A search engine for Ancient Greek, and what it says about Romans 7.

With surprising frequency my professor at Midwestern Baptist Theological Seminary, Dr. Tomlinson, would proclaim in his exegesis classes "I think these verses probably mean…" before enlightening us with a profound insight not found among any commentary — only to conclude "…but I don't know how to prove it."

Tomlinson had, figuratively speaking, spent decades in the first century — reading Ancient Greek literature, legal documents, magic spells, gravestones, mail, bathroom-stall graffiti, and anything else he could get his hands on. This gave him a keen sense of how the language was used and what the culture was like. However, the vast Ancient Greek library of texts that are still in existence consists of over a hundred million tokens. Though an Ancient Greek phrase might be accompanied by a web of associations for Tomlinson, if he had to convince someone else of these associations, he'd frequently be unable to recall the authors, works, books, and chapters that would justify his claims.

And Tomlinson wasn't entirely stuck in the past. He was up to date on the latest technology that assisted in finding Ancient Greek material. But unfortunately, the search tools were, and still are, quite limited.

While these resources are a huge improvement over having no search capability at all, it is still frequently the case that scholars have to dedicate days, weeks, and even months to find the things they are looking for.

A search for a rare word is served well by the resources already available. If a search returns eight hits, the user can easily just read the eight hits. But what is frequently the situation is that a user wants to know something more complex: Does an extremely common word ever get used in some certain way? Does some certain arrangement of grammar and syntax always lead to a certain connotation? When some certain group of words is used in conjunction will that always take some certain semantic domain as its object? The list of questions could go on for days.

For the sake of this post, here are five questions that matter in biblical scholarship — and the difficulty associated with researching them today.

At Cherith Labs we have built Korakes — a search engine for Ancient Greek that can be asked practically anything. It is in early development, but it's already incredibly powerful and accurate — both in precision and recall.

If you want Korakes to only return instances where σύν might be used with instrumentality, it gives you those few hits instead of the thousands. Same with κεφαλή meaning "source." If you only want it to return places where ἐν is modifying καθίζω and the object of the preposition is a personal being, it returns exactly those hits — instead of the thousands of places that καθίζω and ἐν happen to occur near each other.

Such queries are possible because Korakes is interacted with in a manner similar to ChatGPT. You write what you want it to find and it does it. In a future post, I'll go through the details of how it works and give some benchmarks. In the meantime, I'll walk you through one controversy I've recently used Korakes to investigate.

Few passages in the New Testament generate as much sustained debate as Romans 7. Is Paul describing his pre-Christian life under the law, or his ongoing experience as a believer?

One of the central arguments hinges on Paul's use of present-tense verbs. Many readers — including a long tradition of interpreters — believe Paul is describing his past life under the law, not his ongoing experience. But that raises an obvious puzzle: how could Paul be talking about his past if he's consistently using present-tense verbs? The proposed solution is that these are historical presents — a well-attested Greek phenomenon where a past event is narrated in the present tense for vividness, much like an English speaker telling a story might say "So I walk in, and he just looks at me…"

An objection was raised by the formidable grammarian Dan Wallace in his Greek Grammar: Beyond the Basics. He simply observed that the historical present never occurs in the first person in the New Testament. By contrast there are hundreds of historical presents in the third person. This would seem to strongly suggest that the historical present has a severe aversion to the first person, and might avoid it altogether.

I was initially taken aback by the observation. It's a powerful objection. But as the initial surprise wore off, I perceived a potential flaw in the logic: How often does an author of the New Testament even narrate his own past? Perhaps the first-person historical present never occurs because there's almost no situation where it could.

Formulating a method to test my objection was fairly simple. But its execution would be incredibly tedious and time-consuming. It required answering two questions: (1) Do historical presents occur in the first person at all? (2) If so, do they occur rarely enough that we might expect very few or even none in the New Testament? I broke down the investigation into several steps.

Performing these steps, which would have previously taken weeks, is now approachable in a single day. For the sake of this blog post, I won't carefully calculate exact numbers — but I should still be able to walk away with a sense of my claim's veracity.

Within minutes, Korakes confirmed Wallace's observation: no first-person historical present-tense verbs in the New Testament. But it also found several in a work ripe with opportunities for them — Flavius Josephus' autobiographical Life of Josephus.

Overall, Korakes found ten such instances throughout Josephus' autobiography. I suspect there may be more — in its current stage of development, Korakes might withhold instances that are too close together because they seem redundant. This is an area that needs further testing. Whatever the case, the result is clear: historical-present tense verbs do indeed occur in the first person in first-century Greek.

Wallace also observes that historical presents occur far less frequently in the writings of the erudite. Each of the other gospels contains roughly ten times more historical presents than Luke. Like Luke, Josephus would be counted among the highly educated. So one might expect Luke's usage rate to be roughly equal to its frequency in Life of Josephus. And indeed it is.

Two works of about the same size — Life of Josephus at 16,000 tokens, Luke's gospel at 19,000 — with about the same number of historical-present occurrences. One in the first person, the other in the third. This suggests historical presents are independent of grammatical person — they should occur at about the same rate whether in the first or third.

We did steps 1 and 2. We have an idea of how frequently historical presents occur. If this rate were to continue, how many might we expect in the New Testament? To estimate, I asked Korakes:

Korakes found 110 instances meeting that criteria. That's more than I would have guessed. I thought Paul talked about his past a little in a couple of places — but he does it, at least briefly, in every single letter he writes. John, likewise, in all of his letters. Jude. Peter. Acts contains a section that uses first-person verbs. And a big one I forgot about: Revelation — John relays much of his visionary experience in the first person.

For Luke and Josephus, this means a historical present is occurring about once every 1,600 tokens. How many tokens would these 110 instances roughly equate to? In my estimation, the number is not extreme enough to indicate that historical presents definitely should or should not be occurring in the New Testament.

Korakes is an enormous step forward.

In the span of a single investigation, Korakes accomplished what previously would have required weeks of manual labor: it confirmed Wallace's observation that no first-person historical presents occur in the New Testament; it located several clear counterexamples in Josephus, proving the construction is grammatically natural in first-century Greek; and it identified 110 candidates in the New Testament where a first-person historical present could have occurred — the pool against which any statistical assessment would be measured.

The significance for Romans 7 is this. Although Wallace's observation appears true, it's not particularly meaningful. We would not expect to see many — or any — first-person historical-present verbs in the New Testament. The difference between zero in the New Testament and a few in a single passage is unlikely to be statistically significant one way or the other. This does not settle the meaning of Romans 7, but it does likely nullify one particular objection to one particular understanding.

That is a meaningful exegetical result. But perhaps the more consequential takeaway is methodological. Answers like these have always been waiting in the corpus — just out of reach of the scholars who needed them. Questions that once required months of manual labor, or went unanswered altogether, can now be investigated in an afternoon. Whatever one concludes about Romans 7, scholars like Dr. Tomlinson should now find that the gulf between what they can intuit and what they can prove has drastically narrowed.