified bursts of voiced airflow (like “AH”), while consonants involve marked narrowing (or temporary closure, as in “K”) of the vocal tract.
Degrees of interference in the flow of air.  Plosives like “K” interrupt the escape of air through the mouth, fricatives like “Z” make the flow turbulent, and approximants like “W” only modify it slightly.
Nasal sounds allow air to escape continuously through the nose, while oral sounds don't – compare nasal “M” and its oral twin “B”.  The same can also happen in vowels, though that's never distinctive in twenty‐first‐century English.
This is another way of dividing up the consonants: fricatives and plosives (“Z”, “D”, “CH” etc.), which involve serious constriction, are grouped together as “obstruents”, while the highly resonant approximants and nasals (“M”, “L”, “W” etc.) are “sonorants”.
A syllabic sound is one that forms the main peak of a syllable.  This isn't quite the same as “vowel/consonant”, since vowels (or at least semivowels) can be nonsyllabic, and consonants (or at least sonorants) can be syllabic; examples include the initial “W” and final “L”‐sounds in waddle.
The most prominent syllable in a word is said to be stressed; other syllables may have some lesser degree of stress or none at all.  The acoustic property involved is a combination of volume, pitch, and duration.
Rounded vowels are pronounced with the lips in position as for an “OO” sound (or maybe only an “OH” sound); unrounded ones aren't.
These terms describe modifications in the position of (the highest point of) the tongue for a vowel; “EH” is a “front” vowel, “OH” is a “back” one.
…And these describe the other dimension: “AY” is a half‐close vowel, pronounced with the tongue quite high in the mouth; raising it gets you the close vowel “EE”, while lowering it results in half‐open “EH” then fully open “A”.
The “nucleus” of a syllable is its main syllabic element (i.e. the vowel, usually); any sounds that come before the nucleus are the “onset”, and any that come after are the “coda”.

The last thing I ought to say before I switch from “documentary” mode to “speculative fiction” mode is this: if you aren't familiar with Comparative Reconstruction then my predicted sound changes are bound to seem wildly unlikely.  If I'd shown Julius Caesar a schedule of the changes that were to turn Latin into Italian (“PS: beware the Ides of March”) he wouldn't have believed a word of it either.  And yet languages really do behave this way, with “mutations” in the system of sounds adding up to new accents, new languages, new family trees of descendant tongues… witness this Wikipedia entry on one big‐name sound change, Grimm's Law.

∼1385 AD: Ye knowe eek that in forme of speche is chaunge With‐inne a thousand yeer, and wordes tho That hadden prys, now wonder nyce and straunge Us thinketh hem, and yet thei spake hem so, And spedde as wel in love as men now do

(From Chaucer's “Troilus and Criseyde”.  Note that for Geoffrey, “do” and “so” rhymed, but “price” and “nice” didn't.)


I'm using a gerrymandered starting point here: please note that the phonology described below isn't precisely that of any major present‐day US accent (although it is close enough for plausibility – close enough indeed that this section is essentially a summary of existing trends).  Instead it's just one of the accents that will be current in a century or so: the one that happens to be ancestral to the thirty‐first‐century language.

Assimilation in Syllable‐Onset
Any syllable‐onset t, d or st with a following r undergoes assimilation (that is, features of one sound bleed over into the other); the results are the clusters tʃɹ̥, dʒɹ, stʃɹ̥.
Post‐Stress Voicing
There is a well established phonetic trend towards the voicing of any lone obstruent that closes a stressed syllable: e.g. pick is pronounced pɪɡ (note: the compound phoneme as in itch counts for this rule as a single sound).  However, the words pick and pig are still distinct – see below on vowel breaking.
Intervocalic Flapping
As a special case, d or t between stressed and unstressed syllables (perhaps with a preceding sonorant) is phonetically a flap, ɾ – a sound many languages treat as a form of r.  Thus for instance bitty is pronounced ˈbɪɾi.  This “flapping” context never triggers vowel breaking – bitty and biddy are pronounced identically.
Nasal consonants in the coda of a stressed syllable drop out in favour of nasalisation of the vowel; e.g. bond is pronounced bɑ̃d.  Immediately following plosives may also be nasalised (thus bɑ̃n).
Cluster Simplification
Meanwhile, ld not followed by a vowel becomes l (so build is now simply bɪl); nd in unstressed contexts simplifies similarly (thus England becomes ˈɪŋɡlən).
Unstressed Vowel Loss
Unstressed syllables may be lost immediately before or after stress; e.g. abominable turns into 'bom'nable.  Unstressed i, u tend to become nonsyllabic (turning into the approximants j, w), and other unstressed vowels reduce towards the “schwa” ə.
Vowel Mergers
Neutralisations (the technical name given to the blurring of phonemic distinctions) are widespread before sonorants – fewer vowels are distinguished before m, n, ŋ, l, w, j, or especially r.  See below for details.
Stressed Vowel Breaking
Stressed vowels begin to undergo a process called “breaking” before (phonemically) voiced obstruents, becoming generally longer and more lax; thus for instance pick is pɪɡ while pig is pɪɪ̈ɡ.  See below for details.
Occurs before a sonorant in beer, beam, bean, and peel = pill (a recent merger)
In a non‐breaking context, e.g. beat, i
In a breaking context, e.g. bead,
Occurs before a sonorant in bring, pin, him, where it is indistinguishable from ɛ as in pen, hem
In a non‐breaking context, e.g. bit, ɪ
In a breaking context, e.g. bid, ɪɪ̈
Occurs before a sonorant in Mary = merry = marry and in bale, bane, blame, bang (n.b. that last is beŋ, not baŋ)
In a non‐breaking context, e.g. bait, e
In a breaking context, e.g. bayed,
Occurs before a sonorant in bell (and see above on pen)
In a non‐breaking context, e.g. bet, ɛ
In a breaking context, e.g. bed, ɛɛ̈
Occurs before a sonorant in pal, ban, bam
In a non‐breaking context, e.g. bat, æ
In a breaking context, e.g. bad, ea
ɑ (Already to British ears a triple merger of ah/aw/o)
Occurs before a sonorant in bar, borrow, ball, pawn, bomb, bong
In a non‐breaking context, e.g. bot, ɑ
In a breaking context, e.g. bod, ɑə
Occurs before a sonorant in for = fore and in bowl, bone, foam
In a non‐breaking context, e.g. boat, o
In a breaking context, e.g. bode,
Only occurs unstressed (never a breaking context), e.g. initial in about; for unstressed syllables involving sonorants see below on “Syllabic Consonants”.
Occurs before a sonorant in bulk, bun, bum, bung
In a non‐breaking context, e.g. but, ʌ̈
In a breaking context, e.g. bud, ʌ̈ə
ʊ (phonetically quite fronted)
Occurs before a sonorant only as a variant form of syllabic or  – see below on “Syllabic Consonants”.
In a non‐breaking context, e.g. put, ʊ
In a breaking context, e.g. good, ʊʊ̈
u (also phonetically fronted)
Occurs before a sonorant in pool, boon, boom, and woman (a recent shift from ʊm)
In a non‐breaking context, e.g. boot, ʉ
In a breaking context, e.g. food, ʉɵ
Occurs before a sonorant in coin; however, boil is a disyllable, ɔɪ ɫ̩
In a non‐breaking context, e.g. quoit, ɔɪ
In a breaking context, e.g. void,
Occurs before a sonorant in prowl, brown
In a non‐breaking context, e.g. bout, au
In a breaking context, e.g. proud, it tends towards , becoming indistinguishable from ai as in pride
Occurs before a sonorant in bile, mime, brine
In a non‐breaking context, e.g. bite, ai
In a breaking context, e.g. pride, it tends towards , becoming indistinguishable from au as in proud
Syllabic Consonants
The distinctive syllabic in words like fury, poor, murderer can function as a stressed vowel, varying from a merely “tongue‐bunched” ʊ˞ or ɚ to a full ɹ̴̩.
Likewise the syllabic that occurs in pullable first as ʊɫ and then as ɫ̩.
The nasals and can also occur as syllabics as in Adam, Eden, but are always unstressed.


By this time the language has fallen out of fashion; the phonemic analysis given here is the one used retrospectively in the subsequent “Classical” period.  The vocabulary shrinks and is later restocked with borrowings, but many of them are returns, and the basic core of the language remains Germanic.

Lone Consonants in Syllable‐Coda
Most significantly, the post‐stress voicing effect becomes phonemic: pick is now piɡ, while pig is piaɡ (see below on vowel‐breaking).
The rule that converts both d and t to the flap ɾ persists, making the two indistinguishable in that context (sandwiched between a stressed and an unstressed syllable).
Clusters in Syllable‐Coda
Any obstruent forming a cluster with an s or z is lost; thus six becomes sis (but is not voiced to siz).
Clusters in Syllable‐Onset
The sequences tr, dr, and str become tʃr, dʒr, ʃr, while θr as in three shifts to become a new tr.
The sequences tʃj or tj (including such sequences created by vowel‐breaking – see below) become ; ʃj or sj become ʃ; and likewise for dʒj and ʒj.
After any obstruent, former w (but not new w resulting from vowel‐breaking) becomes v; thus quit becomes kvid.
Where a stressed or semi‐stressed vowel is followed by a nasal consonant without a following vowel, the nasal is pronounced only as a modification of the vowel.  In clusters, mb becomes mm, or simple m where there's a following vowel – i.e., ˜m.
Meanwhile mp becomes mb (i.e. ˜b), even outside stressed syllables.
Equivalent changes happen to nd, nt, ŋɡ, and ŋk.  For (former) nt the situation is complicated by the fact that nd may be flapped as ˜ɾ.
The Great Vowel Breaking
There is a major reorganisation and reanalysis of “broken” vowels as sequences, converting what used to be a feature of the following consonants into a pattern of new distinctions within the vowel inventory.
Note throughout that vowels that “want” to gain a preceding or following w or j don't if one is already there; weed becomes wid, not wjid.  Preceding r or l has the same effect (reed breaks to rid), but h gives way itself (heed breaks to jid).
Unbroken i syllable‐finally or before r, l, or a nasal (bee, beer, bill, peel, bean, beam) is unchanged
Otherwise, unbroken former i (beat) becomes i j
Broken former i (bead) becomes j i
Unbroken former ɪ (bit) becomes i
Broken former ɪ (bid) becomes i a
Unbroken former e syllable‐finally or before r (bay, bear) is unchanged; before l or a nasal (bale, bane, blame, bang) it becomes j e
Otherwise, unbroken former e (bait) becomes e j
Broken former e (bayed) becomes j e – note that j e is also a possible broken form for former æ (see a below)
Unbroken former ɛ (bet, bell) becomes e, which also absorbs former nasal ɪ (him, pen, bring)
Broken former ɛ becomes e a (bed)
Broken former ɔɪ (void) becomes w e
a (ranging phonetically from æ to ə)
Unbroken former au before a sonorant (bounce, brown) becomes the disyllable a o
Otherwise, unbroken former au (bout) becomes a w
Unbroken former ai before a sonorant (pint, brine) becomes the disyllable a e
Otherwise, unbroken former ai (bite) becomes a j
Broken former au or ai (proud, pride) becomes a a
Unbroken former æ (bat, ban) becomes a, except after k or ɡ (cap) when it becomes j a
Broken former æ (bad) becomes j a, except after k or ɡ (cab) when it becomes j e
Former ə generally merges with unstressed a (and see an below under “Syllabic Consonants”)
Unbroken ɑ (paw, bot, bar, ball, bomb) is unchanged
Broken former ɑ (bod) becomes ɑ a
Former word‐final ɑ (e.g. the end of beta) merges with unstressed ɑ
ɜ (former ʌ)
Unbroken ʌ (but, bun) becomes plain ɜ
Broken former ʌ (bud) becomes ɜ a
Unbroken o (blow, boat, bore, bone) is unchanged
Broken former o (bode) becomes w o
Unbroken former ɔɪ before a sonorant (coin) becomes the disyllable o e
Otherwise, unbroken former ɔɪ (quoit) becomes o j
ʉ (former u)
Unbroken u syllable‐finally or before sonorants (blue, pool, boon) becomes plain ʉ
Otherwise, unbroken former u (boot) becomes ʉ w
Broken former u (food) becomes w ʉ
Unbroken former ʊ (put) becomes ʉ
Broken former ʊ (good) becomes ʉ a
Syllabic Consonants
Unbroken syllabic and (pert, fur, wolf, full) become ʉr, ʉl – or r̩r, l̩l where followed by a vowel (furry, fully)
Broken and (bird, bulls) are unchanged
Note that the above syllabic forms, along with syllable‐coda r and l (but not the syllable‐onset forms), are pronounced at the back of the mouth.
Former syllabic nasals ( and ) are reanalysed as the sequences am, an (never stressed as ã)


Contrary to the impression you'd get from a detailed account of the chaos the spelling system goes through early in this stage, the Classical period happens to be one of relative stability in the development of the language as a whole, and one that Late American speakers continue to regard as a formal standard.

Stress Shift
Nouns (and to a lesser extent other word‐classes, though not verbs) tend towards regular initial‐syllable stress; thus for instance millennium shifts from ma ˈlen jam to ˈma lan jam.
Interdental Loss
The phonemes θ, ð (as in thigh, thy), long gone in related dialects, finally vanish in American, merging with t and d respectively.
Secondary Split
The above changes undermine the rule that used to explain the distribution of the ɾ sound.  It used to be a post‐stress variant of d, but now occurs where stress has vanished, and fails to occur where the d was formerly ð; so it's left as a short‐lived ɾ phoneme in its own right.
Affricate Loss
The phonemes , are lost, simplifying to ʃ, ʒ.  This happens in onset and coda, including nasal contexts (e.g. inch: ɛnʒ).  The “shibilants” ʃ and ʒ also undergo a phonetic shift towards ɕ, ʑ (technically, dorsal palatals, like Mandarin Chinese sh).
In onset, the voiceless plosives p, t, and k become increasingly strongly aspirated (phonetically etc.) except after s (or ʃ); the phoneme h itself is lost unless preceded by a vowel, where it becomes a voiced approximant articulated almost anywhere from the soft palate back (labelled ɦ here).
Syllable‐Coda Clusters
The nasal ŋ, now uncommon, loses its phonemic status; bang, formerly beŋ, is now interpreted as ending in a nasalising n which has assimilated in place of articulation to an otherwise silent ɡ – bɛnɡ, pronounced bæ̃ŋ.
Remaining consonant clusters are reduced; where there are two obstruents in the coda, the second tends to be dropped (apt becomes ap); but between vowels, pt or kt become tt.
Syllabic Consonants
Syllabic and are lost (see below); syllabic nasals are still equivalent to unstressed an, am.
Rhotic Split
Former syllable‐coda r becomes ɦ in inherited American vocabulary, while new loanwords use r – thus beer has become biɦ while the equivalent Brazilian import is bir.
Former syllabic always becomes ʏɦ; syllable‐onset r remains unchanged.
Lateral Vocalisation
Former syllable‐coda l is lost: il becomes iw, el becomes ɛw, al becomes aw, ɑl becomes ɑw (pronounced ɔw), and ɜl, ol, or ʉl becomes o.
Former syllabic also becomes o.
Vowel Shifts
Except before l as noted above, former ʉ becomes ʏ (still never quite a fully fronted y).
Former e becomes ɛ.
Former ɜ is lost, merging with a; however, in the case of the common sequence ɜa, an intrusive ɦ is inserted and the unstressed vowel then drops out – thus bɜad (bud) becomes baɦad and subsequently baɦd.
Nasal Vowels
By the late Classical period the nasal vowels are more or less separate phonemes in their own right; phonetically they tend to be more open than their oral counterparts, with ĩ becoming , ɛ̃ becoming æ̃, õ becoming ɔ̃, and ʏ̃ becoming ø̃.


The language represented by the examples in the final section.  By this point the Great Wheel of Morphology has come round from a thoroughly analytic to an increasingly agglutinative grammar, but there isn't room here to cover the complexities of Late American verb declensions.

Plosive Affrication
When they occur immediately before a stressed vowel, the aspirated plosives p, t, k tend to become the affricates , ts, kx.
On the other hand, plosives after s (or ɕ) are unaspirated, and thus come to be regarded as voiced; former sp becomes sb, st becomes sd, and sk becomes .
Aspirate Shift
The voiced ɦ settles down as ʁ: phonetically, a uvular approximant (ʁ̞, which is a sort of gentle “ugh” at the very back of the mouth).  When k or ɡ come in contact with it they themselves become uvular (q, ɢ) and the ʁ may approach a gargled “French r” sound.
Flap Loss
The flap phoneme ɾ is lost, turning into r before an a or ʏ, and simply disappearing elsewhere.
Coda Approximant Loss
Syllable‐coda j tends to be lost or moved.  First, js or assimilate to ɕ and jz or to ʑ; then j tends to switch places with a following consonant (e.g. becomes ɡj), even when that consonant is preceded by a nasal and/or followed by r or l (jɡr becomes ɡrj); however this switching (or “metathesis”) does not occur with heavier consonant clusters or word‐finally.  Remaining cases of aj become ɛ; otherwise the j is just dropped.
Syllable‐coda w behaves similarly.  First, wn assimilates to m (and wnd to mb); then w undergoes the same kind of metathesis as j (e.g. wɡl becomes ɡlw).  Remaining cases of aw become ɔ; otherwise the w is dropped.
Syllable‐coda ʁ is slightly different, in that it may assimilate to a following sonorant: ʁl becomes ll before a vowel, l otherwise, and ʁr likewise becomes either rr or r.  In ʁw and ʁj the ʁ is always dropped, and the same happens with ʁn or ʁm if the nasals have not themselves been lost to nasalisation.  Elsewhere, metathesis occurs but is more easily blocked – non‐final ʁɡ may become ɡʁ, but ʁɡr or ʁɡl is unchanged.  Leftovers keep the ʁ, which lowers preceding vowels just like a nasal (so i before ʁ is e, ʏ is ø, u is o).
Nasal Neutralisation
The distinction between (e.g.) former bean and beam is lost.  It was already blurred, since both are pronounced in isolation as bẽ, but when the word is immediately followed by a syllabic sound as in beam of light, which used to revive the final nasal consonant, it now always inserts the same one – thus bẽnəlɛd.
Syllabic nasals are unaffected, though being unstressed they do tend to reduce to plain n or m alongside a vowel.  Plosives nasalised as a side‐effect of a preceding nasal vowel also escape the neutralisation: sɛ̃ɡ (sing) is still pronounced sæ̃ŋ, not sæ̃n.
Back Vowel Raising
Former ɑ becomes ɔ (weakly rounded if at all); former o becomes u.  Note that as a result American at last has a more‐or‐less evenly spaced vowel inventory: front i and ɛ, central ʏ and a, and back u and ɔ (plus their nasal equivalents).

EXAMPLES – Words and Phrases

The examples given below are selected largely on the basis of semantic stability; there's no point using a word like “computer”, which means different things from century to century.  It also simplifies things to start with nouns, which have no confusingly mutable inflected forms.  The spellings used are the closest transliteration I can manage within the limitations of a twenty‐first‐century characterset; fortunately by the thirty‐first century storing information as strings of written words is something of a fossil handicraft anyway (much like calligraphy in the present day), so an “anachronistic” font is as good as any.

If you're wondering about the leading asterisks, those are a slightly warped application of the convention used for “real” reconstructed languages like Proto‐Indo‐European, where the star in front of *oinom is a warning that it's an unattested “best guess” at the PIE for “one” arrived at by deducing the sound‐change rules that separate it from modern languages.

American language › *myeghan lengvaj
Early American ˈmerkn̩ ˈleŋɡwədʒ
Middle American ˈmjerɡan ˈleŋɡvadʒ
Classical American ˈmjɛɦɡan ˈlɛnɡvaʒ
Late American ˈmjɛɡʁan ˈlɛ̃ɡvaʑ
Pronounced: ˈmjɛɢʁn̩ ˈlæ̃ŋvəʑ, i.e. “MYEGrhnn LANGvuzh”
George Washington › *Jwohj‐wᴀjandan
Early American dʒordʒ ˈwɑʃəŋtn̩
Middle American dʒwordʒ ˈwɑʒandan
Classical American ʒwoɦʒ ˈwɑʒandan
Late American ʑwuʁʑ ˈwɔʑandan
Pronounced: ʑwoʁ̞ʑ ˈwɔʑn̩dn̩, i.e. “zhwohghzh WAWZH'n'dnn”
Abraham Lincoln › *Yebraham‐lengan
Early American ˈebrəham ˈlɪŋkn
Middle American ˈjebraham ˈleŋɡan
Classical American ˈjɛbraɦam ˈlɛnɡan
Late American ˈjɛbraʁam ˈlɛ̃ɡan
Pronounced: ˈjɛbɹəʁ̞m̩ ˈlæ̃ŋn̩, i.e. “YEB‐rugh'm LANG'n”
William Shakespeare › *Wiyam‐xexbih
Early American ˈwiljəm ˈʃekspir
Middle American ˈwiljam ˈʃejspir
Classical American ˈwiwjam ˈʃɛjspiɦ
Late American ˈwijam ˈɕɛɕbiʁ
Pronounced: ˈwijm̩ ˈɕɛɕpeʁ̞, i.e. “WEE‐ymm SHESHpaygh”
red, white, blue › *read, *wed, *blu
Early American rɛd wait blu
Middle American ˈread wajd blʉ
Classical American ˈrɛad wajd blʏ
Late American ˈrɛad wɛd blʏ
Pronounced: ˈɹɛəd wɛd blʏ, i.e. “REH‐ud wed blü”
one, two, three, four, five › *wan, *tu, *tri, *foh, *faav
Early American wʌn tu θri for faɪv
Middle American wɜn tʉ tri for faav
Classical American wan tʏ tri foɦ ˈfaav
Late American wã tʏ tri fuʁ ˈfaav
Pronounced: wɐ̃ tsʏ tɹ̥i foʁ̞ ˈfɐəv, i.e. “wu(ng) tsü tree fohgh FUH‐uv”
six, seven, eight, nine, ten › *sis, *seavam, *ed, *naen, *ten
Early American sɪks ˈsɛvn̩ et nain tɪn
Middle American sis ˈseavan ejd ˈnaen ten
Classical American sis ˈsɛavan ɛjd ˈnaɛn tɛn
Late American sis ˈsɛavam ɛd ˈnãɛ̃ tɛ̃
Pronounced: sis ˈsɛəβm̩ ɛd ˈnɐ̃æ̃ tsæ̃, i.e. “cease SEH‐uv'm ed NUH‐a(ng) tsa(ng)”
California, Texas › *Kyafwonyᴀ, *Tesas
Early American ˌkæləˈfornjə ˈtɛksəs
Middle American ˌkjalˈfornjɑ ˈtesas
Classical American ˈkjawfoɦnjɑ ˈtɛsas
Late American ˈkjafwũjɔ ˈtɛsas
Pronounced: ˈkj̥afw̥õjɔ ˈtsɛsəs, i.e. “KYUFFwoh(ng)‐yaw TSESSuss”
Mercury, Venus › *Muhgyurri, *Vinas
Early American ˈmr̩kjr̩ri ˈvinəs
Middle American ˈmʉrɡjr̩ri ˈvinas
Classical American ˈmʏɦɡjʏɦri ˈvinas
Late American ˈmʏʁɡjʏrri ˈvinas
Pronounced: ˈmøʁɢjʏɹɹi ˈvinəs, i.e. “MÖRHGyürrree VEEnus”
Earth, Mars › *Uhd, *Mᴀahz
Early American r̩θ mɑrz
Middle American ʉrð ˈmɑarz
Classical American ʏɦd ˈmɔaɦz
Late American ʏʁd ˈmɔaʁz
Pronounced: øʁ̞d ˈmɔəʁ̞z, i.e. “öghd MAW‐ughz”
Jupiter, Saturn › *Jubwatuh, *Sarun
Early American ˈdʒupətr̩ ˈsætr̩n
Middle American ˈdʒʉwbatʉr ˈsadʉrn
Classical American ˈʒʏwbatʏɦ ˈsaɾʏɦn
Late American ˈʑʏbwatʏʁ ˈsarʏ̃
Pronounced: ˈʑʏbwətøʁ̞ ˈsaɹø̃, i.e. “ZHÜBwatögh SUH‐rö(ng)”
Uranus, Neptune › *Yurranas, *Nettun
Early American ˈjr̩rənəs ˈnɛptun
Middle American ˈjr̩ranas ˈneptʉn
Classical American ˈjʏɦranas ˈnɛttʏn
Late American ˈjʏrranas ˈnɛttʏ̃
Pronounced: ˈjøɹɹənəs ˈnɛttø̃, i.e. “YÜRrranas NET‐tö(ng)”

The rough pronunciation guides above have deliberately not been made too simple – that would risk leaving readers with the impression that Futurese was just a lazy, garbled version of Presentdayese.  In particular those umlauts should serve to remind readers that our successors will have different ideas about what sounds are “basic” and “easy”, and which are “subtle” and “exotic”.

2016 POSTSCRIPT: thanks to A. Z. Foreman for articulating, recording, and publishing some fantastic audio file versions of the above!

And finally: to give an impression of how much else has been going on besides regular sound‐changes, here's a Late American rendition of the Colloquy of Ælfric (as seen previously), followed by a word‐by‐word analysis.  3000 AD American has metamorphosed into something that is clearly a new language, yet recognisably a descendant of English – sentences even have a familiar stress‐timed rhythm.

Mind you, if you've been skipping over the phonetics and only looking at the spellings, you'll get an exaggerated impression of the differences between 2000 AD and 3000 AD, since our present‐day standard orthography is basically mock‐Chaucerian (for instance, we still write knight the way they used to say it: as nit with extra consonants).  As a counterbalance to this, instead of repeating my sample text's 2000 AD version spelled as if it was Middle English, I'll do things the other way round and write it according to Classical American conventions here:

2000 AD: Wi txìldran beg yu, titxar, đat yu xùd titx as tu spik karektli, bikaz wi ar ìgnarant and wi spik karàptli…
3000 AD: *Zᴀ kiad w’‐exùn ya tijuh, da ya‐gᴀr’‐eduketan zᴀ da wa‐tᴀgan lidla, kaz ’ban iagnaran an wa‐tᴀg kurrap…
*zᴀ, pronounced “zaw”
“Us‐all”, analogous in form to the second‑ and third‐person *yᴀ, *dᴀ.
*kiad, pronounced “KKHEE‐ud”
“Kid”, obviously enough.
*w’‐exùn, pronounced “weSHÖ(NG)”
Pronominal prefix (“we”) and finite verb‐stem; a twenty‐fifth‐century slang term, origin unclear.
*ya, pronounced “yuh”
“You”, singular.
*tijuh, pronounced “TEEZH‐ögh”
From “teacher”, now restricted to meaning specifically a language‐instructor.
*da, pronounced “duh”
“That”, as a subordinating conjunction.
*ya‐gᴀr’‐eduketan, pronounced “yagaw‐RED‐üket'n”
Pronominal prefix, auxiliary prefix (from “gotta”) and subordinate verb (“educate” – note the preserved form).
*wa‐tᴀgan, pronounced “wuh‐TSAWG'n”
“Talk”; pronominal prefix and subordinate verb.
*lidla, pronounced “LEEDla”
A back‐loan from Central Hindi, where English “legal” developed the specialised sense “linguistically well formed”.
*kaz, pronounced “kkhuzz”
Conjunction, “because”.
*’ban, pronounced “bnn” (unstressed)
Irregular particle derived from the verb “be”.
*iagnaran, pronounced “EEugnurr'n”
Regularly derived from “ignorant”.
*an, (still) pronounced “'n”
The coordinating conjunction “and”.
*wa‐tᴀg, pronounced “wuh‐TSAWG”
As in the previous clause, but this time in the positive‐indicative form.
*kurrap, pronounced “KKHÜRrrup”
Regularly derived from “corrupt”.