ARGUS — Explanatory Glossary

Common meaning a possible descriptive expression — the absence here concerns an element whose result remains unknown. ARGUS makes it an additional property of an absence already qualified.

In ARGUS an absence with indeterminate result is the case where evidence necessary to a central proposition is missing, but its result could equally confirm or refute the thesis. The analyst retains the qualification obtained at the first gate and adds the note “indeterminate result”. This property bars the opening of the second gate: so long as the possible result of the missing evidence is indeterminate, the analyst cannot demonstrate that integrating it would contradict the conclusion or require a major overhaul of the causal mechanism.

Not to be confused with indeterminate result is not a new qualification replacing strategic omission, mere point of weakness, non-attributable absence or a first-gate indeterminacy. It is an additional property grafted onto the qualification already obtained. Its only effect is to forbid concluding that the absence is structural.

See also structural omission · strategic omission · first gate, second gate · missing evidence · central proposition

Common meaning a possible descriptive expression — a revision may be said to be acknowledged where it is explicitly owned. ARGUS turns that idea into a precisely defined exclusion.

In ARGUS acknowledged revision is the second exclusion of the modal constancy check. It requires two cumulative conditions: the author explicitly revises the proposition posited at the strong degree, and effectively ceases to rely on its force to conclude, adjusting the conclusion. Before granting the exclusion the analyst verifies that the conclusion does not carry the force of the strong pole again, verbatim or reworded.

Not to be confused with acknowledging that an assertion was too absolute is not enough. If the conclusion, or the recommendation that depended on the strong pole, survives intact, the exclusion does not apply: the structure is a disguised retraction and the check triggers.

See also modal constancy · disguised retraction · strong pole, retracted pole · scope partition

Common meaning a possible descriptive expression — propaganda may be adapted to an audience or a context. ARGUS makes it a precise qualification tied to the audience’s credibility codes; the string of words is not claimed as a creation.

In ARGUS adapted propaganda is propaganda that adopts the intellectual, moral or aesthetic codes of the audience it targets. It may therefore be sourced, nuanced, cautious, technically precise or stylistically sophisticated where those are the forms of credibility its addressee expects. The protocol accordingly forbids excluding propaganda on the basis of formal qualities alone: what counts is the function of those qualities in opening or closing interpretation. Since the term propaganda is reserved for Levels 3 and 4 of 7.c, adaptation to the audience never by itself produces that qualification.

Not to be confused with adapted does not denote an additional degree of propaganda. The degree is assigned in 7.c according to interpretive closure. Adaptation of persuasive methods to the expectations of the target audience is among the criteria used to assign that level, but it becomes an indication of closure only when accompanied by asymmetries, strategic omissions, informational circularity, a neutralised counter-argument, or other mechanisms that prevent contrary elements from modifying the conclusion. The only descriptors orthogonal to the level are the rhetorical mode and origin or function.

See also principle of rhetorical destination · degree of interpretive closure · rhetorical mode · sophisticated · interpretive closure

Common meaning a material thing, or that on which an action, a thought or an examination bears.

In ARGUS the analysed object is the whole document submitted to ARGUS; the scope is only the part of it examined under the protocol. The object is stated in one line at the head of the analysis, with what the document says of itself — medium, date, and mention of another version where it signals one. Declared object names the same document from the point of view of that declaration. Every external verification is external to this object. A section excluded from the scope remains internal to the object: consulting it does not for that reason alone become external verification, and its use under the bridge of 0.a bis enters none of the three counters of 3.B. The accessibility of a piece of information is assessed at the publication date of the declared object, not at the date of the analysis.

Not to be confused with the object is not the scope. Excluding a section removes it from the scope, not from the object.

See also scope · external verification · regime applied to excluded sections

Origin the modern psychological sense refers to the anchoring and adjustment heuristic described by Amos Tversky and Daniel Kahneman in “Judgment under Uncertainty: Heuristics and Biases” (1974). The anchoring effect names the persistent influence of an initial value or piece of information on a later judgment.

Established meaning anchoring bias is the tendency of a judgment to remain influenced by a prior piece of information, value or starting point, even where that influence is not warranted by the object being assessed.

In ARGUS anchoring designates the risk that an analytical judgment is influenced by what was seen or assessed before, rather than by the text under examination alone. Step 8 requires the analyst to flag a risk of anchoring where an analysis forms part of a series. For corpora, C.3 distinguishes two risks to be checked and declared separately: sequence anchoring, where texts treated later receive systematically more severe or more favourable judgments without internal justification of their own; and chronology anchoring, where the analyst imputes to an earlier text what later texts have made visible, notably where the order of analysis is not the order of publication.

Not to be confused with observing that one analysis comes after another is not enough to establish an anchoring bias. ARGUS requires looking for a drift of judgment without internal justification, and flagging it as a risk of series bias. Nor is anchoring the same as series conformism, which consists in aligning one’s judgment with previous analyses rather than undergoing, more broadly, the effect of a starting point or an order of treatment.

See also series bias · series conformism · cross-text consistency check · corpus · blind rereading

Origin an English rendering of the German assertorisch, a term formed on the Latin assertorius, which is attested well before Kant in juridical and theological use. Kant gives assertorisch its modal-logical sense in his classification of judgments, which distinguishes problematic, assertoric and apodictic judgments.

Established meaning in Kantian philosophy, an assertoric judgment states a matter of fact without positing it as necessary; it occupies the middle degree of the triad.

In ARGUS an assertoric proposition is one the author posits as a fact; it is the maximal degree of commitment. The ARGUS scale has four degrees — assertoric, probable, hypothetical, suspended — and serves to draw up the modal map of the load-bearing propositions at Step 2, then to conduct the modal constancy check at Step 6.

Not to be confused with ARGUS uses neither apodictic nor problematic. More importantly, assertoric is for Kant a middle degree, because his scale classifies the logical status of a truth; it is in ARGUS the highest degree, because the scale measures the author’s commitment to their proposition. The two scales do not measure the same thing.

See also degree of commitment · polarity · modal map · modal constancy · probable, hypothetical, suspended

Established meaning an asymmetry is an absence of symmetry or correspondence between two comparable elements, structures or relations; by extension the word denotes a non-reciprocal difference of position, property or treatment.

In ARGUS an asymmetry is a difference of treatment between actors, sources, positions or requirements of proof, which becomes a flaw where it is not explicitly and validly justified. In 3.H ARGUS tests in particular whether a flaw charged against one camp is also looked for in the speaker or in the text’s own authorities; in 3.J it asks whether actors are described according to systematically different standards. An unjustified asymmetry may be qualified as performative contradiction, unwarranted epistemic privilege or reflexive blindness.

Not to be confused with not every difference is a faulty asymmetry. Two actors may be treated differently if that difference is explicitly and validly justified. In interest-based explanation, a justification of principle drawn from the text’s own theoretical framework is not sufficient if that framework is applied to one side only. Conversely, an established asymmetry does not by itself entail a diagnosis of propaganda: to reach Level 3 the protocol requires converging mechanisms protecting the conclusion against contradiction.

See also epistemic symmetry · treatment of actors · performative contradiction · interpretive closure

Origin in the statistical and probabilistic use intended here, the notion is a standard one in applied statistics, current well before the literature on judgment under uncertainty borrowed it; in that literature base-rate neglect names the failure to take prior frequency into account.

Established meaning in statistics and Bayesian reasoning, a base rate is the frequency or prevalence of an event, state or category in the reference population, before a particular piece of information is taken into account.

In ARGUS the base rate is the prevalence one must know in order to interpret a claim of detection, screening, attribution or profiling correctly. Appendix 4 asks, under D. Base rate, whether that prevalence is given. A very high detection rate is not by itself sufficient to establish the probability that a designated individual actually belongs to the category sought; without the base rate, the protocol treats the detection rate as unusable. This control is applied systematically to texts dealing with security, health or fraud.

Not to be confused with the base rate is not the detection rate. The first describes the prior frequency of the phenomenon in the population; the second describes the performance of a device or a criterion of detection. A high accuracy announced for the latter does not replace the missing prevalence. And base rate here has nothing to do with the interest-rate sense the phrase carries in banking.

See also Appendix 4 · denominator · prevalence · arithmetical and statistical control · zetetic

Common meaning an examination made to establish that something is correct, safe or as expected; also, in ordinary English, a restraint on something — to hold in check — and, in North American usage, a restaurant bill, a written order to a bank, and a tick placed against an item.

In ARGUS a check is a verification question defined before it is conducted and counted as a single unit. It corresponds to a verificatory proposition or a protocol condition, formulated at the moment the check is undertaken in 3.B. Sub-elements necessary to decide that same proposition are not counted separately; where some are established and others are not, the check is partial. Two checks are distinct only if they bear on two propositions each producing an autonomous result for the analysis. Each check receives a status — successful, partial or unsuccessful — and a mode of access — provided source or external search; the two axes are independent and must not be presented as a single list of categories. Every analysis declares three distinct numbers, one per status, and justifies the number of successful checks if it is nil. The partition announced may not be altered after the fact in order to transform one partial check into one successful check and one unsuccessful check.

Not to be confused with a check is not an external verification. External verification qualifies the character of an operation; the check is its unit of count. An analysis may have conducted several external verifications and still have no successful check to declare, if none established what was sought. Nor is a check a careful reading: it is a question posed in advance and counted, not attention paid.

See also successful, partial, unsuccessful · mode of access · external verification · analysed object

Common meaning a reservation is a restriction placed on what has just been said, a caution expressed.

In ARGUS the closing reservation is one of the four locations where a load-bearing proposition may appear; it is not a degree of commitment. It is recorded by the modal map of Step 2 alongside the opening, the body and the concession. Where the same proposition, on the same referent and at constant polarity, appears at several points under different degrees, each occurrence is noted with its location; the modal constancy check of Step 6 depends on that record.

Not to be confused with the closing reservation establishes no retraction by itself. A low degree at the end of a text may fall under consistent restraint, acknowledged revision, or a scope partition.

See also modal map · concession · modal constancy · retracted pole

Common meaning something granted to an opponent in a discussion.

In ARGUS a concession is first of all a location in the text where a low degree of commitment may appear; it is then called opening or immunising according to the effect it produces. At Step 2 the concession is a location, alongside the opening, the body and the closing reservation, and not a degree of commitment. In 3.K it is opening when it makes a different or more uncertain conclusion possible, and immunising when it gives an appearance of balance while protecting the central conclusion from criticism.

Not to be confused with the presence of a concession does not prove the balance of the text. A concession that constitutes the retracted pole of an established modal retraction cannot be credited to the text in 7.b — neither as an effective counter-argument, nor as an opening concession, nor as a point of rigour.

See also effective counter-argument · locked nuance · modal constancy · strong pole, retracted pole

Common meaning a transparent composition — a restraint is consistent when it holds from one end to the other. The technical qualification and its procedural effect are proper to ARGUS.

In ARGUS consistent restraint is the first exclusion of the modal constancy check: a proposition cautious from one end to the other is not requalified as a retraction merely because a closing reservation restates that caution. The check verifies in particular whether the opening was already cautious. If the cautious degree of commitment is constant, the check remains silent as to any defect; the modal constancy report nevertheless names the exclusion applied, where the proposition was taken up by the check.

Not to be confused with consistent restraint is not a mild retraction. It is an exclusion: where it applies, there is no defect to carry forward to 7.a. A cautious closing formulation therefore does not erase an earlier strong assertion — but neither does it create one where that force was never posited.

See also modal constancy · closing reservation · acknowledged revision · scope partition · Rule 14

Established meaning a correlation is a statistical or empirical association between two variables or phenomena; it does not by itself establish a causal relation.

In ARGUS the protocol uses correlation within the fidelity test against a primary source. Where a source establishes only a correlation and the text turns it into a more general or more certain claim, the check may qualify that restatement as strengthened. The term therefore keeps its established meaning; its ARGUS role is to supply an example of the degree or reach a source can lose when it is reformulated too strongly.

Not to be confused with ARGUS does not create an autonomous “correlation versus causation” test here. The point is narrower: to check whether the text preserves the meaning, scope and degree of certainty of its source. A correlation correctly presented as such is not a flaw.

See also primary source · fidelity test · strengthened · causal chain · empirical soundness

Established meaning in philosophy and linguistics, a counterfactual is a conditional bearing on a hypothetical possibility — for instance “if A had been the case, C would have been the case”. Despite the name, its antecedent is not necessarily false in every technical use; the term is therefore not confined to scenarios contrary to what actually happened.

In ARGUS the counterfactual is the minimal hypothetical version of the text in which an identified flaw is corrected, in order to measure its severity. In the substitution test of 7.a the analyst removes an unjustified claim, restores an omitted piece of information, replaces a distorted claim with its correct version, or repairs a logical leap without adding any argument foreign to the text, and then asks whether the central conclusion can still stand with the same nature of claim. For a quantitative flaw, where the correct value is established the minimal correction is its replacement; where it is not, the quantitative proposition is removed rather than invented. For an established modal retraction, the forcing bears on the claim as announced by the strong pole, not on a charitably weakened version of the thesis.

Not to be confused with the ARGUS counterfactual is neither a free rewriting of the text nor the analyst’s preferred model. It is strictly bounded to the minimal correction of the flaw found. Adding new arguments or fabricating a plausible value would distort the severity test.

See also substitution test · fatal · major · minor · arithmetical control

Common meaning a transparent descriptive expression — divergences bear on coverage where two treatments do not take in the same objects or aspects. ARGUS makes it a precise category of the differential.

In ARGUS coverage divergences are findings raised by one analysis but not mentioned by the other during a cross-audit. They must be treated separately from divergences of position. For each finding absent from one side, the analysis that did not raise it must state whether it rejects it after examination, did not examine it, or considers it beyond what the text permits. Until that answer is given, a finding not mentioned is neither conceded nor contested, and the adjudication rule does not apply to it.

Not to be confused with the absence of a finding in one analysis is not automatically a disagreement. A divergence of position presupposes that the two analyses pronounce on the same point; a coverage divergence signals that only one of them treated it.

See also differential · cross-audit · adjudication rule · divergence of position

Common meaning a transparent but unlexicalised composition — a contract evokes a set of reciprocal expectations, and credibility the fact of being taken as worthy of belief. ARGUS fixes a technical sense of its own.

In ARGUS the credibility contract is the set of codes a given audience expects before holding a text credible. It is formulated in 0.f after the audiences have been identified. At that stage the qualification of the audience and its credibility codes is a working hypothesis, not a conclusion. At the end of Step 3 the analyst must state explicitly whether the examination confirms, qualifies or invalidates that working hypothesis; if Step 3 supplies no element capable of modifying it, this is flagged as a possible limitation rather than treated as automatic confirmation. The protocol cites in particular expertise, quotations, figures, historical references, apparent moderation, irony, moral indignation, direct testimony, simplicity, complexity, technicality and neutrality of tone. The credibility contract then serves as a criterion of selection and interpretation: in 3.B, the fidelity test bears on the decisive claims on which the thesis of the text or its credibility contract depends.

Not to be confused with it does not measure the actual credibility of the text. It describes what the audience expects, not what the text is worth. Satisfying those codes is not proof: ARGUS expressly distinguishes a credibility device from substantiation.

See also reading contract · strategic audience · substantiation

Established meaning deference is the respect shown to someone in consideration of their rank or authority. In public law, judicial deference designates the degree of restraint a court adopts when reviewing the assessment of another decision-maker or body, and in that field it is a term of art with a substantial literature.

In ARGUS deference is the risk that the analyst demands less proof of a claim because its source appears prestigious or authoritative. At Step 8, point (d), the analyst checks whether the status of a source — a recognised institution, a canonical author, a peer-reviewed journal, an established news agency — has lowered their evidentiary requirement on a specific claim and, if so, which. Where a case is identified, the claim concerned must be named; where none is, the explicit statement that no case was identified is required — a general acknowledgement does not satisfy the point.

Not to be confused with deference is not a flaw of the analysed text but a possible flaw of the analysis, and it does not carry forward to 7.a. Nor is it courtesy towards a source: what is at stake is the level of proof actually demanded on an identified claim. And the ARGUS sense is not the judicial one — no court, no reviewing body, and no standard of review is involved.

See also symmetry of verification · standard of proof · Step 8

Common meaning definition test and boundary test may be used descriptively in various fields. Their union into a pair controlling decisive terms, and their conditions of activation, are proper to ARGUS.

In ARGUS the two tests determine the activation of Appendix 2 on the decisive terms of the central thesis. The definition test is positive where the text supplies neither a definition, nor a criterion of application, nor an enumeration allowing one to know what the term covers. The boundary test compares two delimitations of the same term, one restrictive and one extensive, both compatible with the use the text makes of it; it is positive if the conclusion holds under one of the two delimitations and falls under the other. A convention proper to ARGUS: a test is said to be positive when it reveals the defect sought, not when the text satisfies the requirement.

Not to be confused with absence of a definition is not enough to activate Appendix 2 if the boundary of the term does not carry the conclusion. Conversely, a defined term may still trigger the appendix if its delimitation changes the argumentative outcome.

See also empty signifier · empty, full, mixed · substitution test · Appendix 2

Common meaning a descriptive composition — degree measures an intensity, and interpretive closure a closing of interpretation. The technical scale from 0 to 4 and its definition are proper to ARGUS.

In ARGUS the degree of interpretive closure measures the freedom the device leaves the reader to examine the conclusion, to oppose to it the elements present or accessible, and to modify it if contradiction so requires. That is what the scale of 7.c measures, from Level 0 to Level 4. It measures neither the respectability of the author, nor their institutional standing, nor how militant the subject is. The move between levels is decided by functional thresholds, not by mechanical addition of flaws.

Not to be confused with the degree of closure is not a general quality score for the text. A text may be technically precise, stylistically elaborate or issued by a reputable institution and still present a high degree of closure; conversely, militant or frontal writing is not enough to establish a high level. The descriptors of rhetorical mode and of origin or function are orthogonal to the level.

See also Levels 0 to 4 · orthogonal descriptor · rhetorical mode · adapted propaganda · locked propaganda

Common meaning a thing made or adapted for a particular purpose; by extension, a plan or contrivance used to obtain an effect, and, in the study of texts, a technique used for effect.

In ARGUS a device is an organised set of elements whose arrangement produces an identifiable effect in the text. The word has four uses. The opening device — title, first sentences, first paragraph, introductory anecdote, opening question — is never neutral: it imposes a framing, delimits what can be thought or said, and positions the author as an authority; its critical examination is a priority, before entering the argumentation at all, and Step 6 returns to it after the analysis. The device of the text designates the text taken as an arrangement — framing, omissions, presuppositions — in the epistemic symmetry test of 3.H. The credibility device, in 3.K, names rigour concentrated on the verifiable and absent from the decisive, producing an effect of scholarship broader than the actual sourcing. The device of self-legitimation is what an opening declaration becomes, at Step 5, when the body of the text belies it.

Not to be confused with the protocol also uses device in a second and weaker sense — the persuasive devices of 3.K, the particular means of persuasion a text adapts to the expectations of its audience. Those are what a device holds together; they are not themselves the arrangement. The four uses above are arrangements, and only they are examined as such. A credibility device, in particular, is not substantiation: it is precisely what can stand in for substantiation without constituting it.

See also opening device · credibility contract · substantiation · epistemic symmetry · adapted propaganda

Common meaning as a noun, a quantified difference between two magnitudes and, in mechanics, the gear that distributes torque between two wheels.

In ARGUS the differential is the written document produced by the cross-audit at Level 2 of contradiction. The first analysis receives the second and produces not a commentary but a structured comparison: points of convergence, step by step; points of divergence with the position of each analysis and the nature of the disagreement — fact, qualification, severity level, or interpretation; the cases where one analysis concludes and the other declares insufficient elements; coverage divergences; and finally what would settle each divergence, and at what cost. The user then reads that single document rather than two complete analyses, and that is what makes Level 2 practicable.

Not to be confused with the cross-audit is the procedure; the differential is its product. It is neither a commentary on the second analysis by the first, nor anything borrowed from the mechanical sense of the word.

See also cross-audit · coverage divergences · adjudication rule

Common meaning depending on the field, the phrase could denote a trial bearing on a direction or on a steering mechanism. It has no univocal common sense; ARGUS uses it as the proper name of an argumentative test.

In ARGUS the direction test is the two-question test that determines whether a variation in degree of commitment constitutes a directional modal retraction. First question: is the strong degree the one under which the proposition does the argumentative work — as a premise, reinforcing the thesis, or charging an opponent? Second question: does the weak degree — suspended or hypothetical, or a retreat to “probable” from the assertoric — appear where maintaining the strong degree would carry a cost: a section on divergences, a concession, a closing reservation, a reply to an objection? Two “yes” answers trigger the verdict, subject to the three exclusions.

Not to be confused with the direction test is not a count of occurrences, and it does not conclude from the chronological order of the formulations alone. It bears on the respective argumentative function of the two degrees, and on the place where the weakening occurs.

See also directional modal retraction · modal constancy · strong pole, retracted pole · consistent restraint · acknowledged revision · scope partition

Common meaning a technical locution built from existing terms — retraction, modal, directional. No ordinary sense corresponds to the ARGUS procedure; the protocol fixes its meaning.

In ARGUS a directional modal retraction is established where one and the same load-bearing proposition receives a stronger degree of commitment where that strength does the argumentative work, and a weaker degree where strong commitment would carry a cost. The verdict requires two positive answers to the direction test: the strong degree serves as a premise, reinforces the thesis, or charges an opponent; the weak degree appears in a place of retreat — a section on divergences, a concession, a closing reservation, a reply to an objection. The three exclusions of the check must then be set aside.

Not to be confused with a modal retraction is not a logical contradiction. Polarity stays constant: the text does not move from P to ¬P, it weakens its commitment to P. Nor is it an acknowledged revision, which effectively ceases to draw benefit from the strong degree.

See also modal constancy · direction test · strong pole, retracted pole · acknowledged revision · disguised retraction · substitution test

Common meaning descriptively possible in ordinary use — a retraction may be concealed or presented in another form. ARGUS fixes procedural criteria of its own.

In ARGUS a disguised retraction is the structure in which a text withdraws or weakens the force of a strong proposition where maintaining it would carry a cost, while continuing to draw the same argumentative benefit from the stronger formulation. One form arises where a text explicitly revises a strong proposition but keeps the conclusion that depended on its force; another, singled out by the protocol, operates through a frame shift. In the explicit-revision case, an assertion may be acknowledged as “too absolute” while the conclusion re-asserts the same force under a narrower, reworded or re-framed formulation — in particular where the recommendation or the conclusion that the strong assertion supported survives intact. The acknowledged-revision exclusion then does not apply, and the modal constancy check triggers.

Not to be confused with the term does not qualify a hidden intention of the author. ARGUS records a structure in the analysed object: withdrawal of the strong degree on one side, maintenance of its benefit in the conclusion on the other. The history of production, and the authorship of that gap, must not be inferred. Where the discontinuity is manifest in the object itself — between a caption or an image and the body, or in a stylistically discontinuous box — the analyst may add a purely descriptive note locating the contradiction, imputing no authorship and no intention. Near the boundary with acknowledged revision, the verdict is given with signalled confidence and calls for re-reading.

See also acknowledged revision · directional modal retraction · strong pole, retracted pole · frame shift · strategic intent

Common meaning a transparent metaphor — rigour may be said to be displaced where it bears somewhere other than where it would be expected. ARGUS fixes the technical conditions of that displacement.

In ARGUS displaced rigour designates documentary precision concentrated on the claims easiest to verify for the target audience identified in 0.f and absent from those that actually carry the thesis. In 3.K the analyst locates separately the claims accompanied by a precise reference and the unreferenced claims, then compares the two locations: the question is not only whether the text carries references, but where it places its evidentiary effort.

Not to be confused with a text is not in displaced rigour merely because some claims are better sourced than others. The defect requires that precision concentrate where it costs little or carries little of the conclusion, while documentary precision is absent from the claims that carry the thesis. It is a distribution of rigour, not a general lack of sources. Note also that displaced here is not the fidelity qualification of 3.B, which the protocol names shifted.

See also standard of proof · source audit · load-bearing claim · 3.K · degree of interpretive closure

Common meaning a counter-argument is an argument advanced against another; effective, formal and decorative are ordinary. The precise functional opposition between these categories is proper to ARGUS.

In ARGUS an effective counter-argument is an objection, an opposing source or a contrary element that forces the text to modify, limit or reconfigure its thesis. A formal counter-argument, also called decorative, is indeed present in the text but has no effect on its argumentative trajectory: it is absorbed, marginalised or rendered harmless. Rule 16 forbids inferring the balance of a text from the mere presence of a nuance, a concession or an opposing source; the analyst must examine the effect it actually produces on the conclusion.

Not to be confused with formal does not mean non-existent. The contrary element may be quoted accurately and even developed at length; it remains formal if the device prevents it from having an effect on the thesis. Conversely, an effective counter-argument does not require the text to abandon its conclusion: it remains effective if the text keeps its conclusion after genuinely limiting, reconfiguring or qualifying it. An effective counter-argument that constitutes the retracted pole of an established modal retraction falls under the prohibition on crediting in 7.b set out at strong pole, retracted pole.

See also opening concession, immunising concession · locked nuance · interpretive closure · degree of interpretive closure · strong pole, retracted pole · Rule 16

Origin a concept developed by Ernesto Laclau in the mid-1990s, notably in “Why Do Empty Signifiers Matter to Politics?”, collected in Emancipation(s) (1996). Its theoretical genealogy runs through structural and post-structural theories of the signifier; the literature relates Laclau’s empty signifier to the floating signifier formulated earlier by Claude Lévi-Strauss, while noting that Laclau distinguishes the two.

Established meaning in Laclau’s discourse theory, an empty signifier is a particular signifier whose particular content is effaced far enough for it to represent an absent totality or articulate a chain of heterogeneous demands. It is distinguished from the floating signifier, whose meaning remains disputed between competing fixations.

In ARGUS an empty signifier is a term whose rhetorical and political power comes from the absence of stable semantic content, which lets heterogeneous and even contradictory interpretations project themselves onto it. ARGUS is concerned above all with strategic empty signifiers, those whose vagueness fulfils an identifiable argumentative function: to rally different audiences, to avoid refutation, to produce an effect of self-evidence or urgency, or to disqualify an opponent. Appendix 2 does not activate merely because a term is vague: its boundary test must show that the conclusion varies according to the delimitation retained for the decisive term.

Not to be confused with an empty signifier is not simple imprecision of language. The protocol distinguishes the vague term with no strategic function from the term whose vagueness carries the argumentation. And no word is “empty” by nature: the status empty, full or mixed qualifies its use in the text examined.

See also empty, full, mixed · boundary test · falsifiability · framing · Appendix 2

Common meaning three ordinary adjectives. Empty and full are opposites; mixed denotes what partakes of both.

In ARGUS empty, full and mixed are the three values of the “status” field in the analysis grid for an empty signifier, in Appendix 2. Empty: absence of definition, interchangeable usage. Full: defined, stable, non-substitutable without loss of meaning. Mixed: empty in some places, full in others, which signals a potential tension. The status is read together with the third field of the grid, the function in the argument — to rally, to avoid refutation, and so on.

Not to be confused with the status qualifies the use of a term in this text, not the term in itself. No word is empty by nature, and empty is not a value judgment: an empty signifier may be used with no strategic intent.

See also empty signifier · perimeter of the sayable

Established meaning epistemic qualifies what concerns knowledge, its justification, its conditions, or the reasons one has for holding a proposition true.

In ARGUS epistemic appears chiefly in the expression epistemic symmetry. In 3.H, ARGUS asks whether a flaw the text attributes to an object — bias, opacity, irrationality, lack of legitimacy, need for external justification, or any other breach of a standard of rigour — can also be found in the speaker, in the group or instance they privilege, or in the device of the text itself; if so, the text must acknowledge it and draw the consequences, or validly justify the difference in treatment. The same test applies to interest-based explanations: where the text explains an opposing position by the interests, institutional position or funding of whoever defends it, it must apply the same reading to its own authorities. An unjustified asymmetry is flagged as an argumentative weakness, under the qualifications of performative contradiction, unwarranted epistemic privilege or reflexive blindness.

Not to be confused with epistemic does not mean merely “intellectual” or “scholarly”. Epistemic symmetry does not require different objects to be treated as identical; it requires the difference in treatment to be explicitly and validly justified.

See also epistemic symmetry · deference · source audit · performative contradiction · asymmetry

Common meaning release from an obligation that would otherwise apply.

In ARGUS an exemption never exempts from the analysis; it exempts only from the mandatory stop, in two cases the protocol keeps strictly apart. First case: the user has already fixed the scope in the request; the analysis adopts it without a new validation, states at the head “scope fixed by the user in the request; no additional validation required”, and treats it as validated for the purposes of 0.a bis. Second case: the user explicitly waives validation without fixing a scope; the analyst then sets the scope and states at the head “scope not validated by the user, set unilaterally by the analyst”. This second wording tells the reader that the most decisive decision in the analysis received no external control. The two wordings are mutually exclusive for one and the same scope decision. In both cases, if the scope retained is restricted, the analysis recalls that this may make some findings inaccessible and that widening remains possible.

Not to be confused with the two cases are not interchangeable. Confusing them would present as validated a decision that was not, or the reverse.

See also mandatory stop · scope · relevance

Common meaning a check carried out by someone or something outside what is being checked.

In ARGUS a verification is external when it goes outside the analysed object, even where the source was already provided; it does not become external merely by coming from the internet. An attachment supplied with the article is external to the text analysed even if it required no search; conversely, a section of the same declared object remains internal even when it lies outside the scope. Every external verification is explicitly notified as a move outside the strict scope of the text, tied to the declared object rather than another version, sourced, and accompanied by an acknowledgement that the original text did not contain the information added. The verification is counted only if the analysis supplies the source actually consulted, in a form that lets a third party find it again. Knowledge remembered by the model is not one, even when accurate. Used on its own, “number of external verifications” designates successful checks only. Verification becomes mandatory where an element is decisive to the demonstration and a simple check would establish it; it remains bounded to such elements and does not replace internal examination. Beyond about ten checks for one analysed object, the analysis flags that the exercise is tending towards fact-checking — a signal of proportionality, not an exemption from Rule 13.

Not to be confused with external qualifies the relation to the analysed object, not the technical channel of access.

See also analysed object · mode of access · check · provided source

Common meaning five ordinary adjectives. Their organisation into five of the seven qualifications of one and the same fidelity test, and the boundaries between them, are proper to ARGUS.

In ARGUS these five qualifications describe the relation between a decisive claim of the text and the primary source to which it relates, where the fidelity test is activated. Faithful: the meaning, scope and degree of certainty of the source are preserved. Strengthened: a hypothesis, reservation, correlation or limited finding becomes a more general or more certain claim. Weakened: the text attenuates or restricts what the source establishes. Shifted: the words or data are exact but attributed to a referent, comparison, period or scene the source did not address. Inaccurate: on a verifiable and decisive fact, the source says something else. These qualifications are not necessarily exclusive — a claim may be both strengthened and shifted — and where they cumulate the analysis names the principal qualification and the secondary ones. The severity of a flaw is not contained in the label: it is determined in 7.a by the substitution test.

Not to be confused with strengthened does not mean simply “more convincing”, and weakened does not mean “worse”. They describe a gap in force or reach relative to the source. Faithful is incompatible, on the same element, with any defect qualification; it records instead that the source was preserved. And an available primary source is not enough to open this grid: the lineage must first be established.

See also primary source · fidelity test · own interpretation signalled · not decidable · substitution test · correlation

Established meaning in logic and argumentation, a fallacy is a defective pattern of reasoning that may wear the appearance of a valid or persuasive one. Named fallacies form a large and only loosely standardised catalogue.

In ARGUS the protocol uses fallacy in its argumentative sense without giving a general definition of its own. In 3.I it expressly calls a fallacy the presentation of the absence of counter-argument in a selected sample as proof of validity. The activation checkpoint for Appendix 2, at the end of Step 2, also uses “fallacy” as an example of a term that may remain vague without becoming strategic: if two compatible delimitations of the term change no conclusion of the text, the boundary test is negative and the appendix does not activate on that ground alone.

Not to be confused with the mere use of a vague term is not by itself a fallacy nor a strategic empty signifier. In ARGUS the qualification bears on the argumentative operation actually observed and on its effects, not on the presence of a label. Nor is the ARGUS sense the looser ordinary one, in which a fallacy is any mistaken belief: what is at issue is a pattern of reasoning, not the falsity of a conclusion. A list of named fallacies applied without regard to the genre of the text does not, the protocol notes, provide the indexing that a standard of proof provides.

See also logical validity · empty signifier · boundary test · informational circularity

Origin the modern notion of falsifiability is associated with Karl Popper, who makes it a criterion of demarcation for the empirical or scientific character of a theory: a scientific proposition must be capable of conflicting with a possible observation. The expression unfalsifiability in use is a distinction proper to ARGUS.

Established meaning falsifiability, in the Popperian tradition, is the property of an empirical theory or proposition that excludes certain observable states of affairs and can therefore conflict with experience. It does not mean that an isolated observation mechanically entails the rejection of a theory in scientific practice.

In ARGUS falsifiability asks what observation the text would recognise as refuting; unfalsifiability in use appears where the thesis is refutable in its wording but the text absorbs observations of the opposite sign as confirmations too. The flaw is then not in the form of the statement but in the way it is applied: an observation and its contrary may both be treated as confirming. Step 3.F therefore requires three distinct operations: list the observations the text treats as confirming; ask whether an observation of the opposite sign is treated as confirming as well — if so, qualify unfalsifiability in use; and finally ask what observation the text would recognise as refuting, flagging the case where it names none.

Not to be confused with unfalsifiability in use is not the formal immunisation of a thesis stated in irrefutable terms, and the protocol keeps the two apart. Here the wording may be refutable; it is its use that neutralises any practical possibility of refutation. And absence of proof is not, by itself, a problem of falsifiability.

See also empty signifier · logical validity · empirical soundness · formal immunisation · Step 3.F

Common meaning a gate is what one passes through; figuratively, a compulsory passage that lets through or holds back.

In ARGUS the two gates are two successive filters of the test of absences: first establish that the information was accessible and that it is missing; only then examine the effect of that absence on the conclusion. The first gate bears on the accessibility of the information and the finding of its absence. Once the absence is established, it distributes the outcomes of point 3: strategic omission, mere point of weakness, non-attributable absence, or first-gate qualification not determinable as things stand, which concludes nothing. Non-attributable absence and first-gate qualification not determinable as things stand are not defects and do not carry forward to 7.a; the three cumulative conditions of the latter are stated in its own notice. Where both criteria of the first gate are met, strategic omission is established, and it is forbidden to reject that qualification on the ground that the absent element would not destroy the thesis — a consideration belonging to the second gate. The second gate bears on the effect of the absence on the conclusion and opens only for an absence already qualified as strategic; it alone allows an omission to be qualified as structural, which requires demonstrating that the absent element would directly contradict the conclusion or require a major overhaul of the causal mechanism defended.

Not to be confused with the two gates are not two degrees of severity but two distinct criteria applied in order. The second cannot be used to reopen the first. An absence with indeterminate result keeps its first-gate qualification and is merely barred from the second. Nor are they thresholds on one scale.

See also test of absences · strategic omission · structural omission · non-attributable absence · first-gate qualification not determinable as things stand · threshold

Common meaning a long procedural formula, transparent through its components but not lexicalised as a current expression. ARGUS builds it to name a strictly conditioned outcome of indeterminacy.

In ARGUS the first-gate qualification not determinable as things stand applies where the information is absent but the analysis cannot establish whether it existed or was readily accessible at the relevant date. It is not a defect attributed to the text and does not carry forward to 7.a. This outcome is permitted only under three cumulative conditions: a check reasonably necessary to decide the issue was attempted and did not permit a decision, its result being partial or unsuccessful within the meaning of 3.B, or the analysis declares and justifies that no check was accessible under its execution conditions, under the zero-verification regime provided by 3.B; the information that would be needed to decide is named; and the indeterminacy concerns what the analysis can establish, not a decision the analyst has simply left open. Where the element is decisive and a simple check is possible, the attempt is mandatory under Rule 13. An uninvestigated uncertainty is therefore not an admissible indeterminacy.

Not to be confused with this qualification is not a synonym for caution, doubt, or disagreement on the analyst’s part. It presupposes that the question was actually pursued to the limit of the accessible checks. It is also distinct from not decidable in the fidelity test against a primary source, which bears on what the available source permits to be established.

See also partial check · unsuccessful check · Rule 13 · when elements are insufficient · not decidable · object of accessibility

Common meaning an expression attested in several cognitive and discursive analyses for a change in the frame of interpretation. ARGUS specialises its use within the modal constancy check; it does not claim to have created the phrase.

In ARGUS a frame shift designates, within the modal constancy check, the case where a strong proposition is posited on one plane — teleology, tendency or intention, for instance — and then weakened on another — content, or present fact. The protocol treats this as a particularly useful form of disguised retraction and imposes the demonstrate-or-withdraw alternative. If the text claims that the two planes correspond to different referents or perimeters, that distinction must be demonstrated separately for the scope partition to come into play.

Not to be confused with changing vocabulary or level of analysis is not by itself a frame shift. The qualification becomes relevant where that displacement accompanies the weakening of one and the same proposition, and where the distinction invoked between the two planes is not demonstrated.

See also scope partition · disguised retraction · modal constancy · referent · teleology

Common meaning frontal is said of what comes head-on, without indirection; sophisticated, of what is refined, complex or highly developed.

In ARGUS frontal and sophisticated describe the route persuasion takes, not its intensity nor the quality of the writing. They are the two values of the rhetorical mode, one of the descriptors the analyst may add after the level in 7.c, only if established. Frontal: commitment, accusation or mobilisation immediately visible. Sophisticated: persuasion operating mainly through the credibility codes expected by the audience — technicality, expertise, moderation, documentary abundance, concessions or neutrality of tone. These descriptors do not change the level number and must never be used to infer it.

Not to be confused with sophisticated here does not mean “refined”. The protocol uses that ordinary sense elsewhere, when it recalls that propaganda aimed at a cultivated audience may be stylistically sophisticated. The descriptor states the route persuasion takes; nor does it impute intention, the inference of strategic intent being a separate step.

See also orthogonal descriptor · degree of interpretive closure · credibility contract

Established meaning what is implicit is not stated directly but can be understood or reconstructed from what is said, from the situation, or from the organisation of the discourse.

In ARGUS implicit qualifies what the text does not formulate directly but the analysis can reconstruct from its elements. In 0.a, an explicit or implicit thesis is one of the three central criteria of argumentative relevance; it does not by itself establish strong relevance, which requires at least one central criterion and at least one other criterion on the list. Step 1 also examines the author’s implicit positioning, and Step 4 may formulate a hypothesis about an implicit strategic function from textual indications, under the imposed progressivity: indications, hypothesis, degree of confidence.

Not to be confused with an implicit content is not automatically a hidden intention. A thesis or a positioning may be reconstructed in the text without any design of the author being established. Implicit does not necessarily mean presupposed either: the latter designates what the text takes as settled without examination. And the ARGUS sense is not the one implicit carries in phrases such as implicit trust, where it means absolute or unquestioning.

See also presupposition · central thesis · positioning · strategic intent · relevance

Established meaning an inference is the operation by which a conclusion is drawn from one or more premises. The expression inferential leap is used less stably, for a passage from a set of propositions to a conclusion, often where that passage is not fully spelled out; depending on the author it may be descriptive or critical. Outside ARGUS it therefore does not denote a flaw that is necessarily established.

In ARGUS inference is the chain of reasoning that carries the text from its premises to its conclusions; ARGUS looks for the places where that link is insufficiently justified. Under A. Logical Validity at Step 3, the analyst checks whether the conclusions follow from the premises, whether there are unjustified inferential leaps or shifts of meaning, and whether causal chains are demonstrated rather than merely posited.

Not to be confused with not every inference is a flaw. The flaw is the unjustified leap, where the text concludes more than its premises carry. And the inference of strategic intent at Step 4 is an operation of the analyst, subject to its own safeguards; it is not to be confused with the inferences made by the analysed text.

See also premise · logical validity · causal chain · inference of strategic intent

Established meaning the phrase has no single settled technical definition outside ARGUS comparable to that of circular reasoning. In logic and argumentation theory, reasoning is circular when what is to be established reappears among its own supports, whether directly or through a chain of propositions that provides no independent footing.

In ARGUS there is informational circularity when the sources supporting the central thesis belong to the same camp or the same ideological ecosystem, with no genuinely independent counter-argument. Test 3.I asks three questions: do all the sources supporting the central thesis belong to the same camp or ecosystem? Do sources capable of contradicting it exist and, if so, are they cited, summarised or refuted? Does the text present the absence of contradiction as proof of the validity of its thesis? The protocol calls that last operation a fallacy: the absence of counter-argument in a selected sample proves nothing.

Not to be confused with reasoning may be formally valid and its documentary base circular. Test 3.I does not judge logical validity, which belongs to A. Logical Validity at Step 3. One and the same homogeneous corpus may show up both as informational circularity and in the test of absences; it must then be flagged as a single cause with two manifestations, not as two independent flaws.

See also second-order circularity · source audit · test of absences

Origin the English phrase is attested well before its media-critical use; the protocol attributes it to Noam Chomsky, who gives it that sense — critical education about media and indoctrination — and records that its dissemination in French is due to Normand Baillargeon, Petit cours d’autodéfense intellectuelle (Lux, 2005). ARGUS states this lineage explicitly.

Established meaning intellectual self-defence is a pedagogical metaphor for acquiring the tools of critical thought needed to examine arguments, sources, figures and means of persuasion. It does not correspond to a single scientific method or to a discipline with a standardised definition.

In ARGUS intellectual self-defence names an intellectual lineage within which ARGUS places part of its learning function. The protocol does not turn the expression into a step or an autonomous test: it cites it to situate its project historically among earlier critical traditions, and then specifies what ARGUS adds to them — notably the standard of proof indexed to genre, the separation between the degree of closure and the descriptors, the distinction between formal and effective counter-argument, the mandatory section on the text’s qualities, and the self-critical examination of Step 8.

Not to be confused with intellectual self-defence is not the generic name of an ARGUS analysis nor a qualification applied to the text examined. It is a lineage of the protocol and an aim of its learning function: to increase the reader’s autonomy.

See also zetetic · learning function · Step 8 · standard of proof · lineage and distinctive contribution

Common meaning a transparent composition — a closing of the interpretive field. No settled technical locution of this exact form is recorded; closure is a term of art elsewhere, notably in the sociology of technology. ARGUS specialises the sense: an impression of complexity without any real opening of the conclusion.

In ARGUS interpretive closure is the effect by which a text gives the reader the feeling of having crossed complexity while the elements introduced never genuinely threaten the conclusion. In 3.K the check asks explicitly whether the complexity presented opens interpretive outcomes or is organised so as to lead back always to the same conclusion. In a composite text this test bears on the argumentative scope declared in 0.a bis, not on technical or didactic sections: a genuine technical difficulty serving no conclusion is not interpretive closure.

Not to be confused with closure does not mean simplicity, verbal dogmatism, or absence of nuance. A very complex, abundantly sourced and apparently pluralist text may produce closure if all the routes it opens lead back to the same conclusion without allowing it to be genuinely put at risk.

See also effective counter-argument · formal counter-argument · locked nuance · displaced rigour · degree of interpretive closure

Common meaning said of a wall or a member that carries a load rather than merely closing or filling a space. The ARGUS use rests on that image of a supporting function.

In ARGUS load-bearing covers two distinct applications of the same image. A figure is load-bearing whenever the text offers it in support of a central proposition; a proposition is load-bearing when its removal would make the thesis unintelligible or unproven. For the figure, the criterion is the evidentiary function the text assigns to it, not whether the datum is indispensable to the conclusion: a figure presented as evidence commits its verifiability, and the presence of at least one load-bearing figure activates Appendix 4 under the conditions of its §2. Load-bearing propositions are the only ones annotated by the modal map of Step 2.

Not to be confused with being load-bearing is not being conspicuous. A striking figure the text offers in support of no central proposition is not load-bearing; a discreet figure to which the text assigns an evidentiary function is. Proportionality is then applied in the weighting of 7.a, not in refusing activation.

See also arithmetical and statistical control · modal map · decisive term

Common meaning a transparent metaphor — a nuance may be said to be locked if it is held within a determined outcome. The technical category and its criteria are proper to ARGUS.

In ARGUS a locked nuance is a nuance integrated in such a way as to reinforce or protect the conclusion rather than open it. It lets the text appear cautious, complex or balanced without giving the qualifying element the capacity to alter the argumentative trajectory. The notion puts into effect the principle of rhetorical destination: the formal presence of a nuance must be distinguished from its actual function.

Not to be confused with not every nuance that leaves the conclusion intact is locked. It becomes so where its integration takes part in protecting the conclusion against the effect the objection, uncertainty or limit it introduces would normally have. Consistent and honest caution falls under another mechanism.

See also immunising concession · formal counter-argument · interpretive closure · consistent restraint · rhetorical destination

Common meaning a possible descriptive expression — propaganda may be called locked in the sense of closed or protected. In ARGUS it names precisely Level 4 of the scale.

In ARGUS locked propaganda is Level 4 of 7.c, reached where the protection of the conclusion becomes self-immunising or materially closed. The protocol cites as possible mechanisms structural omissions, decisive fabrication or manipulation, exclusion of counter-argument, or a device that turns every contrary observation into confirmation. At this level, internal contradiction can no longer modify the conclusion on its central point without substantial external input.

Not to be confused with Level 4 is not propaganda that is simply “stronger” in a rhetorical or emotional sense. It is distinguished from Level 3 by a functional threshold: at Level 3 the contradiction remains reconstructible but the device makes it markedly more difficult or absorbs it; at Level 4 its central effect is prevented by self-immunising or material closure.

See also degree of interpretive closure · Level 3 — Propaganda · unfalsifiability in use · structural omission · effective counter-argument

Common meaning a stop is an interruption or a cessation; mandatory means imposed by a rule or a circumstance. Outside ARGUS the phrase denotes a halt that must take place.

In ARGUS the mandatory stop is the point at which ARGUS interrupts the procedure before the full analysis, where relevance is not clearly strong, in order to fix the scope explicitly. It occurs after the relevance assessment when relevance is partial, weak or ambiguous — by degree as by composition — and no other step is begun. The stop is not a request for authorisation: it submits a proposal the user can correct, and it contains three elements and three only — the relevance qualification with its form and reason in one line, the proposed scope with sections named, and the regime under which excluded sections would be examined. The stop further reminds the user that there are three outcomes — validate, correct the scope, or abandon the analysis — and that a restricted scope may make some findings inaccessible while widening remains possible; these reminders are not additional elements of the proposal.

Not to be confused with the stop is not a refusal to analyse. A text of weak relevance may be analysed at the user’s express request, with the statement required at the head; that derogation is not a fourth outcome of the stop. The stop bears on the scope alone: soliciting anything else — the severity of findings, or what the user hopes to find — is a breach.

See also relevance · exemption · scope · regime applied to excluded sections

Common meaning a locution compatible with the vocabulary of discourse analysis, where a marker may denote a discursive indication and authority a source of legitimation. ARGUS fixes a functional list of its own at Step 1.

In ARGUS markers of authority are the formulations by which a text invokes a vague but purportedly indisputable instance — “the theoretical point”, “history”, “science”, “everyone knows that”, “any informed person” — so as to close the debate before opening it. The analyst records their function in the initial framing.

Not to be confused with not every reference to an identifiable authority is a marker of authority in the critical ARGUS sense. The notion aims at an appeal to an instance rendered indisputable by its very formulation, not at the normal use of an attributed source or expertise.

See also opening device · framing · source audit · deference

Common meaning a possible ordinary evaluative expression. ARGUS gives it a precise procedural status among the outcomes of the first gate.

In ARGUS an absence is qualified as a mere point of weakness where the information existed at the publication date but was not readily accessible. The text is then weaker without it, but the protocol forbids inferring strategic selection: the accessibility required by the first gate is not established.

Not to be confused with mere does not mean negligible, and weakness does not mean a mitigated strategic omission. The qualification answers a different configuration: the information existed, but the conditions for imputing its absence as strategic selection are not met.

See also strategic omission · non-attributable absence · object of accessibility · first gate, second gate

Common meaning a transparent composition — constancy is the property of holding steady, and modal refers to modality. The phrase has no settled technical sense in ordinary usage corresponding to the ARGUS check.

In ARGUS modal constancy is the check that compares, for one and the same load-bearing proposition, the force with which the text commits to it at different points, where its polarity and its referent remain constant. From the modal map drawn up at Step 2, Step 6 takes up only those load-bearing propositions whose degree of commitment varies. The check looks for a directional weakening: it does not treat a change of polarity as a retraction — that is a contradiction — though it may flag its directional structure; and caution that has remained constant is not a defect.

Not to be confused with modal constancy does not require a text to speak everywhere with the same intensity. A variation is a defect only if the direction test establishes a directional modal retraction and none of the three exclusions — consistent restraint, acknowledged revision, scope partition — applies. The reporting nevertheless requires one modal-constancy line for each load-bearing proposition whose degree varies, with the verdict, or the exclusion applied and which one.

See also modal map · degree of commitment · directional modal retraction · direction test · consistent restraint · acknowledged revision · scope partition

Common meaning strengthening means making stronger or firmer; modal refers to modality. The combination has no univocal common sense corresponding to the ARGUS notion.

In ARGUS modal strengthening is the reverse movement to the retraction checked at Step 6: a load-bearing proposition cautious at the outset is posited more strongly later, for instance to the point of being asserted in the conclusion. This movement is explicitly outside the scope of the modal constancy check. Its sourced form is captured by the fidelity test of 3.B under the finding strengthened; its quantified form falls to Test E of Appendix 4. The remaining form — the author strengthening their own unquantified assertion without new support — falls to the examination of validity as unwarranted escalation.

Not to be confused with modal strengthening is not, by definition, at fault. A text may legitimately reinforce a proposition if the body supplies the new substantiation that justifies that rise in commitment. The defect appears where the reinforcement exceeds what the premises permit.

See also unwarranted escalation · modal constancy · modal map · logical validity · the finding strengthened

Established meaning in logic and linguistics, modality concerns the way a propositional content is presented — for example under the categories of the possible and the necessary, or according to the speaker’s degree of commitment. Typological work commonly uses semantic maps of modality to represent relations between modal values; modal map is also attested in that literature, although it is not the standard label. The ARGUS modal map is a distinct protocol-specific device.

In ARGUS modality designates the degree of commitment with which the text posits a load-bearing proposition at a given point. At Step 2 the modal map records, for each load-bearing proposition, two distinct pieces of information: its polarity, P or ¬P, and its degree of commitment — assertoric, probable, hypothetical or suspended. The annotation is local and neutral: it describes how the text posits the proposition at that point, without yet judging constancy. Where the same proposition, on the same referent and at constant polarity, appears under different degrees, each occurrence is noted with its location — opening, body, concession or closing reservation — and taken up at Step 6 by the modal constancy check.

Not to be confused with modality is neither the content of the proposition, nor its truth, nor its polarity. Two occurrences may have the same content and the same polarity and still differ in degree of commitment. The modal map does not tag every sentence: it is bounded to load-bearing propositions.

See also assertoric · probable, hypothetical, suspended · polarity · referent · load-bearing · modal constancy

Common meaning the way in which something is reached.

In ARGUS the mode of access states how the source used for a check arrived in the analysis, independently of the result of that check. It is one of the two axes of 3.B and takes two values only: provided source, for an attachment or a document already present; external search, for a source that had to be sought out. The mode of access is declared because neither the cost nor the reproducibility for a third party is comparable between the two.

Not to be confused with the mode of access is not the status of the check. The two axes are independent, and must not be presented as a single list of categories at the same level: a provided source that establishes none of what was sought yields an unsuccessful check.

See also check · successful, partial, unsuccessful · provided source, primary source · external verification

Common meaning a transparent juridical or argumentative composition — non-attributable means that responsibility for the absence cannot be laid at the door of the agent considered. ARGUS makes it a precisely bounded outcome of the test of absences.

In ARGUS an absence is non-attributable where the missing information was not available at the publication date of the text, notably because it was published later. This is not a defect: it does not carry forward to 7.a. It is nevertheless flagged in one line because it bounds what the analysis can establish, without turning later information into a retroactive requirement addressed to the text.

Not to be confused with a later element may illuminate the analysis, but it does not retroactively make the information accessible and therefore cannot ground a strategic omission. It may in some cases help to qualify the modality of a claim; that modal defect is distinct from the omission defect and must not be counted with it.

See also strategic omission · object of accessibility · modality · publication date · 7.a

Common meaning in ordinary English, a question is not decidable when it cannot be settled on the elements available. ARGUS keeps that idea but bounds its application strictly.

In ARGUS not decidable is the qualification of the fidelity test where the available source does not permit a conclusion on the relation between the text’s claim and that source. It is a limitation of the analysis, not a defect attributed to the text: it does not carry forward to 7.a and is incompatible, on the same element, both with faithful and with any defect qualification. The qualification therefore presupposes a source that is available but insufficient to conclude. If the primary source to which the text relates is unavailable, this qualification does not apply: the protocol requires declaring the unavailability, naming the claims concerned, and handling the limitation under the “When elements are insufficient” clause of Step 7.

Not to be confused with not decidable is not a general synonym for analyst’s doubt. In the fidelity test it means precisely that the available source does not permit a conclusion. Nor should it be confused with the first-gate qualification not determinable as things stand in the test of absences, which obeys conditions of its own concerning the accessibility of missing information.

See also fidelity test · primary source · when elements are insufficient · first-gate qualification not determinable as things stand · faithful

Common meaning a possible abstract composition — the object of a criterion is what that criterion bears on. ARGUS uses it as a technical safeguard specifying what accessibility bears on.

In ARGUS the object of accessibility is the missing information or proposition itself, and not the mere availability of a person, an institution or a source able to supply it. To conclude that information was readily accessible, the analysis must be able to name that information and the medium, statement, datum or fact that made it available at the publication date. A journalist’s presence at an event, or the possibility of putting a question to an actor, is not sufficient. Within a corpus, another text makes the information accessible only where it is established that it was already publicly available before the moment of publication of the analysed text, or that a distinct earlier source already made the information accessible.

Not to be confused with the accessibility of an interlocutor is not the accessibility of the information. Likewise, two texts published on the same day cannot be ordered by supposition: without a time or other proof of anteriority, one cannot serve to establish that the information was already accessible to the other.

See also strategic omission · non-attributable absence · first-gate qualification not determinable as things stand · corpus · publication date

Common meaning concession is an established rhetorical notion. The adjectives opening and immunising describe here two opposed functions; that functional typology is proper to ARGUS.

In ARGUS an opening concession introduces a limit, an objection or an uncertainty in such a way that a different, narrower or more uncertain conclusion genuinely becomes possible. An immunising concession on the contrary gives an appearance of balance while protecting the central conclusion from criticism. The protocol therefore qualifies a concession not by its formal presence but by its effect on the space of conclusions still open to the reader.

Not to be confused with a concession is not opening because it is long, explicit, or placed in a section of its own. And an immunising concession is not necessarily false: what makes it immunising is the function it fulfils in the argumentation. A concession that constitutes the retracted pole of an established modal retraction, the Step 6 exclusions being set aside, falls under the canonical rule set out at strong pole, retracted pole: the prohibition on crediting in 7.b covers all three routes — effective counter-argument, opening concession, and point of rigour — and the consistency check of Step 8 verifies that it is observed.

See also effective counter-argument · formal counter-argument · directional modal retraction · retracted pole · 7.b · Rule 16

Common meaning a current locution in technical fields for a mechanism permitting an opening. It can also be used descriptively in the analysis of texts. ARGUS gives it a precise discursive sense bearing on the opening segment of a text.

In ARGUS the opening device is the whole formed by the title, the first sentences, the first paragraph, an introductory anecdote or an opening question, in so far as they impose a framing before the argumentation is entered. ARGUS makes it an object of priority examination: the opening delimits what can be thought or said, positions the author as an authority, and may predetermine the argumentative trajectory. Step 6 returns to it after the analysis.

Not to be confused with the opening device is not merely the material introduction of the text. It designates the framing function that opening exercises. A descriptive first sentence is therefore not automatically a strong device; the analyst must examine what it renders thinkable, self-evident or excluded.

See also framing · presupposition · perimeter of the sayable · markers of authority · pre-emptive disqualification

Established meaning an order of magnitude locates a quantity approximately on a numerical scale, often by the nearest power of ten or by comparison at the same scale.

In ARGUS the order of magnitude is the test of quantitative plausibility that extrapolates the stated rate over the full period and relates the result to the reference population or set. It is Test A of Appendix 4. It applies first to hourly or daily rates extended over years, to periodic doubling, and to prolonged compound growth. A result exceeding the reference set establishes impossibility only where the unit counted is non-repeatable; if the set renews itself over the period, the bound must be raised from data in the text or from an admissible external verification. Failing an established bound the calculation is still performed, but impossibility is not asserted, and that silence does not amount to validating the figure. Before concluding, the analyst names the unit counted, the bound retained and the time window.

Not to be confused with the test does not authorise the analyst to substitute their own model for the author’s. ARGUS redoes the calculation with the data and assumptions available in the text or established by an admissible source. Test A cannot be declared unanswerable on grounds of cost or of the weight of the calculation, nor for mere uncertainty about the order of magnitude; it can be so declared only where a minimum necessary datum is missing, under the conditions Appendix 4 provides. And a total exceeding a population proves nothing where the event counted can recur for the same unit.

See also Appendix 4 · non-repeatable unit · reference set · arithmetical and statistical control · base rate

Established meaning in documentation, a descriptor is a word or group of words retained in a controlled vocabulary to represent a notion unambiguously and permit its indexing. In mathematics, orthogonal qualifies directions or objects whose inner product is nil; in statistics the term is used of contrasts or components that are uncorrelated within a defined frame. Orthogonality is not, in general, a synonym for statistical independence.

In ARGUS an orthogonal descriptor is information added to the level in 7.c but independent of it: it describes the level without being able to aggravate it, reduce it, or allow it to be inferred. ARGUS provides two, only if established: the rhetorical mode, frontal or sophisticated, and origin or function. The protocol names them descriptors orthogonal to the level, and that independence is what the word orthogonal records: descriptor and level are two distinct axes.

Not to be confused with a descriptor is not a criterion. A Level 3 accompanied by the descriptor sophisticated is not an aggravated Level 3. Orthogonal does not mean opposed or contrary, but without effect on the value of the other axis.

See also rhetorical mode · degree of interpretive closure · level in 7.c

Common meaning a transparent composition — an own interpretation is a reading attributed to whoever advances it; signalled means made explicit. ARGUS makes it one of the seven qualifications of its fidelity test.

In ARGUS own interpretation signalled qualifies the case where the text adds to a source a reading that is not the source’s but explicitly presents it as its own. The protocol does not treat this as a defect: the text does not lend its reading to the source, and it keeps visible the boundary between what the source establishes and what the text infers from it. It is one of the three qualifications that are not defects and do not carry forward to 7.a.

Not to be confused with signalling an interpretation as one’s own does not make that interpretation true, valid or well substantiated. The qualification records only that the text does not improperly attribute it to the source. Its argumentative soundness may still be examined under the other checks of the protocol.

See also fidelity test · faithful · primary source · inference · standard of proof

Origin the term belongs to J. L. Austin’s work on the distinction between constative and performative utterances, and is attested in his lectures from the mid-1950s.

Established meaning in the philosophy of language, an utterance is performative when uttering it performs, or takes part in performing, an act, rather than merely describing a state of affairs.

In ARGUS performative appears in two distinct constructions. In the performative contract, a standard of rigour, sourcing, method or transparency that the text prescribes for itself enters its reading contract and is applied back to it in addition to the genre norm; it must be declared explicitly after the standard of proof in 0.c. In performative contradiction, the text breaches, through its own discourse or its own device, a requirement it proclaims or applies to others — for instance when it claims to address “everyone” in a register that excludes part of that everyone, or when an unjustified epistemic asymmetry takes precisely that reflexive form. Other unjustified asymmetries may instead be qualified as unwarranted epistemic privilege or reflexive blindness.

Not to be confused with the performative contract and performative contradiction are two distinct uses. The first fixes an additional standard before examination; the second is a flaw found during the analysis. And the ARGUS sense is the Austinian one, not the current colloquial sense in which performative means done for show, or insincere: a performative utterance in this protocol is not an insincere one, and calling a text’s requirement performative says nothing about the author’s good faith.

See also performative contract · performative contradiction · register · epistemic symmetry

Common meaning an unlexicalised composition. Contract evokes an agreement or a set of expectations; performative has a technical sense of its own. ARGUS combines the two in a specific procedural category.

In ARGUS the performative contract is the standard the text prescribes for itself and which ARGUS applies back to it, in addition to the norm attached to its genre. It may be a requirement of rigour, sourcing, method or transparency, and it is declared in 0.c after the standard of proof. The performative contract, like the standard of proof, is restated at the beginning of Step 3. The clause applies in particular to manuals, protocols, codes of ethics, editorial guides, methodological articles and codes of conduct. A text that teaches its reader to demand sources is therefore examined against what it demands, and not only against what its genre permits. ARGUS submits itself to this in an explicitly bounded way: Rule 19 applies Tests 3.C and 3.G to the prose of the analysis, and the consistency check of Step 8 verifies among other things that the analysis does not contradict itself in executing its own requirements.

Not to be confused with the performative contract does not replace the genre norm; it is added to it and can only tighten it. Nor is it performative contradiction, which is a flaw found in the text rather than a standard declared before examination.

See also standard of proof · reading contract · performative

Common meaning a perimeter traces limits, and the sayable is what can be said — but the latter is not a transparent substantivation: the sayable and the boundaries of the sayable are marked locutions of discourse analysis and sociology, of Foucauldian descent, for what can and cannot be said in a given public space. ARGUS uses the phrase as a category for analysing the initial framing.

In ARGUS the perimeter of the sayable is the set of questions a framing renders possible, legitimate or debatable, as against those it renders unthinkable, trivial or morally suspect. Step 1 expressly asks the analyst to identify that boundary, so as to establish what the opening authorises or excludes before the arguments are examined at all.

Not to be confused with the perimeter of the sayable is not the scope declared in 0.a bis. The first describes a boundary produced by the text; the second is a methodological decision by the analyst about which sections are submitted to the protocol.

See also opening device · framing · scope · pre-emptive disqualification

Established meaning in linguistics and in the logical formalisms that oppose a proposition to its negation, polarity distinguishes the positive or affirmative pole from the negative one.

In ARGUS polarity states whether the text posits the target proposition as P or as its negation ¬P, independently of its degree of commitment. It is one of the two elements of the modal map, alongside the degree of commitment. The modal constancy check of Step 6 examines a variation of degree only where polarity and referent remain constant.

Not to be confused with a change of polarity is not a modal retraction. If the text moves from P to ¬P on the same referent at constant strong degree, that is a contradiction, reachable by the examination of validity; the modal check does not treat it as a retraction and can only flag its directional structure. And suspension is not a third polarity: “it is not known whether P” remains of polarity P, at the suspended degree.

See also modality · modal map · referent · probable, hypothetical, suspended · modal constancy

Common meaning a transparent descriptive expression — a disqualification is pre-emptive where it comes before the opposing position has actually been examined. ARGUS makes it a category of the opening.

In ARGUS a pre-emptive disqualification is the operation by which the text rules out an opposing position in advance, before it can be examined on its best arguments. At Step 1 the analyst checks whether the disqualified position is actually defended by someone identifiable, or whether it has been reconstructed in a weak or marginal form — which may lead to the straw man qualification.

Not to be confused with not every criticism placed in the opening is a pre-emptive disqualification. It must close off access to an opposing position in advance, or degrade it before its examination. The author’s conscious intention is not inferred from that finding alone.

See also straw man · principle of charity · opening device · strategic intent

Established meaning in logic and argumentation, a premise is a proposition from which a conclusion is drawn.

In ARGUS a premise is a link on which the text makes a conclusion rest; the examination of validity at Step 3, under A. Logical Validity, asks whether the conclusions follow from the premises. In the modal constancy check, a load-bearing proposition at the strong degree does “the argumentative work” in particular when it serves as a premise, reinforces the thesis, or charges an opponent.

Not to be confused with a premise is not true merely because it is used as a starting point, and a conclusion may be invalid even where its premises are accurate; ARGUS therefore separates the validity of the logical chain from the empirical soundness of the claims composing it. Note also that the plural premises, in the sense of a building or a site, is an unrelated word for ARGUS purposes and never appears in the protocol in that sense.

See also inference · logical validity · load-bearing · empirical soundness

Established meaning a presupposition is an element taken as given upstream of a piece of reasoning or of an utterance, without being itself explicitly established in that passage.

In ARGUS a presupposition is what the text takes as already settled, obvious or shared without submitting it to examination. Identifying it is the first check of Step 1 on the opening device: the analyst asks what the initial framing installs as going without saying, before the argumentation begins. Step 8 then applies the same requirement to the analysis itself, by asking what undemonstrated presupposition the analyst has admitted.

Not to be confused with a presupposition is not simply an implicit idea. The term here aims at a proposition the reasoning treats as settled; a content may be implicit without serving as a starting point held true. And recording a presupposition is not yet concluding that it is false.

See also implicit · opening device · framing · self-critical examination

Origin the expression is introduced by Neil L. Wilson in 1959 in “Substances without Substrata”. Quine takes up a related maxim in his account of radical translation; Davidson then develops the principle of charity as a condition of interpretation, optimising so far as possible the coherence, rationality and truth attributed to the speaker.

Established meaning in the philosophy of interpretation, the principle of charity requires interpreting what others say so as to preserve, so far as reasonably possible, its coherence, rationality and truth. In argumentation analysis the principle is commonly extended to the requirement that a thesis be reconstructed in the strongest version its formulations genuinely permit, rather than in a version easy to refute.

In ARGUS the principle of charity operates at Step 1 as a safeguard on the reconstruction of an opposing position. A position is qualified as a straw man where the text omits its best arguments and retains only its weakest or most marginal version. The finding bears on the reconstruction performed, not on the author’s design; any intention is a separate matter, for Step 4 and its three-level progressivity.

Not to be confused with applying the principle of charity means neither endorsing the opponent nor softening the criticism. It forbids weakening the position artificially before examining it. And charity here has nothing to do with generosity or almsgiving: it is a technical constraint on interpretation.

See also straw man · pre-emptive disqualification · strategic intent · Step 1

Common meaning probable is said of what is held to be likely; hypothetical, of what is posed as a hypothesis; suspended, of what is left pending or undecided.

In ARGUS probable, hypothetical and suspended are three of the four degrees of commitment ARGUS uses to record the force with which a text posits a load-bearing proposition at a given point. The fourth degree is assertoric, treated in a separate notice. At Step 2 the analyst notes, for each load-bearing proposition, its polarity — P or ¬P — and its degree of commitment as the text poses them at that point. Suspended means the judgment is left in suspense. Suspension concerns the degree of commitment to the target proposition, not its polarity: “it is not known whether P” remains of polarity P, at the suspended degree. If the same load-bearing proposition, on the same referent and at constant polarity, appears at several points under different degrees, each occurrence and its location are noted for the modal constancy check of Step 6.

Not to be confused with these terms describe the text’s commitment to a proposition. They are neither the analyst’s judgment on whether that proposition is true, nor polarities: suspension adds no third polarity between P and ¬P. And a concession is a location where a low degree of commitment may appear, not a degree in itself.

See also assertoric · polarity · modal map · concession · closing reservation · modal constancy

Common meaning a provided source is one that has been given to you. A primary source is a first-hand document, as against what comments on it.

In ARGUS provided source says how an item arrived in the analysis; primary source says what relation it bears to what the text is about. Neither implies the other. Provided source is one value of the mode of access in 3.B. Primary source designates the source to which the analysed text relates, and identifying it conditions the fidelity test. The primary, official, contemporary or directly relevant character of a supplied item is not sufficient to activate that test: the text must itself relate to the source — cite, name, summarise, comment on or criticise it — or the analysis must be able to establish that the claim under check takes up its content. A mere concordance of subject, date, wording or figure is not enough; failing that, the item remains an ordinary external verification.

Not to be confused with a source may be provided without being primary for the analysed text, and primary without having been provided by the user.

See also mode of access · fidelity test · external verification

Origin a notion developed by the semiotician Eliseo Verón in his early-1980s work on enunciation in the media. Its canonical reference is the 1985 article published by the IREP, on the analysis of the “reading contract” as a method for studying the positioning of press titles. The reading contract there describes the device by which a publication constructs its own place, that of its addressee, and the relation proposed between them.

Common meaning in ordinary English a contract evokes an agreement between parties. Reading contract is not an everyday phrase; it is an English rendering or calque of the French contrat de lecture and is used as a specialist term in some English-language media and semiotic scholarship on Verón, alongside related expressions such as mode of address.

In ARGUS the reading contract is the set of expectations of proof and method that the genre of the text creates in its reader. It is declared in 0.b with the dominant genre, and it governs the adjustment of proof requirements: an op-ed and a scientific article do not install the same expectations. Where the text itself prescribes a standard of rigour, sourcing, method or transparency, that standard enters its reading contract and is applied back to it, in addition to the genre norm.

Not to be confused with it is neither an agreement struck between author and reader nor an explicit promise by the author; it is what the genre installs, whether the author wills it or not.

See also standard of proof · credibility contract · performative contract

Common meaning the participle of recontextualise, to place or set in another context. Recontextualisation is in addition an established technical term — in the sociology of education after Bernstein, and in discourse studies for the displacement of a text or an utterance from one context to another. ARGUS gives the participle a narrower technical qualification.

In ARGUS a quotation is recontextualised when it is authentic but moved to a scene it did not address. In 3.B, for each decisive quotation, the analyst therefore does not check only the literal accuracy of the words: they also check that the work, the date, the page and above all the context of utterance correspond to what the analysed text makes them say. Recontextualisation is precisely a flaw that can survive literal verification, since the words quoted may be exact.

Not to be confused with recontextualised and shifted are close but not synonymous. Recontextualised qualifies specifically an authentic quotation moved to a scene it did not address. Shifted belongs to the fidelity test against a primary source and covers more broadly words or data that are exact but attributed to the wrong referent, the wrong comparison, the wrong period or another scene.

See also context of utterance · decisive quotation · shifted · primary source · external verification

Established meaning in linguistics and the philosophy of language, the referent is what a sign or expression points to — in reality, in a possible or imaginary world, or in the discursive context, according to the theory adopted.

In ARGUS the referent is what a given proposition actually bears on. The modal constancy check compares two occurrences only if they bear on the same referent and keep the same polarity; it then measures the variation in degree of commitment alone. Where the two occurrences genuinely bear on different referents or different perimeters, a scope partition may exclude the defect, but that difference must be named and demonstrated.

Not to be confused with the referent is not the scope of the analysis. The scope says which parts of the document are treated under ARGUS; the referent says what a proposition is about. And it is not enough for the text to assert that two formulations aim at different objects: that distinction must be established before the scope-partition exclusion comes into play.

See also polarity · scope partition · modal map · modal constancy · scope

Common meaning a regime is the set of rules under which something operates or is treated.

In ARGUS the regime applied to excluded sections is the standard of examination applied to the parts of the object not treated under ARGUS. It is declared with the scope in 0.a bis. The protocol gives four: technical accuracy, internal consistency, pedagogical validity, documentary accuracy. This regime is not ARGUS; it must be named so that the reader knows no part was silently discarded.

Not to be confused with a section outside the scope is not exempt from examination. It is examined under another standard and remains internal to the declared object.

See also scope · analysed object · mandatory stop

Common meaning in ordinary English the word denotes a book in which entries are recorded, and a vocal range. Its linguistic sense — the variety of language a speaker uses according to the situation, formal, ordinary or familiar — is an established technical one, and it is the sense ARGUS works from at 3.G.

In ARGUS ARGUS uses register to compare the language actually used with the audience the text claims to address; the same requirement applies to the analysis itself. In 3.G the register of the analysed text is confronted with its stated addressee: if the text claims to address “everyone”, does it speak a language genuinely accessible to everyone? The gap between the proclaimed “we” and the actual register is a performative contradiction and must be recorded. Under Rule 19 the analysis applies Tests 3.C and 3.G to its own prose. Two prohibitions follow: no comparative praise or blame about a class of texts without being able to establish it, and removal of loaded terms that are not necessary to the finding.

Not to be confused with register is neither style nor quality of writing. It is measured against accessibility for the addressee the text gives itself. The protocol also uses the word in a third sense, the registers of restitution in the corpus appendix.

See also readability · reading contract · performative contradiction

Common meaning the quality of bearing on the question at hand; appropriateness to the matter being considered.

In ARGUS relevance measures not the quality of the text but the degree to which its argumentative working falls within ARGUS. It is assessed in 0.a before any other step, and receives four qualifications: strong, partial by degree, partial by composition, weak or nil. A text is fully relevant if it meets at least one central criterion and at least one other criterion; the two corroborating criteria can never by themselves establish strong relevance. The criteria are assessed over the whole text submitted, not over its most argumentative part; where they are met only in identifiable parts, relevance is partial by composition, even if those parts meet all the criteria. It is partial by degree when one central criterion is met without a second, and weak or nil when no central criterion is met. Where two qualifications remain defensible for the same object, retain partial by composition.

Not to be confused with relevance does not say whether the text is interesting, nor whether it is right. A text of weak relevance is not a bad text: it is a text for which another kind of analysis is better suited.

See also mandatory stop · scope · exemption

Established meaning reproducibility covers several notions according to the field: a description precise enough to repeat a method; obtaining concordant results on a fresh execution; or the possibility for independent analysts to recover the same results from the same data. Representativeness designates, in survey statistics, the property of a sample whose mode of selection permits description of, or inference to, the reference population; it characterises the sample as a whole, not an isolated individual.

In ARGUS reproducibility and representativeness are two distinct requirements bearing on different operations. Reproducibility concerns checks and counts in particular. For the three counters of 3.B it bears on one and the same execution: the partition announced must remain stable, each status must be justified against the unit of the check, and the consistency check of Step 8 must recover the same numbers; a comparison between independent analyses bears on the propositions checked and their results, not on raw equality of counters. For a documentary count, Test G of Appendix 4 requires that the database queried, the query or selection criterion, and the unit of counting be stated; if any one is missing, the count establishes nothing verifiable. Representativeness arises chiefly in corpus analysis: before a cross-text synthesis the analyst asks whether the selection studied represents the phenomenon to be spoken of or only a particular editorial viewpoint, and whether a corpus built on another criterion could lead to a substantially different synthesis. If so, that difference must be reported as a limit of the synthesis, and not as a mere stylistic qualification.

Not to be confused with a reproducible operation may bear on an unrepresentative sample, and a corpus that looks representative may be described too vaguely for a third party to reproduce its selection or its count. The two notions answer different questions: can the operation be redone, and how far does this selection permit generalisation?

See also reproducibility of counting · Appendix 4 · corpus · second-order circularity · selection

Common meaning the extent of what an activity, an enquiry or a rule deals with; the range within which something applies.

In ARGUS the scope is the explicit decision fixing which parts of the object will be analysed under ARGUS and which are excluded from it. It is declared in 0.a bis as a nominative list of the sections treated and those excluded. The protocol holds the scope to be a decision, not a finding: treating a long text as a single argumentative whole is a decision in the same way as excluding a section, and both must be reasoned. The mention “no section excluded” is admissible only with its reason; no scope decision is taken by default or in silence.

Not to be confused with the scope does not describe what the text is about. It states what the analyst decided to examine, and it engages the analyst’s responsibility. A flaw found in a section outside the scope is not, in itself, an argumentative flaw of the thesis: it becomes one through Test 3.H only if the text itself prescribes the norm it breaks. Conversely, out-of-scope data internal to the object may establish a flaw within the scope if it bears directly on a claim made there and alters its truth, reach or representativeness. One further caution is internal to the word: the protocol uses scope for a second, unrelated notion in scope partition, the third exclusion of the modal constancy check, where it names the reach over which a proposition holds. The rule stating that exclusion uses perimeter for the present notion in its own body.

See also scope partition · regime of excluded sections · mandatory stop · composite texts

Common meaning partition denotes a division into parts, and scope the reach over which something holds. The phrase is descriptively intelligible, but no established technical sense corresponding to the ARGUS procedure is recorded here.

In ARGUS there is a scope partition when the two poles examined by the modal constancy check do not bear on the same referent or the same perimeter, provided that difference is demonstrated. It is the third exclusion of the check at Step 6. The analyst must name the referent of each pole explicitly. If the text claims to distinguish two referents — a surface and a substratum, an aim and an efficacy — that distinction must be demonstrated separately; asserted without demonstration, it does not bring the exclusion into play. In the aim-and-efficacy case, an aim held at a constant degree throughout while only the efficacy is left in suspense is silent by construction; the exclusion is therefore invoked only if the aim itself varies, and the analyst then verifies that the body has not already rendered the aim uncertain, which would forbid the reading “aim settled, only efficacy discussed”.

Not to be confused with an exclusion is not a mitigated defect. Where it applies, the check remains silent and nothing carries forward to the severity grid of 7.a. Note also that scope here does not carry the sense it has in 0.a bis, where it names the perimeter of the analysis; the rule stating this exclusion uses perimeter for that notion in its own body.

See also scope · consistent restraint · acknowledged revision · disguised retraction · referent · strong pole, retracted pole

Common meaning circularity and second order belong to several theoretical traditions, notably around self-reference and second-order cybernetics. The exact notion of a corpus that confirms itself is proper to Appendix 3.

In ARGUS second-order circularity is the risk that a synthesis drawn from a corpus inherits the representativeness limits of the initial selection. Before any thematic synthesis, ARGUS asks whether the corpus represents the phenomenon it is made to speak of or only a particular editorial viewpoint on it, and whether an alternative corpus built on another criterion or another editorial line would permit a substantially different synthesis from comparable facts. The test also asks whether the synthesis draws from the combination of texts a conclusion logically permitted by them — legitimate aggregation — or whether it adds an interpretive framework external to the corpus; those two registers must then be separated structurally in the reporting.

Not to be confused with second-order circularity is not informational circularity internal to a text. It sits at the level of the corpus and its synthesis: even individually well-conducted analyses may produce a synthesis whose reach is limited by a homogeneous or unrepresentative selection.

See also informational circularity · representativeness · corpus · cross-text synthesis

Common meaning no prior sense as a bias defined by ARGUS. Conformism denotes a tendency to conform; series attaches that tendency to a succession of objects. ARGUS makes it a precisely controlled reflexive bias.

In ARGUS series conformism is the risk that an analysis following other analyses aligns its judgment on the preceding ones rather than on the text it is examining. At Step 8 the analyst must answer whether such an alignment occurred and, if so, name the precise point; otherwise they state explicitly that no case was identified. A general acknowledgement does not satisfy the point.

Not to be confused with series conformism is not identical to anchoring. Anchoring covers more broadly the influence of an order of treatment or of prior information; series conformism aims specifically at the alignment of a judgment on previous analyses.

See also anchoring · self-critical examination · deference · corpus

Common meaning a counter-check is a check made to verify or contradict a first result; short indicates its extent. In ARGUS the phrase names the first level of the contradiction device.

In ARGUS the short counter-check consists in opening a fresh conversation, without supplying the first analysis, and asking the same AI to address only Steps 1, 3.E, 3.H and 7.a of the same text. The user then compares the four results. The protocol recommends it by default because it captures the essentials of analytical variability at a cost of a few minutes.

Not to be confused with a counter-check is not a revision of the first analysis. Its independence requires that the second execution not see the first. Supply it, and what is obtained is a revision or a commentary, not an independent counter-check.

See also cross-audit · differential · documented human adjudication · reliability of the analysis

Common meaning an audit is a methodical examination of an activity, an organisation, accounts or a system, in order to assess its regularity, quality, working or compliance. Source audit is a transparent composition: a methodical examination of the sources.

In ARGUS the source audit describes who the sources cited or mentioned by the text are, and what interests or positions may bear on what they say; it does not certify them. In 0.d a record is produced for each source, as a bulleted list and not a table, with name, role, institutional affiliation, country, declared position on the subject, and possible funding or links with stakeholders. Where the information is not available in the text, limited external verification is authorised, with citation and explicit notification; it becomes mandatory under Rule 13 when the element is decisive and a simple check would establish it. The operation closes on four questions: how diverse the sources are; whether sources contrary to the thesis are present and, if not, whether the text flags that limitation; whether the sources are sufficiently qualified for the reader to assess their reliability, biography, interests and bias included; and whether the text draws general conclusions from a restricted or partisan sample.

Not to be confused with the source audit does not sort sources into reliable and doubtful. It sets out the elements of position liable to bear on their contribution and leaves the consequence to the analysis. It bears on the sources of the text, not on those of the analysis.

See also primary source · external verification · informational circularity

Established meaning in law, the standard of proof is the degree of persuasion a party must reach for a fact to be held established — beyond reasonable doubt, balance of probabilities, clear and convincing evidence — and it varies with the kind of proceedings.

In ARGUS the standard of proof is the level of internal substantiation the analyst fixes, before examination, as sufficient for this text given its genre. It is written in 0.c before the critical examination. The declared standard is restated at the beginning of Step 3 before the systematic critical examination. The protocol’s example, for an op-ed, is a direct quotation, a reference to an identifiable authority, attributed figures, or reasoned internal coherence, with unattributed claims flagged as unsubstantiated. Once declared, the standard becomes binding: it is forbidden to relax it during the analysis in order to save the coherence of the text, and a fact central to the thesis that does not meet it is classed as unsubstantiated.

Not to be confused with the standard of proof is not the burden of proof. The second says whose task it is to prove; the first says to what level. It is not a general and abstract requirement of rigour either: it is indexed to the genre and holds for one given text. And the ARGUS standard is not the legal one — it is declared by the analyst for a text, not fixed by law for a class of proceedings, and nothing in ARGUS turns on the legal thresholds.

See also substantiation · reading contract · performative contract

Common meaning the four phrases are intelligible through their ordinary words. Their union into a four-audience typology, and the criteria distinguishing them, are proper to ARGUS.

In ARGUS ARGUS distinguishes four audiences that need not coincide. The stated audience is the one the text explicitly claims to target. The probable actual audience is the one the outlet, register, references, technicality, section, mode of dissemination and publication context make the likely addressee. The strategic audience is the one the text primarily seeks to convince, reinforce, mobilise, reassure or immunise against objections. The repelling audience is the audience or camp the text does not seek to convince, but to disqualify, ridicule or exclude from the perimeter of the sayable. The qualification produced in 0.f is a working hypothesis, not a conclusion: at the end of Step 3 the analyst must say explicitly whether the examination confirms, qualifies or refutes it, and where nothing in Step 3 permits it to be modified, flag that as a possible limitation rather than as automatic confirmation.

Not to be confused with these four categories are not necessarily four different audiences. One and the same group may occupy several positions. The repelling audience is the counter-intuitive one: it is not a failed persuasive target, but a group whose function in the device may be precisely not to be convinced.

See also principle of rhetorical destination · credibility contract · adapted propaganda · perimeter of the sayable

Common meaning a locution already attested in work on deception and information manipulation for an omission playing a strategic role. ARGUS gives it a specific procedural criterion based on accessibility and absence.

In ARGUS a strategic omission is established where the missing information was readily accessible at the publication date and it is effectively absent from the text, including in the form of a mention intended to dismiss it. The two criteria have the same polarity: two “yes” answers suffice to pass the first gate and to qualify the omission as strategic. In order of execution the finding of absence is treated first: if the text mentions the element, there is no absence to qualify and the question of accessibility no longer arises. It is forbidden to reject this qualification on the ground that the missing information would not destroy the thesis; that question belongs to the second gate.

Not to be confused with strategic does not mean that the author’s conscious intention is established. The qualification bears on the structure of selection observed: accessible and relevant information is in fact missing. The effect of that omission on the conclusion is a separate question, which determines whether it is structural.

See also first gate, second gate · structural omission · object of accessibility · strategic intent · test of absences

Origin the argumentative sense is native to English — an opponent, or an opposing argument, reconstructed so as to be easily refuted. The image is usually explained as that of a figure of straw set up in order to be knocked down, though dictionaries give no settled etymology for the argumentative use.

Established meaning in argumentation and informal logic, a straw man is a misrepresented version of an opposing position, weakened or distorted so that refuting it is easy and looks like refuting the position actually held.

In ARGUS an opposing position is a straw man when the text omits its strongest arguments and retains only its weakest or most marginal version, in violation of the principle of charity. The finding belongs to point 2 of Step 1, alongside pre-emptive disqualification. It bears on the reconstruction the text actually performs, not on the author’s intention.

Not to be confused with a straw man proposal — a first draft circulated precisely so that it can be attacked and improved — is a legitimate and often useful practice. The two share a name and nothing else: one is a fault of reconstruction, the other a method of work. And finding a straw man does not amount to saying the author deliberately distorted the opposing position; inferring intention belongs to Step 4 and to its own requirements.

See also principle of charity · pre-emptive disqualification · inference of strategic intent

Common meaning a pole denotes a point or a term of an opposition; strong and retracted are ordinary. The pair is not a current lexical opposition; ARGUS gives it a technical function of its own.

In ARGUS the strong pole is the occurrence where a load-bearing proposition carries the highest of the degrees compared and, in the positive case of the direction test, does the argumentative work. It is therefore not necessarily assertoric: it is the strongest of the degrees actually observed. The retracted pole is the weakened occurrence taking part in an established modal retraction. Where a concession constitutes that retracted pole in an established modal retraction, the Step 6 exclusions being set aside, it cannot be credited in 7.b as an effective counter-argument, an opening concession, or a point of rigour. The consistency check of Step 8 verifies that no quality credited in 7.b coincides with a retracted pole.

Not to be confused with the strong pole does not automatically designate the opening of the text, and the retracted pole does not designate every cautious formulation. The two terms take their sense only inside a comparison bearing on the same proposition, at constant polarity and referent, after the direction test has been applied.

See also direction test · directional modal retraction · concession · closing reservation · 7.b

Common meaning a transparent composition — an omission is structural when it bears on the structure of a whole rather than on a detail. The technical qualification and its high threshold are proper to ARGUS.

In ARGUS an absence is a structural omission only if the analyst explicitly demonstrates that the absent element, if integrated, would directly contradict the conclusion or require a major overhaul of the causal mechanism defended. This qualification belongs to the second gate and can be examined only after an absence has already been qualified as a strategic omission at the first gate.

Not to be confused with a strategic omission is not automatically structural. Accessibility and the finding of absence suffice for the first qualification; the second requires a further demonstration bearing on the effect of the missing information on the conclusion. Information that is merely useful, awkward or qualifying does not establish structural character. Nor does the word structural here describe the place of the omission in the text: it describes what integrating the missing element would do to the argument.

See also strategic omission · first gate, second gate · substitution test · severity · absence with indeterminate result

Common meaning to substantiate a claim is to give it support that makes it stand; the noun denotes what a claim rests on.

In ARGUS substantiation is what a claim rests on inside the text. In 0.c the analyst announces, before examination, what will count as sufficient internal substantiation for this genre of text. The protocol gives as examples a direct quotation, a reference to an identifiable authority, attributed figures, or reasoned internal coherence, and provides that unattributed claims be flagged as unsubstantiated. The criterion is therefore announced before being applied, and it depends on the genre.

Not to be confused with substantiation is internal to the text; external verification goes outside it. An unsubstantiated claim may be true: this field does not decide its truth. And substantiation is not a credibility device, which can concentrate rigour on the verifiable while leaving the decisive unsupported.

See also standard of proof · external verification · device · provided source

Common meaning a term already established in linguistics, grammar and education for various procedures in which one element is replaced by another. The ARGUS substitution test is a different procedure.

In ARGUS the substitution test measures the severity of a flaw by constructing the minimal version of the text in which that flaw is corrected, without adding any argument foreign to the text. The analyst removes an unjustified claim, restores omitted information, replaces a distorted claim with its correct version, or repairs a logical leap, and then asks whether the central conclusion can still be maintained with the same nature of claim. If it can no longer be, the flaw is fatal; if it survives with a markedly weakened claim, the flaw is major; if it is unchanged, the flaw is minor.

Not to be confused with the ARGUS substitution test is neither the grammatical test of substitutability nor the educational test of the same name. Nor does it authorise an optimal rewriting of the text by the analyst: the correction must be minimal and tied to the flaw found. Where a correct quantitative value is established, the minimal correction is its replacement; where it is not, the proposition is removed rather than replaced by an invented estimate.

See also counterfactual · fatal, major, minor · structural omission · directional modal retraction

Common meaning successful, which has achieved what it set out to do; partial, which covers only a part; unsuccessful, which has produced nothing.

In ARGUS successful, partial and unsuccessful are the three possible statuses of a check. Successful: the check establishes what was sought and satisfies the counting condition. Partial: it establishes a part only, the established and unestablished parts being named. Unsuccessful: it was conducted without establishing any of the elements sought. It is the result, not the mere production of a source, that decides the status. Every analysis declares three distinct numbers and states what each bears on; if the number of successful checks is nil, it declares and justifies that. These counters are an internal execution trace, not a metric for mechanically comparing two independent analyses.

Not to be confused with an unsuccessful check is not included in the number of successful checks. A source that is available but unfit to establish what was sought does not make the check partial or successful.

See also check · mode of access · external verification

Established meaning in philosophy, teleology is the study or doctrine of ends and finality. A teleological explanation accounts for a phenomenon by reference to an end, a function or a purpose attributed to it.

In ARGUS ARGUS uses teleology only within the modal constancy check, as an example of a plane on which a strong proposition may be posited before being weakened on another plane. The protocol has in view retraction disguised as a frame shift: a proposition asserted strongly on the plane of teleology, of tendency or of intention may then be weakened on the plane of content or of present fact. The term therefore names no autonomous ARGUS test.

Not to be confused with the presence of teleological reasoning is not, in itself, an ARGUS flaw. The flaw appears only where the change of plane serves to keep the benefit of a strong proposition while shedding its evidentiary cost, under the conditions of the modal constancy check.

See also modal constancy · modal retraction · frame shift · intention · tendency

Common meaning the stone at the foot of a door, the entrance to a place; figuratively, the value at which a phenomenon changes in nature.

In ARGUS a threshold is the limit between two levels of the scale of the degree of interpretive closure; ARGUS states that limit as a question, and it is never deduced from a cumulative number of flaws. Each move from one level of 7.c to the next answers a distinct threshold question. A hesitation between two levels is settled by that question, never by an average. In the reporting of 7.c the analyst must in addition give explicit reasons on the upper threshold: why the text crosses it, or why it does not. The word also serves, outside that scale, for the threshold of strong relevance in the applicability test and for the bound of about ten external checks in 3.B.

Not to be confused with a threshold is not a count, and crossing one is not accumulating. Nor is it a gate: the two gates of 3.E are successive filters applied in an imposed order, not two graduations of one scale.

See also degree of interpretive closure · level · first gate, second gate · relevance

Common meaning unaided means without help; a pass is one run through something. ARGUS turns that idea into a practice preceding the analysis.

In ARGUS the unaided pass is a practice recommended to the user before submitting a text to the AI. Without help, and in a few lines, the user answers three questions themselves: what presuppositions does the opening device install, what is missing (3.E), and does the text apply to itself what it demands of others (3.H)? They then compare their three answers with those of the analysis produced. The protocol presents this as the only part of the protocol that cannot be delegated, because its function is the reader’s learning: a user who reads the analysis directly obtains a result, while a user who has first answered for themselves acquires a skill.

Not to be confused with the unaided pass is neither a mandatory step of the AI’s execution of ARGUS nor a pre-analysis meant to influence its result. It belongs to the user’s learning, and it takes its sense precisely from being done before the analysis is seen.

See also learning function · opening device · test of absences · epistemic symmetry

Common meaning a transparent evaluative expression — escalation denotes an intensification, unwarranted that it is not justified. ARGUS specialises the image for an unsupported hardening in the conclusion.

In ARGUS unwarranted escalation is the flaw in which the author strengthens their own unquantified assertion without the body having provided the new support that would justify that reinforcement. ARGUS handles it under the examination of validity at Step 3 as an excess of the conclusive over the premises, reusing the modal map of Step 2.

Not to be confused with not every conclusion firmer than the opening is an unwarranted escalation. If the argumentation does supply sufficient new support, the strengthening may be legitimate. The qualification aims precisely at the absence of that support.

See also modal strengthening · premise · logical validity · modal map · standard of proof

Origin from the Greek zētētikós, “inquiring, examining”, from zēteîn, “to seek”. The word is old and rare in English, where it has been used of an attitude of sceptical inquiry and in mathematical contexts. ARGUS attaches part of Appendix 4 explicitly to that sceptical tradition.

Established meaning zetetic qualifies what proceeds by inquiry and examination. The word is rare in present-day English and carries no settled contemporary sense in critical thinking: the organised sceptical movement that calls itself zététique is French. The approach ARGUS invokes under that name is one of investigating and testing claims through evidence, verification and examination of the reasoning, particularly where claims are extraordinary or fragile.

In ARGUS zetetic is a term of lineage, not the name of a step. ARGUS places part of its arithmetical and statistical control in that tradition, in particular the order-of-magnitude and base-rate tests of Appendix 4. The protocol then distinguishes what it takes from that tradition from what it adds itself.

Not to be confused with ARGUS is not presented as a mere application of the zetetic approach, and the label exempts no operation of the protocol. It describes a partial methodological lineage, not a competing regime of analysis or a verdict on the text.

In English, zetetic also has a historical — and still lexically visible — association with Samuel Rowbotham’s Zetetic Astronomy and later flat-earth movements; that use is unrelated to the sceptical lineage invoked by ARGUS.

See also order of magnitude · base rate · Appendix 4 · intellectual self-defence · lineage and distinctive contribution