From Latin coniunctiōn-em, "joining together" (Oxford English Dictionary)
As a term for the part-of-speech category including and, or, but, etc., conjunction is attested in English since the 14th century. It is a borrowing of the Latin term coniūnctiō, which had been used for the part-of-speech category since ancient times.
An interesting question is how this term came to be used by logicians and logically-oriented semanticists just for and, rather than for the whole part-of-speech category. This sense of conjunction was popularized by Whitehead and Russell (1910). It seems unlikely that they were the first to introduce it, however, since they make no comment to that effect, and simply use the term in that sense without any apparent concern that it will be misunderstood: "Any combination of given elementary propositions by means of negation, disjunction or conjunction...will be elementary" (p. 96). But earlier usage, including by logicians such as George Boole and many others, overwhelmingly favored the more traditional interpretation of conjunction as a broad category including or, if, etc. as well as and.
Perhaps Whitehead and Russell's usage may be seen as building on a point made by grammarians at least back to Priscian (ca. 500). In Book XVI of his Institutiones Grammaticae, Priscian subdivides the class of conjunctions into various subcategories, such as "copulative," "disjunctive," "causal," etc., and introduces the class of disjunctives as follows: Disiunctivae sunt, quae, quamvis dictiones coniungunt, sensum tamen disiunctum... ("Disjunctive [conjunctions] are those which, although they conjoin words, nonetheless disjoin the sense...") The implication is that other conjunctions do "conjoin the sense." It could be that Whitehead and Russell chose to use "conjunction" for just the logical symbol corresponding to and because they regarded their notation as representing propositions — the meanings or senses of sentences — and it had traditionally been recognized that words like and do conjoin meanings, and words like or, if, etc., do not.
Priscian's explanation was repeated by generations of later grammarians, often to explain the terminological oddity of positing a class of "disjunctive conjunctions."
Here is 17th-century grammarian Gerardus Joannes Vossius:
Potestas seu significatio conjunctionum variat. Aliae sunt copulativae: ut, 'et', 'que', 'ac'... Aliae sunt disjunctivae: ut, 'vel', 'aut'. Mirifica autem videtur appellatio, ut conjunctio sit disjungens. Sed conjunguntur voces materialiter; disjunguntur formaliter. Nam ita conjunguntur sententiae, ut iis res disjungi significatur.
I am not very proficient in Latin, so I would welcome any corrections, but I think this can be rendered basically as follows:
The power or meaning of the conjunction varies. Some are copulative: as 'et', 'que', 'ac'... Some are disjunctive: as 'vel', 'aut'. The terminology seems strange, that a conjunction would disjoin. But the words are conjoined materially, disjoined formally. Thus thoughts/sentences are conjoined, in order to signify that a thing is to be disjoined/separated.
18th-century grammarian James Harris says something very similar, but perhaps more clearly:
And now we come to the Disjunctive Conjunctions, a Species of Words, which bear this contradictory Name, because while they disjoin the Sense, they conjoin the Sentences.
In fact, this distinction forms the primary basis for Harris' typology of conjunctions, where the main division is into those that connect meanings and those that do not.
References
- Boole, George (1854) An Investigation of the Laws of Thought. Walton & Maberly.
- Harris, James (1751) Hermes: Or, a Philosophical Inquiry Concerning Language and Universal Grammar. H. Woodfall.
- Priscianus Caesariensis (ca. 500) Institutiones Grammaticae.
- Whitehead, Alfred North and Bertrand Russell (1910) Principia Mathematica, vol 1. Cambridge University Press.