2.1. Eldad and Modad

In search of an exegetical tradition, Allison points to a group of texts that arguably draw upon the lost Jewish apocryphon Eldad and Modad.Footnote ¹⁴ First Clement 23.3 and 2 Clem 11.2 independently reproduce a quotation from an unnamed source (which they respectively call ἡ γραφή and ὁ προφητικὸς λόγος) that denounces those who question the eschatological fulfilment of divine prophecy: ‘Wretched (ταλαίπωροι) are the doubleminded (οἱ δίψυχοι), who doubt (οἱ διστάζοντες) in their hearts (τῇ καρδίᾳ, 1 Clem τὴν ψυχὴν).’ Hermas, who preserves the only explicitly identified line of Eldad and Modad (Vis. 2.3.4: ‘the Lord is near to those who return’), also attests a nexus of terms found in 1 and 2 Clem’s quotation (δίψυχ- κτλ., διστάζω, ταλαίπωρος).Footnote ¹⁵ The Letter of James also preserves a citation from an otherwise unknown ‘scripture’ (γραφή): ‘Does the spirit (πνεῦµα) that he [sc. God] made to dwell in us set its heart on envy (φθόνος)?’ (Jas 4.5).Footnote ¹⁶ The story about Eldad and Modad in Numbers 11 similarly ‘features jealously and a divinely bestowed spirit’.Footnote ¹⁷ The same key lexemes in Hermas, 1 Clem and 2 Clem also occur in the near vicinity of James’ unknown scripture (Jas 4.8–9: δίψυχοι… ταλαιπωρέω). The only other occurrence of δίψυχος in James (1.8) is placed in apposition to the term διακρίνοµαι (1.6): ‘the διακρινόµενος… is an ἀνὴρ δίψυχος’. Allison suggests that such correspondences present the start of a Christian discourse about doubt and doublemindedness. Δίψυχος and διστάζω are connected in Hermas,Footnote ¹⁸ 1 Clem,Footnote ¹⁹ the shared source behind 1 and 2 Clem, and James, only ‘where James has διακρίνοµαι, Hermas [and the rest have] διστάζω’.Footnote ²⁰ Thus, in Allison’s estimation, the vocabulary for ‘doublemindedness’ and ‘doubt’ derive from Eldad and Modad, and James is early evidence that διακρίνοµαι is indeed a synonym for διστάζω.

However, there are difficulties with the line of argumentation. Allison’s basis for identifying the quotation in 1 Clem 23.3/2 Clem 11.2 (the text that collocates δίψυχος and διστάζω) with Eldad and Modad is not particularly strong. He gives three positive reasons:Footnote ²¹ (1) The 1 and 2 Clem source is concerned with unfulfilled prophecy, with 2 Clem calling its source ὁ προφητικὸς λόγος. In Numbers 11, Eldad and Modad were prophets.Footnote ²² (2) The quotation in 1 and 2 Clem has an eschatological orientation. In later Rabbinic tradition, Eldad and Modad were ‘prophets of the last things’.Footnote ²³ (3) Hermas ‘almost certainly knew the source’ of 1 and 2 Clem, because of the way the authors similarly pair δίψυχος with διστάζω and ταλαίπωρος.Footnote ²⁴ Unfortunately, points (1) and (2) are so general that the 1 and 2 Clem source could well be attributed to any prophetic, eschatologically orientated writing. The proclivity to identify unknown quotations with ‘a passing reference to a lost work from the church fathers’ based on ‘loose thematic correspondences’ is surely to be avoided.Footnote ²⁵ As for (3), it may well be that Hermas betrays knowledge of the 1 and 2 Clem source, but this does not constitute evidence that that source is in fact Eldad and Modad, especially given that there is no shared vocabulary between the actual quotations.

The proposed connections between Eldad and Modad and the Letter of James are also somewhat tenuous. To my mind, at least two other options are more likely for identifying the unknown scripture of Jas 4.5: (1) either the text as we have it is corrupt and should be amended (πρὸς φθόνον → *πρὸς τὸν θεόν) to bring it closer to Psalm 41.1 LXX;Footnote ²⁶ or (2) the citation formula in verse 5 (ἡ γραφὴ λέγει) is pointing to the scriptural reference in verse 6 (Prov 3.34). Timothy Gabrielson has recently highlighted several examples of delayed citation, where a parenthetical comment is placed between a citation formula and the scriptural citation.Footnote ²⁷ One such example is found in the Epistle of Barnabas:

The scriptural citation is introduced (λέγει), followed by an ‘interpretative gloss’, then a resumptive citation formula (λέγει δὲ Κύριος).Footnote ²⁹ Gabrielson writes, ‘In a resumptive citation, one formula introduces the preparatory information, and the second the quotation proper… The second formula is added precisely because of the delay in quotation.’Footnote ³⁰ This is exactly what we find in James. The citation formula (ἡ γραφὴ λέγει) is followed by an additional preparatory comment (Jas 4.5b–6a);Footnote ³¹ a resumptive citation formula (διὸ λέγει) then directly precedes the scriptural quotation (4.6b, citing Prov 3.34). Both options (textual emendation and delayed citation) preclude the need to identify Jas 4.5 (or vocabu-lary in the surrounding context) as deriving from Eldad and Modad. In any case, Allison’s assumption that διστάζω and διακρίνοµαι are effectively synonyms has no lexicographical support, either in post-classical Greek in general, or in the texts he associates with Eldad and Modad in particular (except for his reading of James, which, however, is part of the explanandum, not the explanans). For this reason, Allison must concede: ‘As to why διακρίνοµαι took on the meaning “doubt”, the question is open.’Footnote ³²

3.1. Δίψυχος and Doubt in the Christian Reception of the Two Ways Tradition

The Jewish TW tradition may provide the earliest attestation of the δίψυχ- word group, seen in the overlapping material between the Didache, Barnabas and the Doctrina Apostolorum.Footnote ³³

Huub van de Sandt and David Flusser offer a possible reconstruction of the Jewish TW source:

Any proposed reconstruction must remain tentative, given the complications inherent to the TW’s textual transmission. Nevertheless, as John Kloppenborg writes, ‘the agreement between Barnabas and the Didache in their use of διψυχεῖν suggests that this detail in fact belongs to the earliest strata of the TW tradition’.Footnote ³⁵ The fact that Doctr. Apost. 4.4 has dubito, which commonly translated the δίψυχ- word group in fourth-century Latin tradition, would suggest the source did contain a Greek term for intrapersonal division (διψυχέω, as attested in Did 4.4 and Barn 19.5), rather than a Greek term for ‘doubt’ (such as διστάζω).

The Jewish TW source connects several key ideas, reflected in the vocabulary of these three later Christian writings (Doctr. Apost. 4.3–4; Did 4.3–4; Barn 19.4–5, 19.11b–12a): ideas about ‘judging’ (κρίνω), ‘partiality’ (πρόσωπον λαµβάνειν), ‘convicting’ (ἐλέγχω) and ‘transgression’ (παράπτωµα) all belong to a judicial domain, perhaps within the setting of the synagogue.Footnote ³⁶ The TW’s use of διψυχέω (Did 4.5; Barn 19.5) may have a similar sense to the terms δίγνωµων and δίγλωσσος, also attested in the TW source (Did 2.4; Barn 19.7; Doctr. Apost. 2.4: nec eris duplex in consilium dandum, ‘you should not be double in giving advice’) – less an issue of doubt than prevarication, failing to render a clear judgement. As Kloppenborg notes: ‘διψυχεῖν appears to connote equivocation and partiality when it comes to reproof, probably based on the fear of reproving one of higher social status’. In any case, it is clear that ‘TW’s usage [of διψυχέω] has nothing to do with ambivalence in prayer’.Footnote ³⁷

Despite the judicial setting of this particular TW admonition, later Christian writers would nevertheless transpose the admonition into a new context of prayer and petition, with the δίψυχ- language taking on more explicit connotations of doubt.Footnote ³⁸ The Apostolic Church Order specifies prayer as the context for TW’s warning against doublemindedness: ‘do not be double-minded in your prayer’ (ἐν προσευχῇ σου µὴ διψυχήσεις) (13.2),Footnote ³⁹ as does the Epitome: ἐν προσευχῇ σου µὴ διψυχήσῃς (10).Footnote ⁴⁰ So too the Coptic Apocalypse of Elijah:Footnote ⁴¹ ‘No one is able to enter the holy place if he is double-minded (ⲛ̅ϩⲏⲧ ⲥⲛⲁⲩ). The one who is double-minded (ⲛ̅ϩⲏⲧ ⲥⲛⲁⲩ) in his prayer (ⲡⲣⲟⲥⲉⲩⲭⲏ) is darkness to himself’ (Apoc. Elijah 1.25–7).Footnote ⁴² Pseudo-Ignatius’ letter to Hero further connects doublemindedness, prayer and doubt: ‘do not be double-minded in your prayers: for blessed is the one who does not doubt’ (µὴ γίνου δίψυχος ἐν προσευχῇ σου· µακάριος γὰρ ὁ µὴ διστάσας) (Ad Her. 7).Footnote ⁴³ The Apostolic Constitutions modifies the tradition to a similar end, placing the TW admonition next to Jesus’ words to Peter during the walking on water episode (Matt 14.31):

In Mandate 9, Hermas directly connects doubt, prayer and doublemindedness in a way not seen throughout the rest of Hermas.Footnote ⁴⁵

This developing discourse of prayer would come to inform the Gospel of the Lots of Mary, a recently discovered Coptic miniature codex dating to the fifth/sixth century ce.Footnote ⁴⁷ Despite its designation as a ‘gospel’ (ⲉⲩⲁⲅⲅⲉⲗⲓⲟⲛ) (G. Lots Mary, incipit), the text is primarily concerned with the practice of sortilege (divinatory lots). In Oracle 6, we find another example of the Christian appropriation of the TW admonition: ‘The Lord has heard your request… Only do not become careless, (saying) “this matter will not happen”. Yes, it will happen.’ The admonition is repeated in Oracle 24, with the language of doublemindedness: ‘Stop being of two minds, O human, whether this thing will happen or not. Yes, it will happen! Be brave and do not be of two minds.’ Doublemindedness (ϩⲏⲧ ⲥⲛⲁⲩ) is a major theme of this sortes text, as can be discerned from the incipit: ‘The one who will go forward with his whole heart will obtain what he seeks. Only do not be of two minds.’Footnote ⁴⁸ The admonition against doublemindedness is paired with similar injunctions against doubting (ⲇⲓⲥⲇⲁⲍⲉ/ⲧⲓⲥⲇⲁⲍⲉ): ‘At all events, do not doubt (ⲙⲟⲛⲟⲛ ⲙ̅ⲡⲣ̅ⲇⲓⲥⲇⲁⲍⲉ) because nothing is impossible for God’ (12); ‘Do not doubt in your heart (ⲙ̅ⲡⲣ̅ⲧⲓⲥⲇⲁⲍⲉ ϩⲙ̅ ⲡⲉⲕϩⲏⲧ) nor walk in the counsels of vain people’ (15). AnneMarie Luijendijk has noted that this is a significant development within the tradition, since earlier sortes made no appeal to ‘doubt’. By appropriating the discourse of prayer, with its attendant admonitions against doublemindedness and doubt, the author of this divinatory text is able to effectively ameliorate potential anxieties associated with the uncertainty of sortilege: ‘if you received an inappropriate answer, you must have doubted the lot.’Footnote ⁴⁹

If the TW admonition against doublemindedness (in its earliest recoverable form) concerned prevarication in judicial decisions (perhaps due to partiality towards those of higher status), in its Christian reception, we find the admonition transposed into a new context – prayer. It is this new context where the associations between ‘doublemindedness’ and ‘doubt’ are developed more fully. This discourse would eventually undergo another transposition into entirely different contexts; in the case of the Gospel of the Lots of Mary, anxieties surrounding the practice of sortilege are addressed with ‘the theology of steadfastness in prayer’.Footnote ⁵⁰ Although one discourse among others,Footnote ⁵¹ the Christian discourse about prayer helped to forge semantic associations between the concepts of doubt (διστάζω) and doublemindedness (δίψυχος κτλ.) in a way that would prove important (as we will see) for the semantic associations of the term διακρίνοµαι.

3.2. Διακρίνοµαι and Δίψυχος in the Letter of James

Like several of the Christian texts cited above, the Letter of James also references doublemindedness in the context of prayer. However, James does not associate the δίψυχος person with ‘doubt’ (διστάζω), but ‘division’ (διακρίνοµαι):

The idea of ‘being divided’ fits well with the pattern of community division repeated throughout the Letter. While James shows little concern for the inner disquiet of the doubter, who struggles to trust that God will grant their request, he does show repeated concern for his audience’s divided attitudes and behaviour towards God and others within the community: he writes to admonish those who ‘think they are pious’ but cannot ‘bridle their tongues’ (1.27); who ‘hold faith in Jesus’ while ‘showing partiality’ to the rich (2.1); who affirm the Shema but do not demonstrate their faith with works (2.18–19); who ‘bless our Lord and Father’ while ‘cursing those made in God’s image’ (3.9–10); who assume friendship with God while seeking to be ‘a friend of the world’ (4.4); who wait for ‘the parousia of the Lord’ while ‘grumbling against one another’ (5.7–9). When James advises the one who lacks wisdom to ask ἐν πίστει µηδὲν διακρινόµενος, he is not criticising the quality of their πίστις towards God, but the incongruity between their πίστις and actions and attitudes towards other members within the community. Thus parity between faith and works emerges as a major theme within the Letter (1.22–5, 2.14–26).

By rejecting the ‘special NT meaning’ of διακρίνοµαι, we see that the πίστις-διακρίνοµαι collocation does not present an antithesis between ‘faith and doubt’, but rather an antithesis between ‘loyalty and division’. This reading also highlights James’ connection between the use of διακρίνοµαι in 1.6 (‘ask with loyalty without being divided’) and 2.4 (‘have you not made divisions amongst yourselves?’). It similarly highlights James’ connection between how to rightly petition God in 1.5–8 (‘with loyalty, without being divided’) and the picture of community division and asking God wrongly in 4.1–9.Footnote ⁵² This shows that James’ concerns about ‘division’ are deeply connected to issues of community strife (4.2), especially the issue of partiality towards the rich and the dishonouring of the poor (2.1, 4, 6). In this respect, James reflects similar ethical concerns to the TW source.Footnote ⁵³

James is the first known author to collocate διακρινόµενος (1.6) with δίψυχος (1.8). While references to ‘doubt’ are absent in Jas 1.5–8, in the reception of James, patristic authors will nonetheless begin to associate διακρίνοµαι with the concept of ‘doubt’.

(1) Origen (third century) explicates a reference to διακρίνοµαι in Jer 15.10 LXX (‘A man condemned and at variance (διακρινόµενον)’) with the key vocabulary of Jas 1.6–8:

Those who ‘doubt’ (ἀµφιβάλλω) Jesus are δίψυχοι, which Origen collocates with the verb διακρίνοµαι. Here, διακρίνοµαι and ἀµφιβάλλω are not synonyms (ἀπιστοῦσι/ἀµφιβάλλουσι are juxtaposed with διακρίνοµαι/καταδικάζω), but they do both express actions associated with the δίψυχοι.

(2) Nilus of Sinai (fourth century) paraphrases Jas 1.6–8: ὁ διακρινόµενος ἐν ταῖς προσευχαῖς καὶ διστάζων, ἀνήρ ἐστι δίψυχος, ‘the one who is divided in [his] prayers and doubts is a double-minded man’ (Epist. 3.167). The collocation of διακρίνοµαι and διστάζω may suggest a synonymous relation; at the very least, Nilus clearly associates the two.

(3) Cyril of Scythopolis (sixth century) writes that one who has faith can do all things if they preserve in faith ‘unwaveringly (ἀδίστακτος), not yielding to unbelief (ἀπιστία), but always renewing and strengthening himself (ἐνδυναµόω) in faith, µὴ διστάζων ἢ διακρινόµενος. For the διακρινόµενος is like a wave of the sea, blown and tossed by the wind, and so forth’ (Vita Euthymii 5.15). Cyril references Rom 4.20 (Abraham ‘strengthened (ἐνδυναµόω) himself in faith’) and Jas 1.6, again collocating διστάζω and διακρίνοµαι.Footnote ⁵⁴

(4) Pseudo-Andrew (eighth century), in his Catena to James, quotes an extract from Hermas, Mand. 9 (quoted above) in a scholion to Jas 1.8, promoting a reading of James in light of Hermas’ discourse on δίψυχια, διστάζω and prayer.Footnote ⁵⁵

(5) Theodore the Studite (eighth/ninth century), after quoting Jas 1.5–8, writes:

(6) In Theophylact’s (eleventh century) exposition on the Letter of James, the commentator glosses διακρινόµενος with a synonym (διαστέλλω) and a reference to the TW tradition, substituting διψυχέω (‘to be of two minds’) for ἐνδοιάζω (‘to doubt’):

The patristic reception of Jas 1.6–8 shows a clear correlation of the terms διακρίνοµαι and δίψυχος with the concept of doubt (διστάζω, ἀµφιβάλλω, ἐνδοιάζω). While none of these texts indisputably present διακρίνοµαι as a synonym for ‘doubt’, they clearly show that διακρίνοµαι (through its association with δίψυχος) is understood as belonging to a discourse in which ‘doubt’ is a central concept.Footnote ⁵⁶

3.3. Διακρίνοµαι and Δίψυχος in Coptic Translation

The discursive connections between διακρίνοµαι, δίψυχος and the concept of ‘doubt’ become even more intertwined in the Coptic translation traditions. Coptic translators use ϩⲏⲧ ⲥⲛⲁⲩ (‘two hearts’)Footnote ⁵⁷ for δίψυχος and δίψυχοι in Jas 1.8 and 4.8 (Sa),Footnote ⁵⁸ for δίψυχοι in 1 Clem 23.3 (Ach)Footnote ⁵⁹ and for δίψυχε in Hermas Mand. 12.4.2 (Sa).Footnote ⁶⁰ ϩⲏⲧ ⲥⲛⲁⲩ also renders Sir 1.28 (LXX: ἐν καρδίᾳ δισσῇ), itself a possible reference to the Semitic idiom of the ‘double-heart’ (לבוב/שתי לבבות) that may stand behind the Greek term δίψυχος.Footnote ⁶¹ The Apocalypse of Elijah and the Gospel of the Lots of Mary both likely go back to Greek Vorlagen, and it seems reasonable to posit that their use of ϩⲏⲧ ⲥⲛⲁⲩ is based on the δίψυχ- word group.Footnote ⁶²

However, beyond its attestation as a rendering of a Greek loan word for ‘doublemindedness’, ϩⲏⲧ ⲥⲛⲁⲩ took on the sense ‘to doubt’.Footnote ⁶³ The Pistis Sophia describes how the soul of the prophet Elijah descended into the body of John the Baptist, yet the disciples ‘doubted (ⲣ̅ϩⲏⲧ ⲥⲛⲁⲩ) now at that time’, not recognising that John was in fact Elijah (1.7).Footnote ⁶⁴ In the Gospel of Mary, the disciples face the prospect of martyrdom with trepidation:

In Apocryphon of John, John begins to feel uncertainty about the saviour when confronted by a pharisee; as he questions, a child appears to him in a vision, asking, ‘John, why do you doubt (ⲛ̅ϩⲏⲧ ⲥⲛⲁⲩ)? Why are you afraid’ (21.15, BG 8502.2 (BCNH 35:66)).Footnote ⁶⁶ The child assures John, ‘I have come to teach you. Understand my lessons; Share them with any others who have received the spirit.’Footnote ⁶⁷ In the Middle Egyptian translation of Matt 14.31, διστάζω is rendered with the same idiom:

It is this connection between ‘doublemindedness’ and ‘doubt’ that results in the use of ⲣϩⲏⲧ ⲥⲛⲁⲩ in Coptic to translate διακρίνοµαι in Mark 11.23 (Sa); Matt 21.21 (Sa, M); Rom 4.20 (Sa); and Did 11.7 (Fa).

The clear semantic associations between ‘doubt’ and ‘doublemindedness’ in the Coptic rendering of διακρίνοµαι help us understand the Latin translations of διακρίνοµαι as dubito and haesito in Mark 11.23, Matt 21.21, Rom 4.20, Acts 10.20 and Jas 1.6.Footnote ⁷⁴ The oldest Latin version of the Shepherd of Hermas (Vulgata) likewise translates δίψυχος κτλ., διστάζω and διακρίνοµαι with dubitatio and dubito,Footnote ⁷⁵ while the Doctrina Apostolorum renders διψυχέω with dubito.Footnote ⁷⁶ In all these instances, we find translations that reflect an exegetical tradition that associates the terms διακρίνοµαι and δίψυχος with the concept of ‘doubt’. This association was made possible by the Christian transposition of the TW admonition against doublemindedness into the new context of prayer, as well as the influential pairing of the two Greek words in the Letter of James. While the reception of these texts had a significant influence on the Christian discourse of prayer, the Christian idiom had little effect on the broader use of διακρίνοµαι in post-classical Greek. It is only after Christianity had achieved cultural dominance that the sense of διακρίνοµαι as ‘doubt’ found its way into the major lexicons of the Byzantine period.Footnote ⁷⁷

4.1. The Distributional Hypothesis

In the mid-1950s, the Structuralist linguist Zellig Harris (1909–1992) made an important observation about the relationship between word meaning and word context (or distribution):

Harris’ insight (reformulated positively in terms of ‘similarity’) would come to be known as the ‘Distribution Hypothesis’: ‘Similarity of meaning correlates with similarity of distribution.’Footnote ⁷⁹ Harris was thinking of paradigmatic relations (words one might typically find in a thesaurus: synonyms, hyponyms, etc.), but this could also be applied to syntagmatic relations (words that co-occur within a sentence, including co-hyponyms), as John Firth’s famous dictum articulates: ‘You shall know a word by the company it keeps.’Footnote ⁸⁰ This means that word meaning can be explicated by analysing patterns of distribution; or in other words: ‘people learn how to use words by observing how words are used’.Footnote ⁸¹

The Distributional Hypothesis necessitated a radically different approach to the study of semantics, one grounded in the empirical analysis of natural language. If word meaning is correlated with word distribution within a corpus, then it would also be important to consider what went into one’s corpus (since an unrepresentative corpus could skew distributional patterns). But it also meant that different kinds of corpora could be analysed; if distributional patterns differ between linguistic corpora, this would open up new ways to study sublanguage discourse,Footnote ⁸² such as the difference between a lexeme’s distribution in post-classical Greek (its global distribution) versus the writings of Greek-speaking Christians (its distribution within a sublanguage).

4.2. Encyclopaedic Information

Structuralists like Harris understood semantic relations (mapped through distributional analysis) to ‘exhaust word meaning,’Footnote ⁸³ to the exclusion of encyclopaedic information. This exclusion is traditional: ever since Frege’s (1848–1925) distinction between ‘sense’ and ‘reference’,Footnote ⁸⁴ it has been common to discriminate between a word’s ‘dictionary meaning’ (its ‘sense’ or ‘denotation’) and ‘encyclopaedic meaning’ (cultural knowledge related to the word, but not part of its formal definition). The distinction seems intuitive; I do not need to know anything about the Apollo missions to correctly understand the sense of the word spacecraft. Thus, cultural knowledge of the Apollo missions can be characterised as encyclopaedic – non-essential to the ‘dictionary definition’ of the word. Yet despite its intuitiveness, the dictionary-encyclopaedia distinction is very difficult to maintain.Footnote ⁸⁵ When does something stop being inherent to a word’s definition and become ancillary, ‘encyclopaedic’ information?

Cognitive linguist Ronald Langacker famously illustrated this with the word ‘banana’. What features should belong to its ‘dictionary’ definition? Information about its shape, colour, taste, smell? What about ‘more abstract domains of knowledge, that bananas are eaten, that they grow on trees, that they come from tropical areas’, that slipping on a banana skin is a classic schtick in comedy routines?Footnote ⁸⁶ As it turns out, all such information about bananas is encyclopaedic – what matters for semantics is the saliency of the information for the linguistic community. That ‘bananas are eaten’ is likely a more salient feature of the lexeme ‘banana’ than their function in slapstick comedy. But there might exist linguistic communities that make use of the fruit for neither food nor comedy, in which case a whole different set of features will likely be more semantically salient. From this, it is important to grasp two things: (1) all semantic specifications are encyclopaedic; (2) the saliency of encyclopaedic information is relative to the use of the lexeme in linguistic communities (sublanguages).

4.3. A Cognitive Linguistic Explanation: Distributional Shift for Διακρίνοµαι in the Sublanguage

How does this account of semantics help explain how διακρίνοµαι became ‘doubt’?

Firstly: from the Distributional Hypothesis, we learn that a word is known ‘by the company it keeps’. In the Jewish TW tradition, the word διψυχέω κτλ. kept company with words from a judicial context (κρίνω, πρόσωπον λαµβάνειν, ἐλέγχω, παράπτωµα). In the tradition’s Christian reception, δίψυχος would keep new company with a set of words relating to prayer and doubt. This new distributional pattern for δίψυχος in the Christian sublanguage resulted in doubt becoming a more salient encyclopaedic association for the lexeme.

Secondly: if ‘people learn how to use words by observing how words are used’, then the meaning of διακρίνοµαι is shaped by its collocation with δίψυχος (a collocation first attested in James, subsequently influencing patristic reception). Readers within this sublanguage were able to infer a new semantic relation between the two lexemes where one had not previously existed. At first, this may have been a loose co-hyponymy, ‘division’ and ‘double-minded’ being related terms to the superordinate category of doubt. Yet the reception and translation of διακρίνοµαι shows that at some point, readers began to infer a synonymous relation between the two terms. This explains why we find Greek authors collocating διακρίνοµαι and διστάζω (e.g., Nilus of Sinai), Latin authors rendering διακρίνοµαι as dubito and haesito, and Coptic authors translating διακρίνοµαι as ⲣϩⲏⲧ ⲥⲛⲁⲩ (‘to be of two minds’ = to doubt).

Finally: this corpus-based approach to semantics is attentive to the diachronic idiosyncrasies of sublanguage discourse. The distributional patterns of διακρίνοµαι within the shifting contexts of early Christian discourse influence the saliency of the various encyclopaedic specifications associated with the lexeme. For some Christians, the concept of doubt was a particularly salient feature. For the wider linguistic environment (post-classical Greek), it was not, and other concepts remained more salient (divide, distinguish, judge, etc.). It was only once the Christian sublanguage achieved cultural dominance that we find the new sense of διακρίνοµαι attested more broadly (such as in the Byzantine lexicons).

So then, from a cognitive linguistic perspective, the historical data for διακρίνοµαι can be accounted for in terms of a distributional shift in the sublanguage.