actually — a reconstruction

§1The object of reconstruction

The object of reconstruction is not a single dictionary gloss but the whole historical object that ends in the English adverb actually. It includes the Indo-European root *h₂eǵ- and its cognate set, the Latin verb agere, its participial stem act- and the noun stem actu- of actus, the Latin adjective actualis and its Scholastic technical value, the transmission of the adjective into French and English, the formation of the English -ly adverb, and the later semantic-pragmatic development through veridical, counterexpectational, corrective, mitigating and self-repair uses. The reconstruction therefore has a formal backbone and a semantic-pragmatic extension: the formal backbone accounts for where the word comes from, while the semantic-pragmatic extension accounts for what the word has become. The two are kept distinct because a formal ancestor does not by itself determine a later discourse function, and a later discourse function does not license a formal etymology.

Functions: reconstruct the formal ancestry of English actually from PIE *h₂eǵ- through Latin agere, the stem act-, the noun stem actu- of actus, Late Latin actualis and the English adjective actual, model the semantic-pragmatic trajectory from 'in act, in deed' to veridical, counterexpectational, corrective, mitigating and self-repair uses, distinguish genetic inheritance, morphological derivation, borrowing, conceptual translation and semantic change in the word's history, situate the word in the Aristotelian-Scholastic actus/potentia terminology and in the Latin-French-English transmission channels, account for the phonological and orthographic form of the modern adverb, relate the English adverb to its Indo-European cognates and to its Romance and Germanic parallel branches

§2Evidence, scope, and degrees of certainty

The evidence base is heterogeneous, and its scope differs from link to link. For the Indo-European root, the comparative method rests on the cognate set Greek ἄγω, Sanskrit ájati, Old Norse aka, Armenian acem and Old Irish aigid, which anchor the reconstruction of *h₂eǵ- as a verb 'drive, lead, set in motion'. For the Latin formal chain, the morphology is transparent: agere supplies the perfect-participle stem act-, the fourth-declension noun actus is built on that stem, and the -ālis adjective actualis is built on the noun's stem actu- rather than on the perfect-participle stem or on the citation form agere. For the philosophical value, the evidence is the translated Aristotelian vocabulary: Greek ἐνέργεια and δύναμις are rendered by Latin actus and potentia, and the resulting actus/potentia opposition is what gives actualis its Scholastic sense 'actual as opposed to potential'. For the transmission into English, the record shows at least two channels: Old French actuel, with the French reflex -el of Latin -ālis, and learned Latin or Medieval Latin actualis, with Latinizing -al; because the channels converge on the same English form, their relative contribution is not distinguished by the record. Medieval Latin actualiter is a further possible source or reinforcement for the English adverb. For the later pragmatic stages, the evidence is the usage record of English: historical quotation material, present-day corpus patterns, and translation parallels. These stages are not all separately dated in a single text tradition; they are inferred from recurrent contextual values and from the way the adverb behaves in dialogue. The phonology is comparatively direct: yod coalescence accounts for /tj/ > /tʃ/ in the modern pronunciation, and schwa syncope accounts for the compressed variant. No confidence label is assigned in this section; the evaluation layer, not the generation layer, computes any score from the structure of this evidence.

backbone: PIE *h₂eǵ- > Latin agere > act- > Late Latin actualis > Old French actuel and Middle English actual > English actualy/actually > veridical, counterexpectational, corrective, mitigating and self-repair uses.

§3Earliest reconstructible ancestor: *h₂eǵ-

(to drive, lead, set in motion)

The root is reconstructed from the Indo-European cognate set that includes Greek ἄγω, Sanskrit ájati, Old Norse aka, Armenian acem and Old Irish aigid. Its semantic core is 'drive, lead, set in motion', a value that survives in the concrete vehicular sense of several cognates and is generalized in Latin agere to 'do, perform'. The Latin verb is the pivot of the whole reconstruction: it continues the root by regular sound change, and its perfect-participle stem act- is the base on which actus is built, while actualis is built on the noun stem actu- of actus. The root is therefore not an ornamental antiquity placed behind the English word; it is the first reconstructible state of the formal chain that makes Latin agere, Latin actualis and English actually members of one descent series. The falsification condition above states the kind of evidence that would break that series at its root.

Falsified by: If a securely identified Anatolian cognate of Latin agere — for example a Hittite or Luwian verb meaning 'drive, lead' — were found to reflect h₁- or h₃- rather than h₂- in its initial segment, as diagnosed by the colouring of the following vowel, and if the form could not be explained as a loan or as an inner-Anatolian innovation, the root h₂eǵ- would not hold.

§4–8Derivational & transmission chain

1. *h₂eǵ- (to drive, lead, set in motion) — uncertain origin → Proto-Indo-European

The root is not attested, but the cognate set supports the reconstruction of an Indo-European verb *h₂eǵ- with the sense 'drive, lead, set in motion'. No earlier source is reconstructible, so the chain begins from an explicitly uncertain origin.

2. agere (to do, perform; to drive, lead) — PIE *h₂eǵ- → Latin

Latin agere continues the root by regular sound change. Its sense range includes both the concrete 'drive, lead' and the generalized 'do, perform', and it is the source of the participial stem that carries the later derivatives.

3. act- (participial stem 'driven, done, performed') — Latin agere → Latin derivational stem

The derivational stem act- is the perfect-participle stem of agere. Lachmann's law accounts for the long ā, and this stem — not the citation form agere — is the base to which the nominal suffix attaches. The adjective actualis is built not on act- but on the noun stem actu- of actus.

4. actus, -ūs (a doing, a driving; act, deed, performance) — Latin act- → Latin

actus is the verbal abstract noun built on act-. It names the deed or performance, and it supplies the lexical form that is later reused in the philosophical sense 'actuality', though that technical value has its own conceptual history.

5. actualis (pertaining to action, practical; (scholastic) actual as opposed to potential) — Latin act- → Late Latin

actualis is the -ālis adjective of the stem actu- of actus. Its ordinary sense is 'pertaining to action'; within the Scholastic actus/potentia frame it narrows to 'belonging to what is in act, actual as opposed to potential'.

6. actuel (real, actual; present, current) — Late Latin actualis → Old French

Old French actuel transmits actualis into French in a semi-learned form. It is one channel by which the Latin adjective reaches English, and its -el ending is later remodelled in English to Latinizing -al.

7. actual (pertaining to action, practical; real) — Old French actuel and/or Late Latin actualis → Middle English

Middle English actual reflects both the Old French channel and the learned Latin channel. Because the two channels converge on the same English form, the record does not distinguish their relative contribution.

8. actualy/actually (in act, in deed; practically) — Middle English actual → English

The English adverb is formed on actual with -ly and inherits the sense 'pertaining to action'. Medieval Latin actualiter may have supplied or reinforced the adverb, whose sense and structure are identical to it.

9. actually (in reality, in fact) — Middle English actualy/actually 'in act, in deed' → English

The veridical sense develops out of 'in act, in deed' as the axis of contrast shifts from intention to appearance. actually comes to mean 'in reality, in fact'.

10. actually (in fact, contrary to expectation) — actually 'in reality, in fact' → English

In contexts where the stated fact runs against an assumption, the veridical adverb is recruited to code counterexpectation. The value 'as it happens, contrary to expectation' becomes conventional.

11. actually (turn-initial corrective marker) — actually 'in fact, contrary to expectation' → English

The counterexpectational value narrows interactionally: in turn-initial position actually codes that the proposition contradicts the addressee's preceding utterance or assumption.

12. actually (hedging/mitigating marker) — turn-initial corrective actually → English

The corrective force is attenuated under face-work. actually is recruited to soften requests, impositions and disagreement, functioning as a hedge.

13. actually (self-repair marker) — turn-initial corrective actually → English

The corrective function is re-targeted from the interlocutor to the speaker's own prior utterance. actually marks self-repair or afterthought in initial or postposed position.

§5Cross-disciplinary interface

Greek: energeia and dynamis — the conceptual origin, which is not a lexical ancestor

The tradition through which the concept behind actually becomes a technical one begins in Aristotle's contrast between ἐνέργεια and δύναμις, worked out most fully in Metaphysics Θ and presupposed throughout De anima. Energeia names the state of being at work, the realized exercise of a capacity; dynamis names the capacity itself, the ability to be or to act before it is exercised. Aristotle also uses ἐντελέχεια alongside ἐνέργεια for realized completion, and the competition between the two Greek terms is the first sign that the concept is being stabilized by terminological work rather than simply inherited.

The point that must be stated flatly, because it governs the whole section, is that nothing Greek is an ancestor of the English word. English actually does not descend from ἐνέργεια; it descends from Latin act-, and the Greek term participates in the word's history only as a concept to be translated. The Greek material is nevertheless indispensable, because it is what makes actus mean more than 'a doing'. A second, easily missed consequence is that the same Greek concept also produces a different English word by a different route: ἐνέργεια yields English energy through the Latinized energia of the early modern scientific vocabulary. Energy and actually are therefore conceptual siblings with unrelated formal histories, and an etymology of either that ignored the other would misdescribe the intellectual history of the pair.

Latin: actus and potentia as conceptual translation, not genetic inheritance

Latin possessed a native noun actus, the verbal abstract of agere, meaning 'a doing, a driving, a deed'. What the philosophical tradition did was to load that native noun with the value of ἐνέργεια, and to load potentia with the value of δύναμις, so that actus and potentia became the two poles of a translated opposition. Two distinct relations therefore converge on the single Latin form actus, and they must not be run together: actus is genetically descended from agere by regular morphology, and it acquires its philosophical sense by conceptual translation from Greek. The first relation is genealogical; the second is not, and no amount of semantic similarity makes ἐνέργεια a forebear of English actually.

The translation is instructive in its technique. It is sense-for-sense, not morpheme-for-morpheme: actus does not reproduce the ἐν + ἔργον structure of energeia, so it is a semantic loan rather than a loan-translation of the kind that produced Latin conscientia as a calque on Greek συνείδησις, from which English conscience was later borrowed. Competing renderings existed in Roman philosophical prose — actio and efficientia are both reported for energeia in some authors [unverified as to the exact distribution] — but the tradition settled on actus, and it is the settled choice that makes possible the later adjective actualis in its technical sense.

Arabic and Hebrew: quwwa/fiʿl and kōaḥ/pōʿal as reinforcement channels

The Aristotelian pair did not travel to the Latin West on a single road. Alongside the direct Greek-to-Latin route stands an Arabic elaboration: al-quwwa and al-fiʿl render dynamis and energeia in the falsafa tradition, and the adverbial formula bi-l-quwwa wa-l-fiʿl 'in potency and in act' is the standard Arabic way of predicating the distinction of a thing. Al-Fārābī and Ibn Sīnā develop the pair in their own metaphysical and psychological writing, and Averroes' commentaries are the form in which the elaborated doctrine reached the Latin schools.

From Arabic the pair passed into Hebrew in the Tibbonide translation tradition, where כֹּחַ and פֹּעַל calque al-quwwa and al-fiʿl, and where the phrase be-kōaḥ u-ve-pōʿal reproduces the Arabic formula exactly. Thirteenth- and fourteenth-century Latin versions made from Arabic — among them the 1328 Latin Tahāfut of Kalonymus ben Kalonymus, translated from Arabic — then fed the pair back into scholastic Latin; a later 1527 Latin version made from Hebrew by Calo Calonymos likewise carried the pair. The honest assessment of this channel's weight is that it is a reinforcement and refinement, not the origin: the Latin actus/potentia couple is already in place in the direct translation tradition, and what the Arabic-Hebrew route supplies is additional technical formulae and argumentative context, above all around actus and potentia in created beings. How much of the scholastic Latin usage of actualis was shaped by this route rather than by the direct one is not settled in the literature, and the present reconstruction does not claim to settle it.

Scholastic Latin: actualis, actualiter, and the formula in actu

Within the actus/potentia frame the adjective actualis takes on the value 'belonging to what is in act, actual as opposed to potential', and its regular adverb actualiter means 'in act, actually'. These are the two forms that matter for English: actualis is the form transmitted to French and English as an adjective, and actualiter is the formal and semantic double of the English adverb. Its ordinary sense, 'pertaining to action, practical', survives alongside the technical one; the technical value narrows the adjective to the act pole of the opposition rather than replacing the older sense outright.

The register's evidence is overwhelmingly formulaic, and it is the formulae, not the prose, that carry the term. In actu signato and in actu exercito distinguish what is in act as an object signified from what is in act as performed; actus purus names the wholly realized act that scholastic theology identifies with God; actus essendi names the act of being by which a thing is; in potentia supplies the opposed formula; and the derived technical adjectives appear in gratia actualis and peccatum actuale, actual grace as against habitual grace and actual sin as against original sin. These phrases instantiate related but not identical oppositions: in actu signato and in actu exercito oppose signified to performed, actus purus and actus essendi concern act as realized versus potential, gratia actualis opposes transient to habitual grace, and peccatum actuale opposes committed to original sin. It is precisely because these fixed phrases carry the term that the adjective can cross into English theology without the noun actus being borrowed with it.

English theology and law: actual grace, actual sin, actual notice, actual bodily harm

Two English registers take the adjective up in fixed formulae and thereby keep the old opposition alive after the metaphysics has been forgotten. In theology, actual grace is the transient divine assistance given for a particular act, opposed to habitual or sanctifying grace, and actual sin is the sin a person commits, opposed to original sin. Both keep the act-versus-state opposition intact: actual marks what is done or present here and now as against what is dispositional, inherited or imputed.

In law the same structure reappears in a secularized form, and this register supplies the sharpest evidence, because the formulae are definitional. Actual notice is notice in fact received, opposed to constructive notice, which the law imputes whether or not it was received; actual malice is the knowing falsehood or reckless disregard for truth proved to have been present, opposed to malice implied by law; actual damages are losses proved to have occurred, as against nominal or punitive damages; and actual bodily harm is the statutory description of injury that is more than merely transient or trifling. In each of these, actual opposes constructive, implied or nominal — that is, it opposes what is really there to what is posited or presumed. The legal register thus reproduces, in wholly non-metaphysical terms, the act/potency opposition out of which actualis first acquired its technical sense, and it does so in phrases of the form actual X versus constructive X in which the word is a constitutive element rather than a decoration.

English philosophy and modal logic: actual world, actuality operator, actualism, actual occasion

Modern English philosophy re-technicalizes the word a second time. Leibniz's contrast between the actual world and the infinitely many possible worlds gives English a fixed phrase in which actual does the work the scholastics gave to actus; twentieth-century modal logic formalizes the contrast by introducing an actuality operator — written A or @ — whose natural-language reading is 'actually', so that the English adverb becomes the official gloss on a formal symbol. Two-dimensional semantics then exploits precisely the difference between 'actually' read with respect to the world of utterance and 'actually' read with respect to a world of evaluation, which makes the English adverb a technical device for fixing reference to the actual world. Alongside this, actualism names the metaphysical thesis that only actualia exist, as against possibilism, and Whitehead's process philosophy uses actual occasion and actual entity for the units of realized process.

The important point for the etymology is that these are not decorative modern uses. They reinstate the oldest value of the word — 'in act, really, as opposed to merely possible or merely apparent' — in a register where the opposition is defined with formal precision, and they do so in formulaic phrases (actual world, actuality operator, actualism, actual occasion) that behave exactly like the scholastic in actu formulae they unknowingly continue.

§7Borrowing & transmission

The adjective actual is loan material; actually is the native English derivative built on it with the suffix -ly, so not every part of the adverb is borrowed. Old English does not offer a reliable reflex of this root in the relevant sense 'drive, lead'; the frequently cited acan means 'ache' and its affiliation is disputed. A speaker who saw a shared root there might be tempted to derive English actual from the Germanic stock; that derivation is excluded, because English actual shows the noun stem actu- of actus with a Latin-derived suffix, and no Germanic form contributes to it. The Germanic cognates are evidence for the Indo-European root, not for the English word.

The adjective reached English through more than one channel, and the record does not allow the channels to be apportioned. One channel is French: Old French actuel shows the learned stem actual- with the French reflex -el of Latin -ālis, and this is a semi-learned transmission in which the stem is kept Latin while the ending is French. A second channel is Latin itself, taken over in clerical, academic and legal writing as actualis with the Latinizing ending -al. Because English remodelled -el to -al on the Latin model, both channels converge on the same English shape, and the surviving English -al cannot by itself tell us which channel supplied a given occurrence. A third, weaker channel is later reinforcement of the sense from Latin-literate usage in the sixteenth and seventeenth centuries, when the adjective's philosophical value 'actual as opposed to potential' was still live in school Latin. The honest formulation is therefore that English actual reflects Old French actuel, learned Latin actualis, and possibly later Latinizing reinforcement; their relative contribution is not distinguished by the record, and any narrative that makes the word pass through exactly one of them and then get 're-Latinized' from the other overstates what the spellings can show.

The adverb has its own, partly separate transmission. English actually is regularly formed from actual by the productive native suffix -ly (from Old English -līce), and on that derivation alone it inherits the adjective's sense. But the formation has an exact Latin counterpart in Medieval Latin actualiter, identical in structure and in meaning, and it is not possible to determine whether English actually is an independent -ly formation, a loan-rendering of actualiter, or — most probably — an independent formation that the Latin adverb reinforced. Since the two routes yield the same string and the same sense, the question is not decidable from the English record alone; it might be settled by a dated survey of the earliest English occurrences showing whether they cluster in text types translated from Latin.

Finally, the parallel branches must not be mistaken for sources. French actuel and actuellement, German aktuell, Italian attuale, Spanish actual and actualmente are cognates or independent learned loans of the same Latin adjective; none of them supplies Modern English actually, and Modern English actually shows no trace of the temporal sense ('current, present') that French, German and Spanish conventionalized.

§9Semantic trajectory

S1 — 'in act, in deed, in practice' — the deed-versus-word value

The adverb enters English as the -ly derivative of the adjective actual, whose inherited Latin value is 'pertaining to action, practical' (E13), with possible learned reinforcement from Medieval Latin actualiter (E14). At this stage the adverb predicates of a state of affairs that it obtains in act, in deed, in practice, as against what obtains only in name, in intention, in promise or on paper. The mechanism of the stage's formation is adverbial suffixation itself: -ly converts a relational adjective into a predicate of occurrence, so that 'pertaining to action' becomes 'in act'. The specific increment over the adjective is therefore not a new concept but a new predicative force — the word now asserts that a thing is so done, not merely that it concerns doing. Illustrative example (constructed to display the sense, not a quotation): 'The pledge was actually fulfilled, not merely promised.' The opposed term throughout this stage is intention, name or promise, i.e. the axis is deed versus word. A note on granularity: the frozen label carries both 'in act, in deed' and 'practically'. These are one function, not two — 'practically' here is the same value seen in a different syntactic environment (modifying a predicate rather than a whole state of affairs), not an independently conventionalized sense, and no separate stage is warranted.

S2 — 'in reality, in fact' — the veridical value opposed to appearance

The first genuine change is a semantic broadening with a change of the opposed term (E15). The contrast partner shifts from 'in intention, in name' to 'in appearance, seemingly, ostensibly', and correspondingly the domain of the adverb widens from deeds performed to facts obtaining: 'in act' generalizes to 'in fact'. The mechanism is broadening by loss of the action restriction on the predicated subject-matter, together with the substitution of a new antonym that licenses the wider range. The increment is that the word can now be predicated of any state of affairs whatsoever, not only of actions, and that its function is veridical assertion rather than deed-assertion. Illustrative example (constructed): 'The pool looks shallow but is actually two metres deep.' Here the adverb does not say that anything was done; it says that the reality behind an appearance is otherwise than the appearance. The parallel drift of the adjective to 'real, existing in fact' (E20) models and supports this adverbial value, but the adverbial shift is not reducible to it: the adverb's veridical force is a predicate-level assertion, whereas the adjective's is a modifier. A note on granularity: the frozen label pairs 'in reality' and 'in fact'. These are translational glosses of a single veridical function, not two stages; no record separates them, because the difference between them is one of register of the gloss, not of contextual value.

S3 — 'as it happens, contrary to expectation' — the counterexpectational sentence adverb

Out of the veridical value develops the coded counterexpectational value, by conventionalization of an invited inference (E16). The mechanism is this: in a recurrent context type — a proposition whose truth is asserted against an assumption already available in the discourse — the inference 'this is not what was expected' is repeatedly invited, and it becomes part of what the word itself encodes rather than something the hearer computes from context. The increment is that the word no longer merely asserts reality but presents the asserted reality as deviating from an expectation; the expectation need not be anyone's explicit belief and no interlocutor need be addressed. Illustrative example (constructed): 'I expected the examination to be hard, but it was actually quite easy.' The clause containing actually is true and unremarkable in itself; what the adverb adds is the framing of its truth as a departure from an expectation. This stage is the pivot of the whole later development, and it is the stage at which the word becomes a sentence adverb taking scope over the proposition rather than a predicate adverb. A note on granularity: the frozen label reads 'in fact, contrary to expectation'; the 'in fact' component is the translational approximation inherited from S2, not a second co-present function. The counterexpectational value is a single function, and it is separated from S2 precisely because a proposition can be veridically asserted without any counterexpectation (S2), while the S3 use is odd where no expectation is in play.

S4 — turn-initial other-correction — actually bound to the addressee's prior turn

The counterexpectational value then narrows interactionally (E17). The mechanism is scope narrowing of a specific kind: the expectation that the proposition contravenes is no longer an expectation floating in the context but is attributed to the addressee's preceding utterance or assumed belief, and the adverb is bound to turn-initial position as the slot from which that attribution is made. The increment is that the word now codes other-correction rather than mere counterexpectation — it does dialogic work, treating the addressee's last turn as the thing being set right, and it typically carries a contrastive fall-rise on the adverb. Illustrative example (constructed): A: 'You must be exhausted after that journey.' B: 'Actually, I feel fine.' The same proposition without the adverb would be a bare contradiction; with it, the contradiction is framed as a correction of an assumption the first speaker has just made. This stage is distinguished from S3 by the presence of a prior turn and by the addressee-orientation of the correction: S3 can occur in monologue and in written expository prose, S4 cannot.

S5 — mitigating actually — the corrective edge attenuated under face-work

From the corrective use develops the mitigating use by pragmaticalization under face-work (E18): the corrective edge is attenuated, and the marker is recruited to soften requests, impositions, disagreements and other face-threatening acts. The mechanism is the routinization of the corrective frame in contexts where the speaker's chief problem is not the addressee's false belief but the imposition the utterance itself creates; the corrective semantics is bleached and the face-work becomes the word's principal contribution. The increment is diagnostic and worth stating sharply: in this use the proposition need not contradict anything at all. A request introduced by actually corrects no prior assertion — the marker is doing pure mitigation, signalling that the speaker is aware the request runs against the addressee's plans. Illustrative examples (constructed): 'Actually, could we possibly move Thursday's meeting to the Friday?' and, softening disagreement, 'Actually, I think the second figure is the right one.' S5 is a sibling of S6, not a stage after it: both branch from S4 (E18 and E19 respectively), and the branch matters, because mitigation keeps the other-directed orientation while self-repair reverses it. The two are not merged here for the reason given under S6: they occur in disjoint environments.

S6 — self-repair actually — correction re-targeted at the speaker's own prior utterance

The second branch from S4 is self-repair (E19). The mechanism is a re-targeting of the corrective function from the addressee's utterance to the speaker's own just-uttered plan or statement, with postposed position conventionalized alongside the initial one. The increment is twofold and both parts are functional, not positional accidents: the target of correction becomes the self rather than the other, and the marker acquires a postposable slot that the other-correction use does not admit. Illustrative examples (constructed): initial — 'Actually, let's take the coast road'; postposed — 'I'll come at six — actually, make it seven.' In the second, the adverb follows the material it revises; in the first, it pre-announces the revision. This stage is kept separate from S5 because the environments are disjoint rather than overlapping: the mitigator occurs before requests and impositions and needs no prior utterance to repair, while the self-repair marker requires a preceding utterance of the speaker's own and tolerates final position. A single form with two environment-bound functions is two stages, and merging them would hide exactly the distributional fact that distinguishes them.

Sentence-level function

Actually is a sentence adverb whose removal changes discourse function, not merely the referential or predicative content of the clause: without it a turn loses its corrective, mitigating or revising framing while keeping its proposition intact. It is therefore treated here as a discourse-functional item and the four sub-functions are answered separately below.

Mitigation / face. Actually often serves as a softener of face-threatening acts, though its effect is conditional on intonation and the relation between speakers. Minimal pair: 'Move your car' versus 'Actually, move your car'. The propositional request is identical; what the second adds is a frame in which the speaker presents the imposition as a correction of the way things stand rather than as a demand on the addressee, and signals awareness that the request may cut across the addressee's plans. The same holds for disagreement in mitigating registers: 'This is wrong' versus 'Actually, this is wrong' — the second can lower the cost of contradicting the addressee by coding the contradiction as a small adjustment of the record rather than a confrontation, but the effect may be lost or reversed with marked intonation or unequal status. The mitigating value is recruited precisely where the corrective value would be too strong, which is why the two must be kept as separate stages rather than one 'hedging/correcting' stage.

Self-repair / postposed. Actually marks the speaker's own revision of a plan or statement and occurs in both initial and final position. Initial: 'Actually, let's meet at noon.' Final/postposed: 'Let's meet at noon — actually.' Medial parenthetical: 'The figure — actually, the revised figure — is forty.' The initial position pre-announces the revision, so the hearer processes the following material already flagged as a replacement; the postposed position is an afterthought, added once the original formulation has been uttered, and it typically carries a fall-rise and a slight break in tempo. The availability of the postposed slot is the single clearest syntactic difference between the self-repair use and the other-correction use, which is primarily initial-position-bound.

Negation / interrogation. Under clausal negation actually normally survives outside the scope of the negator, because it is a speaker-oriented adverb: 'It isn't actually raining' is commonly interpreted as 'in fact it is not raining', but it can also be read as 'it is not really raining' with a contrast such as drizzle or mist, so the scope is not unique. There is, however, a genuine scope ambiguity with the veridical value, since 'He's not actually a doctor' can mean either 'in fact he is not a doctor' or 'he is not really a doctor' (a denial of the stronger claim). In questions the pattern contrasts with the declarative: 'Is he actually coming?' asks about the reality of the proposition and typically carries a sceptical, counterexpectational colouring that the plain 'Is he coming?' lacks; 'What actually happened?' asks for the reality as against the reported version; and turn-final 'Is he coming, actually?' signals that the speaker is reopening a matter already treated as settled. Turn-initial actually inside a question ('Actually, is he coming?') does not primarily correct a prior claim but shifts or reopens the topic, which shows that the corrective force is positionally conditioned rather than inherent in the word.

Turn-initial position. Occupying the very first slot of a turn does three things at once, and they are separable in the record. It frames the whole upcoming utterance pre-emptively — the hearer knows before hearing the clause that it is to be read as a correction or a shift. It claims the floor at a transition-relevance point, which is why turn-initial actually is frequent in multiparty talk where more than one speaker is bidding for the next turn. And it marks the upcoming material as a correction of, or a departure from, what has just been said or assumed. Initial and medial uses differ in force: initial actually takes scope over the turn and is heard as corrective or topic-shifting, whereas medial actually ('I actually think the second figure is right') is more integrated, emphasises the truth of the predicate, and can lose the corrective force altogether and shade into a mild intensifier. In some varieties — British English is the usual impression, though this is [unverified] here — turn-initial actually is additionally recruited as a general hedging preface even where nothing is being corrected.

§10Phonological & orthographic development

The formal history from Latin to Modern English is short and rule-governed at every step, so the honest report is a list of named processes rather than a statement that nothing irregular happened.

First, the stressed vowel. Latin actualis had a long ā in the stem; in the English borrowing of actual the vowel stood in a closed syllable and was shortened, exactly as in act, fact and pact; the resulting Middle English short /a/ was fronted to /æ/ in the seventeenth century, giving Modern English /ækt/ and /ˈæktʃuəli/. The shortening is regular loanword behaviour in this prosodic position, not a sporadic change.

Second, the ending. Latin -ālis gives Old French -el, but not by vocalization of preconsonantal /l/; the -el results from the Old French change of stressed /a/ in open syllable to /e/, as in mortālem > mortel and tālem > tel. The vocalization of /l/ is a different path: -als- > -aus- gives forms like -aux, not -el. Old French has actuel and mortel; the Anglo-Norman and Latinizing -al preserves the Latin shape unchanged. This is the only formal difference between the two borrowing channels, and English generalized -al.

Third, yod coalescence. The pronunciation /ˈæktʃuəli/ does not come from Latin /tj/: Latin āctuālis has a vowel /u/ after /t/, not /tj/. The sequence /tj/ arises later, through the French and English development of the vowel to a yod (e.g. French /y/ borrowed as /ju/), and it is this later /tj/ that coalesces to /tʃ/, a regular English process seen in nature, virtue, fortune and creature. The coalescence is therefore fully regular here: it is not a peculiarity of this word but the normal fate of the later /tj/ sequence in English loans. The pre-coalescence pronunciation /ˈæktjuəli/ survives as a conservative or spelling-influenced variant, chiefly in older RP and among speakers who do not coalesce.

Fourth, reduction and syncope. The post-tonic syllables reduce to schwa, and in this high-frequency item the schwa is frequently syncopated, giving /ˈæktʃəli/ and then /ˈæktʃli/. This is the same reduction seen in finally, usually and literally, and its extent correlates with frequency and with speech rate rather than with region.

Fifth, deaffrication in rapid speech. The variant /ˈækʃli/, with /tʃ/ simplified to /ʃ/, is a connected-speech reduction of the same high-frequency item; it is promoted by frequency and rapid speech, but the present evidence does not exclude regional variation or possible lexicalization.

The attested variants, then, are: /ˈæktjuəli/ (conservative, spelling-influenced), /ˈæktʃuəli/ (the standard full form), /ˈæktʃəli/ and /ˈæktʃli/ (reduced), and /ˈækʃli/ (rapid). The full form is the citation form; the reduced forms predominate in connected speech.

The orthography stands in a double tension with the pronunciation, and both tensions should be named rather than passed over. First, spells the Latin consonantism /kt/, while the later /tj/ coalescence produces /tʃ/ after the /k/; the spelling has not followed the sound, and this is exactly what licenses the spelling pronunciation /ˈæktjuəli/ as a variant. Second, the word is written with although English has no contrastive geminate /lː/, so the double letter is purely orthographic: it reflects the morpheme boundary actual + -ly, and it does not mark vowel length. The Middle English spelling actualy with a single is recorded as a variant in the word's early history; no causal explanation is offered for the later doubling. The written word also carries four syllables' worth of letters for what are, in ordinary speech, three or four: the mismatch between the conservative spelling and the reduced pronunciation is the most conspicuous formal fact about the modern word.

§11Formal network model

§11.1 Node inventory

§11.2 Edge types used

G, M, B, C, S, P

§11.3 Edge list

§11.4 Graph (computed from the edge list — no model participation)

L0  N1  PIE *h₂eǵ- ▲
L0  N8  Medieval Latin actualiter
L0  N15  Greek ἐνέργεια (energeia)
L0  N16  Greek δύναμις (dynamis)
L1  N2  Latin agere ▲
       ← N1 (G) [E1]
L1  N17  Latin actus 'actuality' (technical sense) ▲
       ← N15 (C) [E5]
L1  N19  Greek ἄγω (agō)
       ← N1 (G) [E23]
L1  N20  Sanskrit ájati
       ← N1 (G) [E24]
L1  N21  Old Norse aka
       ← N1 (G) [E25]
L1  N23  Armenian acem
       ← N1 (G) [E27]
L1  N29  Latin potentia ▲
       ← N16 (C) [E6]
L1  N30  Old Irish aigid
       ← N1 (G) [E28]
L2  N4  Latin act- (derivational stem) ▲
       ← N2 (M) [E2]
L2  N18  Scholastic actus / potentia opposition ▲
       ← N17 (C) [E7]
       ← N29 (C) [E8]
L3  N3  Latin actus, -ūs ▲
       ← N4 (M) [E3]
L4  N22  Latin actu- (noun stem of actus) ▲
       ← N3 (M) [E26]
L5  N5  Late Latin actualis ▲ ★
       ← N22 (M) [E4]
       ← N18 (S) [E9]
L6  N6  Old French actuel ▲
       ← N5 (B) [E10]
L6  N28  German aktuell
       ← N5 (B) [E30]
L7  N7  Middle English actual ▲
       ← N5 (B) [E11]
       ← N6 (B) [E12]
L7  N27  Modern French actuel(lement)
       ← N6 (S) [E29]
L8  N9  'in act, in deed; practically' ▲
       ← N7 (M) [E13]
       ← N8 (B) [E14]
L8  N24  Modern English actual (adjective)
       ← N7 (S) [E20]
L9  N10  'in reality, in fact' ▲
       ← N9 (S) [E15]
L9  N25  English /ˈæktʃuəli/ ▲
       ← N9 (P) [E21]
L10  N11  'in fact, contrary to expectation' ▲
       ← N10 (S) [E16]
L10  N26  English /ˈæktʃ(ə)li/
       ← N25 (P) [E22]
L11  N12  turn-initial corrective actually ▲
       ← N11 (S) [E17]
L12  N13  hedging / mitigating actually
       ← N12 (S) [E18]
L12  N14  self-repair actually
       ← N12 (S) [E19]

30 nodes / 30 edges, 1 connected component(s). ▲ = cut vertex (deleting it splits the undirected graph; what each one detaches is listed in §11.5); ★ = highest betweenness. Levels = longest distance from a source node (cycles broken at back edges, listed in §11.5).

§11.5 Centrality and paths (computed, reproducible)

• Size. 30 nodes, 30 edges.
• Connectivity. 1 connected component(s) on the undirected projection.
• Most connected nodes. N1 (PIE *h₂eǵ-; degree 6).
• Highest betweenness (brokers between traditions). N5 (Late Latin actualis; 0.6724), N7 (Middle English actual; 0.4901), N9 ('in act, in deed; practically'; 0.4557), N22 (Latin actu-; 0.4433), N3 (Latin actus, -ūs; 0.4138).
• Cut vertices. 16 of 30 nodes, deleted, leave the undirected graph in more than one piece (each deletion recomputed).
• Load-bearing — deletion splits the graph into at least two pieces of ≥ 2 nodes, so it separates two real sub-chains (piece sizes shown): N5 (Late Latin actualis) → 13 | 10 + 5 + 1; N22 (Latin actu-) → 20 | 9; N7 (Middle English actual) → 19 | 9 + 1; N3 (Latin actus, -ūs) → 21 | 8; N4 (Latin act-) → 22 | 7; N2 (Latin agere) → 23 | 6; N9 ('in act, in deed; practically') → 21 | 5 + 2 + 1; N10 ('in reality, in fact') → 25 | 4; N11 ('in fact, contrary to expectation') → 26 | 3; N18 (Scholastic actus / potentia opposition) → 25 | 2 + 2.
• Terminal-branch nodes — deletion only detaches nodes that no other edge passes through, so the descent chain survives without them: N1, N6, N12, N17, N25, N29.
• Being a cut vertex is a property of the graph's shape — a long descent chain makes most of its nodes cut vertices. It is not evidence that the argument depends on those nodes.
• Longest transmission path. N1 → N2 → N4 → N3 → N22 → N5 → N6 → N7 → N9 → N10 → N11 → N12 → N13 (12 edges, edge ids E1, E2, E3, E26, E4, E10, E12, E13, E15, E16, E17, E18). On the cycle-free version of the graph there are 2 distinct paths from its first to its last node.

All quantities above are computed by src/network_metrics.py from the edge list; nothing here is written by the generating model.

§12Collateral family

The derivational family is large and its shape is set by Latin stem morphology rather than by English word formation. Four stems carry it. The act- stem, the participial stem of agere, supplies act, action, active, activity, actor and actuate. The noun stem actu-, taken from the fourth-declension actus, carries the whole actual- group: actual, actually, actuality, actualize, actualization, actualism. The agent- stem, the present-participle stem, supplies agent and agency. Agile and agility are separate derivatives from agilis, not from the agent- stem; agenda comes from the gerundive stem agend-. The frequentative agit- supplies agitate and agitation. And prefixed Latin verbs of the same root supply exact (exigere), exigent, cogent (cogere), ambiguous (ambigere), intransigent (transigere). Navigate is not one of these prefixed verbs; it is a compound of navis + agere ('to drive a ship'). All of these are semantically distant enough that English speakers do not perceive them as relatives. Actuary belongs here too, from Latin actuarius, the accountant or shorthand writer — a reminder that actus named not only a deed but a written record of one.

Two structural points deserve emphasis. First, the family entered English by three different routes and at three different periods: the core nouns through Old French (act, action, activity), the learned adjectives and verbs through both semi-learned French and learned Latin (actual, actuate, actuality), with actual received through semi-learned Old French actuel and learned Latin actualis, the record not apportioning their shares, and the philosophical noun actus as a modern re-borrowing that lives on mainly in the phrase in actu, in Thomist writing, and in the common-law phrases actus reus and actus Dei. The same Latin form can therefore appear in English twice with different histories, which is why actual and actus are both present in the list without being variants of one another. Second, the family's semantics split along a line that recapitulates the act/potency contrast: actual and actuality name what is realized, actualize names the bringing into realization, actuate names the causing of operation (and, of a person, the supplying of a motive), while act and action name the doing itself. The contrast between what is in act and what is only possible, which is the scholastic core of actualis, is thus still legible in the modern derivational set.

At the level of the parallel branches the family's most instructive fact is a false friend. English actual means 'real, existing in fact', but French actuel, German aktuell, Italian attuale, Spanish actual and Portuguese atual all mean 'present, current, topical'. The same Latin adjective was conventionalized along different lines in different languages: English kept the veridical value and developed the counterexpectational and interactional values from it, while French, German and Spanish kept the temporal value that Latin actualis also had ('in act, hence present'). The result is systematic misreading in both directions. An English speaker who says 'actually' and a Spanish speaker who says actualmente are not saying the same thing: actualmente means 'currently, at present', and using it to translate actually in its corrective sense produces a sentence that is merely temporal. Conversely, a French or German speaker learning English must learn that actual does not mean 'current', and that the adverb has acquired a dialogic corrective function that no cognate shares. Etymology here predicts the shared form and the shared core, and predicts nothing at all about the modern meanings — which is precisely the lesson the cognate set teaches.

Members: actual, actually, actuality, actualize / actualise, actualization, actualism, actualist, actualistic, act, action, active, activity, actor, actress, actuate, actuation, actuator, agency, agent, agenda, agile, agility, agitate, agitation, exact, exigent, cogent, ambiguous, intransigent, navigate, actuary, actus (English philosophical loan), in actu (borrowed phrase)

§13Hypotheses

H1 — Medieval Latin actualiter as a reinforcement channel for the English adverb

The English adverb is regularly formed from actual by -ly (E13), but the formation has an exact Latin counterpart in actualiter (E14). This hypothesis proposes that actualiter supplied collocational and semantic models for actually without supplying its morphology. [unverified] The proposal is derived from the coexistence of the native -ly formation and the Latin -iter adverb in the frozen network, and from the fact that both mean 'in act, actually'.

Prediction: In the dated quotation record of Middle English and early Modern English, the earliest occurrences of actually should cluster in texts translated from Latin or in scholastic, theological and legal prose, and should share collocations with actualiter such as 'in act' and 'actually X' where X is a technical predicate. A native -ly formation with no Latin model would instead appear first in original English administrative or literary prose with no such clustering.

Falsified by: If the earliest English occurrences of actually are found in original native texts with no Latin source and show no collocational overlap with actualiter, the reinforcement hypothesis fails; the -ly formation would then be independent of the Latin adverb.

Derivation: N8, N7, N9, E14, E13

H2 — The corrective use precedes the mitigating and self-repair branches

The frozen trajectory places N12 (turn-initial correction) before N13 (mitigation) and N14 (self-repair), with E18 and E19 as separate branches from N12. This hypothesis makes the chronological claim that the dialogic corrective use is earlier than both branches, not merely structurally prior. [unverified]

Prediction: A dated corpus of Early Modern English drama, trial transcripts and familiar correspondence should show turn-initial corrective actually (N12) before pre-request mitigating actually (N13) and before postposed self-repair actually (N14); monologic philosophical and scientific prose should show the veridical and counterexpectational values but not the corrective use.

Falsified by: If the earliest examples of mitigating or postposed self-repair actually predate any turn-initial corrective use in the dialogic corpus, or if corrective actually is confined to monologic written prose, the chronological claim fails.

Derivation: N10, N11, N12, N13, N14, E16, E17, E18, E19

H3 — The French channel supplied the earliest English adjective

The frozen network records both Old French actuel with -el (N6, E10) and learned Latin actualis with -al (N5, E11) as inputs to Middle English actual (N7), and E12 says their relative contribution is not distinguished by the record. This hypothesis proposes that the French channel is chronologically earlier in English. [unverified]

Prediction: A manuscript survey of Middle English spellings should show -el forms such as actuel in Anglo-Norman-influenced legal and administrative texts earlier than -al forms in Latin-literate scholastic texts; the -al forms should become dominant only after the general adoption of Latinizing -al.

Falsified by: If the earliest Middle English stratum shows -al forms only and no -el forms, or if -el forms are later than -al forms, the priority of the French channel does not hold.

Derivation: N6, N7, E10, E11, E12

H4 — English actually once carried the temporal value 'at present, currently'

French actuel and actuellement conventionalized the temporal value 'present, current' (N27, E29), while English actual developed the veridical value (E20). This hypothesis proposes that English actually also had the temporal value in its early history and lost it under pressure from the veridical sense. [unverified]

Prediction: A dated corpus of Early Modern English news, letters and travel writing should show actually used as 'currently, at present' in contexts where French actuellement would be the translation equivalent, with the temporal use declining as the veridical use expands; historical dictionaries should record the temporal sense.

Falsified by: If no attestation of temporal actually can be found in the English record, and all early examples mean 'in act' or 'in fact', the hypothesis fails.

Derivation: N6, N7, N9, N10, N27, E12, E15, E29

H5 — The Arabic-Hebrew channel shaped the scholastic formulae of actualis

The frozen cross-disciplinary section records the Arabic pair al-quwwa/al-fiʿl and the Hebrew pair kōaḥ/pōʿal as reinforcement channels for the Latin actus/potentia opposition (E5–E8), while leaving their weight unsettled. This hypothesis proposes that the Latin formulae in actu and actualis were shaped in part by Arabic-to-Hebrew translation patterns; the Latin Tahāfut renderings belong to a later sixteenth-century Calonymus stage, not to Kalonymus ben Kalonymus [unverified].

Prediction: Latin translations made from Arabic and Hebrew — for example versions of Averroes and the Latin Tahāfut traditionally attributed to Kalonymus ben Kalonymus — should show actualis and actualiter in formulae that calque bi-l-quwwa wa-l-fiʿl and be-kōaḥ u-ve-pōʿal, and these formulae should enter scholastic Latin through those translations or alongside them.

Falsified by: If the Latin actus/potentia formulae are fully attested in direct Greek-to-Latin translations before any Arabic-Hebrew translation and show no distinctive calque patterns from Arabic or Hebrew, the Arabic-Hebrew channel did not shape the scholastic formulae.

Derivation: N15, N16, N17, N18, N29, E5, E6, E7, E8

H6 — The compressed pronunciation is a frequency-driven modern development

The frozen phonology records /ˈæktʃuəli/ (N25) and the compressed /ˈæktʃ(ə)li/ (N26) with schwa syncope (E22). This hypothesis proposes that the compressed form spread in high-frequency conversational use rather than being an old or dialectal variant. [unverified]

Prediction: Pronouncing dictionaries and recorded speech corpora should show the full form /ˈæktʃuəli/ dominant in earlier material and the compressed form /ˈæktʃ(ə)li/ increasing later, with higher rates in high-frequency conversational registers and lower rates in formal read speech.

Falsified by: If nineteenth-century pronouncing dictionaries already record /ˈæktʃli/ as a standard pronunciation, or if early recordings show compressed forms as frequent as modern conversational corpora, the frequency-driven modern-development hypothesis fails.

Derivation: N25, N26, E21, E22

H7 — Mitigating actually is more advanced in American than British English

The frozen trajectory derives mitigation (N13) from corrective actually (N12) by pragmaticalization under face-work (E18). This hypothesis proposes a sociolinguistic asymmetry: the mitigating use is more frequent and more conventionalized in American English than in British English. [unverified]

Prediction: Comparable spoken corpora of American and British English should show higher rates of pre-request and downgrading actually in American English, and historical corpora should show mitigation collocations such as 'actually could you' first in American sources.

Falsified by: If comparable spoken corpora show no transatlantic difference in mitigation rates, or if British sources show the mitigation collocations earlier than American sources, the asymmetry hypothesis fails.

Derivation: N12, N13, E18

§13aRejected hypotheses

Rejected: English actually is a native Germanic formation continuing the root *h₂eǵ- through Old English, with no Latin mediation.

Reason rejected: Old English shows no verb of this root in the required sense — the acan that is sometimes cited means 'to ache', and its affiliation with the root is disputed — the root has no derivatives in actual-, and the word's shape is Latin actu- plus -al plus -ly, with no Germanic stem or sound correspondences. The falsification condition for native descent — a Germanic stem and regular Germanic phonology — is met against it.

Rejected: English actual was borrowed only from Latin actualis, with no French channel.

Reason rejected: Old French actuel is attested with the French reflex -el of Latin -ālis, and Middle English spellings in -el show the French channel in the record. The condition that would support a Latin-only route — absence of French forms and of -el spellings — is not met.

Rejected: English actually is a direct borrowing of Latin actualiter, not an -ly formation.

Reason rejected: The English form ends in native -ly, not in Latin -iter, and no English form actualiter is borrowed as the ordinary adverb. The morphological condition for direct borrowing is met against it; actualiter can at most have reinforced the -ly formation.

Rejected: Modern English actually continues French actuellement in form and sense.

Reason rejected: French actuellement means 'at present, currently', while English actually means 'in fact, contrary to expectation'; the English form has -ly, not French -ment. The semantic and morphological conditions for direct continuation are met against it.

Rejected: The counterexpectational use of actually is borrowed from French actuellement.

Reason rejected: French actuellement is temporal and does not code counterexpectation; French uses en fait and en réalité for the corrective sense. The condition for a French source — a French counterexpectational adverb of the same form — is not met.

Rejected: Latin actualis was built by adding -ālis directly to the nominative noun actus (yielding *actusālis) rather than to the noun stem actu-.

Reason rejected: The suffix -ālis attaches to a stem, and the parallel adjectives manuālis (manus, stem manu-), spirituālis (spiritus, stem spiritu-), and rītuālis (rītus, stem rītu-) show that the stem vowel -u- of the fourth-declension noun is retained before -ālis. Adding -ālis directly to the nominative actus would yield *actusālis, which is not the observed form; the observed form reflects the noun stem actu-.

Rejected: The PIE root is h₁eǵ- or h₃eǵ- rather than *h₂eǵ-.

Reason rejected: Latin a- and Greek ἀ- reflect the a-colouring of h₂; h₁ would yield e-colouring and h₃ o-colouring, and no Anatolian cognate supporting either rival is attested in the record used here. The vowel-colouring condition is met against h₁ and *h₃.

Rejected: The in actually represents a geminate consonant.

Reason rejected: English has no contrastive geminate /lː/, the pronunciation has a single /l/, and single spellings are recorded in the word's early history. The phonetic condition for gemination is met against it.

§14Conclusion

The history of English actually is a descent series with a semantic overlay, and the two must be reported separately because they are warranted differently. The formal backbone is regular at every step: PIE h₂eǵ- 'drive, lead, set in motion' is continued by Latin agere (laryngeal loss and a-colouring); the perfect-participle stem act-, with Lachmann's law lengthening, is the base of actus; the -ālis adjective actualis is built on the noun stem actu- of actus, with -ālis suffixed to actu- (actu- + -ālis > Latin actuālis), not to the bare act- stem (which would give actālis) or to the nominative actus; actualis reaches English through Old French actuel with French -el and through learned Latin with -al, the two channels converging on one English shape, their shares unrecoverable from the spellings; the adverb is the -ly formation on actual, with Medieval Latin actualiter as a possible formal and collocational reinforcement rather than a morphological source. The English word is therefore loan material, not inherited material: Old English offers no reliable reflex of the root in this sense — the form usually cited, acan, means 'ache' and its affiliation is disputed — so nothing Germanic feeds into actual, and the Romance and Germanic branches — French actuel and actuellement, German aktuell, Spanish actual and actualmente — are parallel developments, not sources. What makes the word interesting is not this backbone but what happens on it. Latin actus acquired the value of Greek ἐνέργεια and potentia the value of δύναμις by conceptual translation, sense-for-sense rather than morpheme-for-morpheme, and it was within the resulting actus/potentia opposition that actualis specialized to 'belonging to what is in act, actual as opposed to potential', with actualiter as its regular adverb. That opposition then travelled in formulaic form — in actu signato and in actu exercito, actus purus, gratia actualis, and later in English law's actual notice against constructive notice and actual bodily harm, and in philosophy's actual world, actuality operator and actualism — and it is these phrases, not the philosophical prose, that kept the technical value alive after the metaphysics was forgotten. The semantic trajectory that follows is a chain of independently attestable states, each with its own mechanism and its own increment: 'in act, in deed' opposed to intention or name, formed by adverbial suffixation; 'in reality, in fact' opposed to appearance, by broadening with a change of the opposed term; 'as it happens, contrary to expectation', by conventionalization of an invited inference, at which point the word becomes a sentence adverb; turn-initial other-correction, by interactional scope narrowing binding the value to the addressee's preceding turn; and then two branches from that corrective stage rather than a single line — mitigation under face-work, where the proposition need not contradict anything at all, and self-repair, where the target of correction becomes the speaker's own prior utterance and a postposed slot is conventionalized. The branching is a substantive claim, not a presentational convenience: the two uses are not confined to disjoint environments, and a speaker can mitigate while self-repairing. The phonology is short and fully rule-governed — vowel shortening in a closed syllable, yod coalescence of /tj/ to /tʃ/, schwa syncope, deaffrication in rapid speech — and it leaves the modern word in a visible double tension with its own spelling, for a coalesced palatal and for a single /l/. Two things should be said plainly about what this reconstruction does not settle. First, the chronological order of the branches is undated: the claim that corrective actually precedes both mitigation and self-repair is a structural inference from the network, and it is exposed in the hypotheses as a dated-corpus bet rather than asserted. Second, the weight of the Arabic-Hebrew channel in shaping the scholastic formulae of actualis is not determined here; the Arabic al-quwwa/al-fiʿl and the Hebrew kōaḥ/pōʿal certainly carried the distinction and certainly fed it back into Latin through translations made from Arabic and Hebrew, but whether they shaped the Latin formulae or only reinforced them cannot be read off the forms, and the reconstruction claims only the reinforcement. What the word finally teaches is the disjunction between etymology and meaning: the same Latin adjective settled into a temporal value in French, German and Spanish and into a veridical and then interactional value in English, so that actuel and actually, formally the closest of cognates, are systematic false friends. The formal chain predicts the shared shape and the shared core, and predicts nothing about the modern discourse functions — which are the product of English-internal semantic and pragmatic change and must be reconstructed as such.