ἐκ and ἀπό with temporal expressions in the papyri

One question asked at the Greek Prepositions Workshop at Tyndale House, Cambridge in 2017 was about the use of ἀπό and ἐκ in temporal expressions beyond the corpus we used. The source of our data was the Cascadia Syntax Graphs available in Logos Bible Software for the New Testament (SBLGNT) and the Deutero-Canon (Swete). So the corpus is Greek and it’s certainly grammatical language data (to the extent that the concept of grammaticality is a useful ones), but there are questions about how frequency of particular phenomena exist in beyond the corpus. Much of the Deuterocanon is translation, which, while still Greek, exists at a different degree of naturalness. The grammar does not change, but the frequency patterns can. Rare usages or expressions might appear more commonly in translation texts than they do in the natural language. In turn, if we have texts that desire to emulate historical important translations or texts (such as the KJV in English), those otherwise rather uncommon language phenomena might receive more attention than otherwise. It’s possible/probable that this happens in the New Testament, but the degree to which that is true is debatable. Typically that sort of emulation of revered or respected texts in later writing/speech is done a the conscious level of language use, but the majority of grammatical distinctions in a language exist at the subconscious level. As a result, emulation of “LXX Greek” is more likely to involve larger chunks of text, particular prominent lexical choices, idiomatic expressions, and so forth.

Still, the question is an important one even on more subtle usage such as the topic at hand here. The fact of the matter is that there is a great deal we simply do not know about grammar and language variation in the papyri as compared to the more literary texts of the Koine period and also the New Testament. The existing grammars of the papyri deal with phonology and morphology. They say very little about actual syntactic structure.

Now, the question at hand involves the preference for ἀπό for times expressions denoting an arbitrary point in time that initiates a time span. Since both our prepositions profile temporal expressions via SOURCE, the landmark of the preposition is always the initial point in time. Time spans, then, can be expressed in a couple ways. They can express an initiated temporal starting point without reference to any sort of concluding point if time.

  1. From that day, he only ate chicken for lunch.

They can additional refer to well-defined spans of time with an explicit beginning and end.

  1. From January to March, all she did was write chapter 2 of her thesis.

Our data for the NT and Deuterocanon say that both of these sentences would use ἀπό in the temporal prepositional phrase.

Conversely, time expressions that refer to temporal event quantification invariably use ἐκ. These would be sentences like:

  1. Mary sat down to dinner for a second time, having been interrupted by the phone.
  2. After your third semester of Greek, come talk to me about an independent study.

Again, our data shows consistent usage of ἐκ in such contexts and never ἀπό.

These contrasts are interesting because Luraghi (2003) states that ἐκ was the preferred and dominant preposition for all types of temporal expressions in Homer. She states that this changes starting with Herodotus, but she does not distinguish between the two types of temporal expressions above. We do not know when the division of labor we find in the New Testament and Deuterocanon began or how natural it is. Eventually, by the time of Medieval Greek, ἀπό is simply the preferred preposition across the board according to Bortone (2010).

Does what we see in our data appear in the papyri? This is a very useful question, but it is difficult to assess efficiently and quickly in a comprehensive manner, since we do not (currently) have papyri treebanks. Nevertheless, we can still dip our toes in, using the Duke Database of Documentary Papyri (hence forth: DDoDP). These are a few fixed expressions we can use for proximity searches in the papyri. For this little examination here, I choose ἀπό/ἐκ+ἡμέρας (from+day) and then ἀπό/ἐκ δευτέρου (from+second). There are others we could use. For event quantification, any count quantifier works well. For time spans, ἀπό/ἐκ+νῦν would also work or ἀπό/ἐκ+ὥρας or even ἀπό/ἐκ+ἀρχῆς. That last one falls into the domain of idiomatic expressions, however, for translation-like language and continues to be quite uncommon.

A search for ἀπό followed by ἡμέρας produces ~100 accurate hits, including the following from P. Flor. 1 1 (= Chrest. Mitt. 243), dated to CE 153:

  1. κατὰ μῆνα ἕκαστον ἀπὸ τῆς ἐνεστώσης ἡμέρας ἐπὶ μῆνας εἰκοσιτέσσαρας
    according to each month from the initial day until the 24th of the month

From a cursory survey of those ~100, ἀπό seems quite consistent in the papyri in aligning itself with the usage in the New Testament and Deuterocanon data.

What about ἐκ? Like the NT & DC, ἐκ has limited distribution with time span temporal expressions, but it’s quite uncommon. Searching for ἐκ+ἡμέρας, I find a single result in DDoDP: P. Oxy. 49 3507. It’s circa CE308-312:

  1. γυμνασιαρχ[ία]ς … ἃς ἐκ περισσοῦ ἡμέρας ἠναγκάσθη ποιῆσα[ι π]έρυσι
    the position of gymnasium ruler, which from the remaining days he was compelled to perform last year.

The two options for explaining the linguistic choice here would be either (1) a desire to communicate in a more formal register or (2) that the use of περισσοῦ here is a sufficient condition to evoke the container schema, which ἐκ then profiles. The context implies the latter—the remaining days are a bounded period of time in and of itself without the need for an explicit temporal goal expression, but the letter itself appears to be relatively formal, so both could be involved. Either way, that we can find over 100 instances of ἀπό compared to this single example of ἐκ is instructive for the preferences of language users.

When we turn to the other type the situation is a little different because event quantification is simply the less common form of temporal expression generally. Searching for ἐκ δευτέρου in the DDoDP, we find six clear instances. From CE ~285, we have the following example (P. Oxy. 12 1410):

  1. ὅστι̣[ς] / δὲ ἐκ δευτέρου ἀνεδέ(δοτο) / [α]ὖ̣θ̣ι̣σ̣ δεκαπρώτ[είᾳ -ca. ?-]
    “Anyone who has been reappointed financial officer for a second time…”

Doing a similar search for ἀπό+δευτέρου and we do get positive search results, but none of them are event quantification temporal expressions. Instead, we find that they are all still temporal span expressions just as we would anticipate with ἀπό, such as the following (P. Oxy. 51 3613):

  1. ἀπὸ τοῦ δευτέρου ἔτους τῆς Αὐρηλιοῦ τοῦ ἐν θεοῖς βασιλείας
    from the second year of Aurelius the divine king

There’s no event quantification, here. Instead, Aurelius’ second year is simply the initial starting point for the span of time, i.e. “form the second year of Aurelius [onward].

So the data patterns we find for temporal expressions with ἐκ and ἀπό in the New Testament and Deuterocanon continue to hold in the papyri―at least in the limited set of constructions we examined here. A more comprehensive study of these two preposition in the papyri is certainly needed.