Lecture 5-3: Poetic Vocabulary, Rare Words, and the Limits of the Lexicon
Contents 37
- Lecture 1-1: What This Course Assumes, and What It Is For
- Lecture 1-2: How to Work Through a Verse Without Drowning
- Lecture 1-3: The Tools on Your Desk: Lexicon, Grammar, Concordance, Apparatus
- Lecture 2-1: The Shape of a Hebrew Sentence: Word Order and the Narrative Chain
- Lecture 2-2: Reading Genesis 22:1-14
- Lecture 2-3: Repetition, Wordplay, and What Translation Cannot Carry
- Lecture 3-1: The Verb System in Action: Aspect, Sequence, and the Perfect Problem
- Lecture 3-2: Reading 1 Samuel 17 as Connected Discourse
- Lecture 3-3: Direct Speech, Particles, and the Texture of Dialogue
- Lecture 4-1: Legal Hebrew: Casuistic and Apodictic Formulations
- Lecture 4-2: Reading Deuteronomy 6:4-9
- Lecture 4-3: Motive Clauses and the Rhetoric of Command
- Lecture 5-1: Parallelism, and Why It Is Not Repetition
- Lecture 5-2: Reading Psalm 23 and Psalm 1
- Lecture 5-3: Poetic Vocabulary, Rare Words, and the Limits of the Lexicon
- Lecture 6-1: Prophetic Forms: Oracle, Woe, Vision Report
- Lecture 6-2: Reading Amos 5:18-27
- Lecture 6-3: When the Text Is Hard: Corruption, Conjecture, and the Apparatus
- Lecture 7-1: Koine on the Page: Mark's Sentences and Why They Read as They Do
- Lecture 7-2: Reading Mark 4:35-41
- Lecture 7-3: Aspect Over Time: The Present, Aorist and Imperfect at Work
- Lecture 8-1: A More Literary Koine: Luke's Preface and the Septuagintal Style
- Lecture 8-2: Reading Luke 15:11-24
- Lecture 8-3: Participles: The Engine of Greek Narrative
- Lecture 9-1: Reading a Letter as Argument: Connectives and Signposts
- Lecture 9-2: Reading Galatians 2:15-21
- Lecture 9-3: Cases in Dispute: The Genitive and What Hangs on It
- Lecture 10-1: Long Sentences: Subordination, Anacoluthon, and Keeping the Thread
- Lecture 10-2: Reading Romans 3:21-26
- Lecture 10-3: Conditional Sentences, Questions, and the Diatribe
- Lecture 11-1: Simple Vocabulary, Difficult Meaning: The Johannine Style
- Lecture 11-2: Reading John 1:1-14
- Lecture 11-3: The Article, and What It Does and Does Not Prove
The two previous lectures assumed you could find out what the words mean. You look a word up, the lexicon gives you a gloss, and you get on with the real work of deciding how line B relates to line A. For most of the Hebrew Bible that assumption is sound. In poetry it fails more often than a beginner expects, and it fails in a particular way: the lexicon still gives you a gloss, printed in the same typeface and with the same air of authority as every other gloss on the page, and nothing on the page tells you that this one is a conjecture and the one above it is a report.
Three things make poetry harder on the vocabulary than narrative. First, poetry draws on a partly separate stock of words — a register that the prose of Genesis and 1 Samuel mostly avoids, so a reader who learned vocabulary from narrative arrives in the Psalter with the wrong two thousand words. Second, poetry is where the rare words cluster, including the words that occur exactly once in the entire Hebrew Bible, for which there is by definition no second context to compare. Third, poetry is where the ordinary props of interpretation are thinnest: no narrative frame telling you what is going on, no legal formula fixing the sense, and, as 5-1 showed, fewer grammatical markers.
So this lecture is about what you actually do when you meet a word you cannot securely identify. It teaches you to read a lexicon entry as an argument rather than as a report; to weigh an appeal to a cognate word in Ugaritic, Akkadian or Arabic; to use the ancient translations as evidence for how the word was understood before anyone wrote vowels; and to recognise the point at which a scholar has stopped reporting and started proposing. It ends with the principle that governs the rest of the course and is developed at length in 12-2: a word's history does not determine its meaning. Its use does. That principle is associated above all with James Barr, and the reason it is planted here rather than in chapter 12 is that poetry is where the temptation to violate it is strongest.
What these 50 minutes give you
- Distinguish a rare word from a *hapax legomenon*, and explain why the second is a smaller and less alarming category than it first appears.
- Name at least four items of the poetic vocabulary register and the prose words they stand alongside, and say what follows for how you build a reading vocabulary.
- Read a lexicon entry structurally — headword, cognate paragraph, numbered senses, occurrence list, hedging marks — and state, for any given gloss, what kind of evidence it rests on.
- Set out the case for and against arguing from a cognate in another Semitic language, name the main practitioners and the main critic, and apply a checklist to a specific proposal.
- Explain, using Psalm 22:16 and one other example, how the ancient versions function as evidence, and what they can and cannot show.
For dogs have surrounded me. A company of evildoers have enclosed me. They have pierced my hands and feet. — Psalms 22:16
World English Bible
I can count all of my bones. They look and stare at me. — Psalms 22:17
WEB
Like silver dross on an earthen vessel are the lips of a fervent one with an evil heart. — Proverbs 26:23
WEB
yet they had a file for the mattocks, and for the plowshares, and for the forks, and for the axes, and to set the goads. — 1 Samuel 13:21
WEB
what is man, that you think of him? What is the son of man, that you care for him? — Psalms 8:4
WEB
Two further specimens, chosen because in each of them an English version is printing a guess:
And one line where the poetic register is on display:
Verse numbers again. As 5-2 warned, the Psalms are numbered differently in Hebrew and English Bibles because of the superscriptions. The verse printed above as Psalms 22:16 is 22:17 in a Hebrew Bible, and that is the number under which you will find it discussed in every commentary that works from the Hebrew. Both numbers will be used below, always marked.
A note on pointing in this lecture. Where a Hebrew form is printed with its vowel points below, the pointing is that of the standard and can be checked in your Hebrew Bible. Where the argument is about the vowels, or where the reading is a variant known only from a manuscript or reconstructed from a translation, the form is printed in consonants only and labelled as such. That is not carelessness. Writing vowels onto a variant that never carried them would make a conjecture look like a text, which is the exact mistake this lecture is about.
1. A rare word and a word that occurs once are not the same problem
Take the two categories apart before doing anything else, because they are constantly run together.
A rare word occurs a handful of times. You have several contexts, they can be compared, and if they are varied enough — one in a law, one in a psalm, one in a narrative — you can often triangulate a sense with reasonable confidence. The work is ordinary lexicography and it usually succeeds.
Said once
A hapax legomenon — Greek for "said once," and the plural is hapax legomena — occurs exactly once. There is no second context. Whatever you conclude about it, you conclude from one sentence, plus whatever you can bring in from outside.
Now the deflation, which matters. The count of such words in the Hebrew Bible is often quoted at something over a thousand, and quoted in a tone suggesting that a thousand words of Scripture are unknown. Frederick E. Greenspahn's study Hapax Legomena in Biblical Hebrew (1984) took the list apart and made the decisive distinction: most words that occur once nonetheless have relatives inside Hebrew itself — another noun from the same root, a verb of which the word is plainly a participle, a place name built on it. Those are absolute hapax legomena only in a bookkeeping sense; their meaning is recoverable from Hebrew alone. The residue that has no relative anywhere in the Hebrew Bible is a much smaller set — Greenspahn's figure is in the low hundreds — and it is that residue where the real difficulty sits. Check Greenspahn's figures yourself before quoting them; they are frequently misreported, including, quite possibly, here.
This has a practical consequence. When a commentary says "the word is a hapax legomenon," your next question is not "so nobody knows?" but "does it have a root, and is the root attested?" If the answer is yes, you are in ordinary territory. If the answer is no, then everything the lexicon prints for that word came from somewhere outside Hebrew, and you need to know where.
2. The poetic register: a vocabulary you did not learn in Genesis
Biblical Hebrew, like most literary languages, has words that belong to elevated or poetic speech and are avoided in plain prose. Learning Hebrew from narrative and then opening the Psalter is a bit like learning English from a newspaper and then opening a hymnal: the grammar is the same and the diction is not.
Here are five pairs. The left column is the ordinary prose word; the right is the word that turns up in its place in poetry.
| ordinary | poetic | sense |
|---|---|---|
| / (ish / adam) | (enosh) | "man, human being" |
| (derekh) | (orach) | "way, path" |
| (davar) | (millah) | "word, utterance" |
| (ra'ah) | (chazah) | "see, behold" |
| (am) | (le'om) | "people, nation" |
Psalm 8:4, quoted in §2, opens with אֱנוֹשׁ — "what is enosh, that you remember him?" The word is not a different kind of human being. It is the same referent in a higher register, and translations that try to make something of it ("frail mortal man") are importing an etymological story we will come to in §8.
Two warnings about this table, both important.
First, the distribution is a tendency, not a rule. אֹרַח turns up in prose occasionally and דֶּרֶךְ is extremely common in poetry, including in Psalm 1 as 5-1 showed. Any statement of the form "Hebrew poetry uses X and prose uses Y" is a statistical claim, and you should treat a commentary that infers something from a single occurrence of a poetic word in a prose passage with the reserve you would give any argument from one data point.
Second, and more usefully: many of these poetic words come in fixed pairs that recur across many poems, one member in line A and the other in line B. Heaven and earth, and . Silver and gold, and . The prose word in line A answered by its poetic partner in line B. Adele Berlin, whom 5-1 introduced, treats these as the lexical-semantic level of , and the crucial observation for this lecture is that the same pairs occur in Ugaritic poetry, several centuries earlier and in a different Northwest Semitic language. That is a real, striking and well-attested finding. It is also, as §5 will show, the finding out of which an entire school of over-reading grew.
3. What a lexicon entry is actually claiming
Open BDB or HALOT at any entry and look at its architecture rather than at the gloss you came for. An entry is not one statement. It is four or five different kinds of statement stacked in a fixed order, and they are not equally well founded.
The headword. The lexicon prints a lemma — for a verb, conventionally the third masculine singular qal perfect, or in BDB the three root consonants. Choosing the lemma is already an analysis. For a form attested once, deciding which root it belongs to is often the whole problem, and the entry's placement announces a conclusion before you have read a word of it.
The arrangement. BDB, following a nineteenth-century convention inherited from Wilhelm Gesenius, arranges the whole dictionary by root, so that all the nouns and verbs judged to share three consonants sit together under one head. This is superb for seeing morphological families and it silently asserts that words sharing a root share a meaning — which is sometimes true and sometimes not. HALOT and the Dictionary of Classical Hebrew arrange alphabetically by word, which makes the book easier to use and drops the implied claim. If you are looking a word up in BDB, notice that you have been shown its relatives whether or not the relationship bears on the sense.
The cognate paragraph. Near the top of a BDB or HALOT entry you will usually find a compressed line of comparative material: an Arabic verb, an Akkadian noun, a Ugaritic form, sometimes Ethiopic or Syriac, in their own scripts or in transliteration. This paragraph is not the evidence for the senses printed below it. It is a separate statement about the word's ancestry. In an entry for a common word it is decoration. In an entry for a word occurring once, it may be the only thing the sense rests on, and the entry will not say so.
The numbered senses. These are the heart of the entry and they are the most misread part of it. The numbered senses are a classification of the attested occurrences — the lexicographer has read every passage where the word appears and sorted them into groups. They are not a menu of meanings available to the word on any occasion, and they are not ranked by "literal" versus "figurative." Sense 3 is not a deeper or truer meaning than sense 1; it is a different group of passages. When a preacher says "this word can also mean," what has actually happened is that a different group of passages has been pointed at.
The occurrence list, and the hedges. BDB uses a dagger (†) to mark entries where every occurrence is listed. Both lexicons use a vocabulary of doubt that students skim past: dub., "meaning uncertain," a gloss inside brackets, a question mark, "prob.," "perh.," "so most," "text corrupt." Learn to see these. They are the lexicographer telling you, in the space available, that the confident-looking English word three lines up is a best guess.
The Dictionary of Classical Hebrew, edited by David J. A. Clines (eight volumes, 1993-2011), made a deliberate and instructive decision here: it prints no etymology and no cognates at all, and derives its account of a word entirely from the word's behaviour within the Hebrew corpus — what it is the subject of, what it takes as object, what it stands in parallel with, what modifies it. It also covers Hebrew outside the Bible, in the Dead Sea Scrolls and the inscriptions. The result reads oddly at first and is extremely useful for exactly the words this lecture is about, because it shows you what can be said on Hebrew evidence alone. Where DCH is thin, the other dictionaries were leaning on something external.
Should dictionaries use etymology?
This is a point where readers will not agree. Clines's exclusion of etymology has been criticised as throwing away real information — Semitic comparative philology is a genuine discipline with genuine results, and a dictionary that refuses to use it is refusing evidence. The reply is that etymological material in a dictionary is read by students as though it were semantic evidence, whatever the editors intend, and that the cost of that misreading exceeds the value of the data. Both sides are arguing about the same fact and disagreeing about what a dictionary is for. Keep more than one lexicon on the desk and the question mostly stops being urgent.
4. Cognates: two cases where comparison worked
Comparative philology is not a vice. Two examples where evidence from outside Hebrew genuinely solved a problem, both worth knowing because they establish the standard by which §5's excesses should be judged.
Proverbs 26:23, and a Ugaritic word for glaze. The Masoretic text reads two words which look like "silver of dross," כֶּסֶף , and WEB prints "Like silver dross on an earthen vessel." That is a faithful rendering of the received text and it makes an odd image: dross is worthless, silver is not, and the composite phrase has to be strained into "silver-like dross" or "dross-silver." When the texts from Ras Shamra were published, a word spsg appeared in them with the sense "glaze" — itself apparently borrowed from Hittite. H. L. Ginsberg proposed dividing the Hebrew consonants differently, as a single word כספסגים rather than two, giving "like glaze on an earthen vessel": a smooth bright coating over cheap clay, which is exactly the image the second half of the proverb needs for lips that cover an evil heart. Many modern versions have adopted it.
Notice what makes this a good argument, because §5 depends on the contrast. It requires no change to a single Hebrew consonant — only a different word division, and word division in an unpointed consonantal text is not part of the text in the way letters are. The proposed sense fits the verse's own logic without being read into it. The cognate's meaning is independently established in its own corpus. And it explains the received reading: a scribe meeting an unfamiliar word divided it into two familiar ones. That is a repair, not an invention.
1 Samuel 13:21, and a stone in the dirt. The verse contains a word written , and until the twentieth century nobody knew what it was. WEB, following the older versions and the guess that the word was related to "mouth," prints "they had a file." Then excavation turned up small stone weights inscribed with those very letters, and the word turned out to be a unit of weight — about two-thirds of a shekel — so the verse is quoting a price: the charge for sharpening a tool was a pim. Modern versions print "the charge was a pim" or convert it to a fraction of a shekel.
This one is worth dwelling on because the evidence was not linguistic at all. It was an object with writing on it, dug out of the ground, and it did what no amount of comparative reasoning had managed. It is also a caution in the other direction: the older versions were not stupid, they were working without the stone, and their guess was reasonable given what they had. Every gloss in your lexicon for a word of this kind is somebody's reasonable guess given what they had.
A checklist. From those two cases and their opposite, here is a serviceable test for any cognate argument you meet:
- Is the cognate's meaning securely known in its own language, from enough contexts to be sure? Ugaritic is a small and damaged corpus; a Ugaritic word attested twice cannot settle a Hebrew word attested once.
- Do the sound correspondences hold by rule, or do the consonants merely look similar? Similar-looking consonants across Semitic languages are cheap.
- Does the proposed sense fit every occurrence of the Hebrew word, or only the awkward one?
- Does the proposal explain why the ancient versions read as they did, or does it simply ignore them?
- How many simultaneous changes does it need? A new sense is one move. A new sense plus repointing plus redivision plus an emended consonant is four, and four moves will let you extract almost any meaning from almost any line.
5. Cognates: when comparison runs away
This is a point where readers will not agree — though on this one the disagreement has largely settled, and it is instructive to see how.
Two names stand for the maximal use of comparative material. G. R. Driver, at Oxford, published a long series of articles proposing new senses for difficult words in the Prophets and elsewhere, mostly on the basis of Arabic — a language with an enormous attested vocabulary, which is precisely what makes it dangerous as a source of parallels. Mitchell Dahood, at the Pontifical Biblical Institute in Rome, produced the three-volume Psalms commentary in the Anchor Bible series (1966, 1968, 1970) and used Ugaritic on nearly every page. Dahood's premise was defensible and in part correct: the Psalms are Northwest Semitic poetry, Ugaritic is Northwest Semitic poetry from the same broad tradition, and the fixed word pairs mentioned in §2 show a real continuity. His practice was something else. Across three volumes he proposed hundreds of new meanings, re-divided words, re-analysed prepositions and suffixes on Ugaritic analogies, and produced translations of familiar psalms that no other scholar recognised.
James Barr wrote the systematic reply, Comparative Philology and the Text of the Old Testament (1968). Barr did not argue that comparative philology is illegitimate — he was himself a Semitist and the book is full of cases he accepts. He argued that the method had no controls: that given the size of the Arabic lexicon and the freedom to repoint an unvocalised text, a determined scholar can generate a plausible-looking new sense for almost any difficult word, and that a proposal which merely removes a difficulty is not thereby established. He also pointed out the asymmetry that makes the field hard to police: a new proposal is publishable and interesting, while "the traditional reading still seems best" is neither.
The verdict of the intervening half-century is fairly clear, and it is not a rejection of the enterprise. Very few of Dahood's specific proposals have entered the standard lexicons or the mainstream translations. What survives is the general point about the word pairs and the poetic tradition, which is now common property. Peter C. Craigie's Psalms 1-50 (Word, 1983) is often recommended as the model of the sober use of the same material — he had worked on Ugaritic himself, cites it where it helps, and declines it where it does not. Moshe Held's work on Hebrew and Akkadian is another model, and the discipline it enforces is worth stating: Held required that a comparison match a whole pattern of usage, the same verb with the same kind of subject and object in the same kind of construction, rather than a bare correspondence of root consonants.
So: read Dahood. He is exhilarating and his notes show what a maximally philological commentary looks like. Do not take a reading from him without checking whether anyone else has adopted it, and be aware that a sermon illustration traced back far enough sometimes lands in his footnotes.
6. The ancient versions, and what they can show
When the Hebrew is hard, one of the most valuable things you can do is find out how it was read by people much closer to it than you are. The Greek translation of the Hebrew Bible — conventionally the , though it is a collection of translations of very different character — was made from the third century BCE onwards. The Syriac Peshitta, the Aramaic Targums and the Latin Vulgate follow. All of them predate the Masoretic vowel points by many centuries.
What they can show:
- How the consonants were divided and vocalised by a reader who had the living tradition. If the Greek renders a word with a verb and the Masoretic pointing makes it a noun, you have two readings of one written form — exactly the situation 5-2 examined at Psalm 23:6.
- Where a translator was baffled. A Greek rendering that simply transliterates the Hebrew letters into Greek characters, or that supplies a vague word, tells you the difficulty is old.
What they cannot show:
- What the word meant. A translator faced with an unknown word guessed, as you would. An early guess is still a guess, and the Greek translators of some books were plainly working with limited Hebrew.
- What stood in their Hebrew manuscript. A difference between the Greek and the Masoretic text may be a different Hebrew text, or a different reading of the same text, or a translator's paraphrase, and disentangling these is the substance of chapter 12-1.
7. Working it through: Psalm 22:16 (Hebrew 22:17)
Now to the specimen, which brings together everything above and which you should handle with more than usual care, because it is a verse people fight about.
The Masoretic text of the verse, as printed in a Hebrew Bible:
כִּי סְבָבוּנִי הִקִּיפוּנִי כָּאֲרִי וְרַגְלָי
ki sevavuni kelavim / adat mere'im hiqqifuni / ka'ari yaday weraglay
Working translation of the first two clauses, which are not in dispute: "For dogs have surrounded me; a company of evildoers has encircled me." סְבָבוּנִי is a qal perfect third plural of , "to go around, surround," with a first-person singular object suffix; הִקִּיפוּנִי is a hiphil perfect third plural of with the same suffix, and the two verbs are a standard parallel pair. עֲדַת (adat) is the construct of , a company or assembly — the same noun that appeared in Psalm 1:5, "the assembly of the righteous."
Then the third clause. And here the working translation has to stop, because as pointed, the clause has no verb.
כָּאֲרִי (ka'ari) is transparently the preposition כְּ, "like," with the noun (ari), "lion," which is a normal poetic word for a lion and occurs elsewhere in this very psalm. So the pointed Hebrew reads, word for word, "like the lion, my hands and my feet." That is not a sentence. Something has to be supplied, and every translator who follows the Masoretic pointing supplies something different — most often a verb of mauling or binding taken from the lion image, so that the line becomes "like a lion, they are at my hands and my feet," or the noun phrase is read as an exclamation.
Now the other evidence.
- The Greek renders the clause with a verb: ὤρυξαν μου καὶ μου (oryxan cheiras mou kai podas mou), "they dug my hands and my feet." ὤρυξαν is an aorist of , "to dig, gouge," and this is the ancestor of "they pierced" in WEB and in most English versions.
- A group of medieval Hebrew manuscripts read, in consonants, כארו rather than כארי — a final waw rather than a final yod, two letters that are notoriously similar in several Hebrew scripts.
- A Psalms scroll from Naḥal Ḥever, found in the Judaean desert and older than any Masoretic manuscript by many centuries, reads at this point, again in consonants, כארו. It is written without vowels, as all such manuscripts are, and I print it that way here for the reason given in §2.
- The other ancient versions are reported in different forms in different handbooks. Do not take them from a commentary's summary; if you need them, look them up in a critical edition and see the sigla for yourself.
So what would כארו be? And here is the honest answer: nobody has produced a fully satisfying parse of it. It has been taken as a form of , "to dig" — the verb the Greek translators evidently had in mind, whose qal perfect third plural is (karu, "they dug") — but that requires accounting for the aleph, which does not belong to that root. It has been taken as a form of a root כאר otherwise unknown in Hebrew, glossed from cognates as "to bind" or "to be foul, shrivelled." It has been taken as a scribal variant of the lion reading. Each of these proposals runs into one of the difficulties on the §4 checklist.
Set the situation out plainly, because that is the skill this lecture is teaching:
| reading | difficulty | |
|---|---|---|
| Masoretic pointing | כָּאֲרִי "like the lion" | the clause has no verb; something must be supplied |
| variant consonants | כארו | no secure parse; the aleph is unexplained on the "dig" analysis |
| Greek | ὤρυξαν "they dug" | shows how it was read, not what it was |
Three observations to keep this in proportion.
First, the argument here is about a lexical and text-critical problem, not about a doctrine, and the two should not be soldered together. Psalm 22 is quoted in the passion narratives of the Gospels — verse 1 at the crucifixion, and verse 18 about the dividing of garments — and this clause is not among the quotations. That is a checkable fact about the New Testament text, and it means that far less rides on the word than the volume of the argument suggests. A student who learns that the case for a doctrine depends on a disputed Hebrew letter has been taught something false about how doctrine is established, and something that will not survive their first serious commentary.
Second, the evidence genuinely runs in more than one direction, and both directions have to be reported. The variant consonants are attested early, in a manuscript much older than the Masoretic tradition, and the Greek rendering is older still. Those are strong witnesses. Against that, the Masoretic reading is the harder one — a clause with no verb is not what a scribe would create — and the harder reading has a real claim, which is one of the criteria chapter 12-1 will examine.
Third, and most usefully for this chapter: notice what the lexicon does with a word like this. If you look up a proposed root כאר in a dictionary that admits such entries, you will find it printed with a gloss and, if the editors are careful, with a hedge. The gloss is not a report of the word's use in Hebrew, because it has no established use in Hebrew. It is an inference from a cognate, made in order to solve this verse. The dictionary is quoting the problem back to you as though it were the solution, and only the hedge — a bracket, a question mark, a "dub." — marks the difference. That circle is the single most important thing to be able to see, and once you can see it, you will start noticing it in entries that have nothing to do with disputed verses.
8. Etymology is not meaning
Everything above converges on one principle, and this is where the course states it for the first time.
Use, not origin
A word's meaning is determined by its use, not by its origin. Where a word came from — its root, its cognates, its history — is a fact about the past. What it means in a given sentence is a fact about how the community using it employed it at that time. The two can coincide and frequently do not, and there is no method for deriving the second from the first.
Everyone accepts this instantly about their own language. English "nice" descends from a Latin word for "ignorant"; "goodbye" is a contraction of "God be with you"; a "silly" person was once a blessed one. Nobody thinks a speaker who calls a meal nice is it ignorant. The etymology is true and irrelevant.
The reason it needs stating for Hebrew is that Hebrew's morphology makes the etymological story unusually visible. Most words wear three root consonants on the outside, related words cluster around them, and BDB helpfully prints them together on one page. It is very easy to move from "these five words share a root" to "these five words share a core idea" to "therefore this word, here, carries that idea." That move is invalid, and it is the engine of an enormous amount of preaching.
James Barr made this unavoidable in The Semantics of Biblical Language (1961), a book aimed principally at the theological dictionaries then in fashion, which built theological accounts of concepts out of the etymologies and root-clusters of the words. Barr's demonstration was that the method produces conclusions the texts do not support, and that its conclusions tend to arrive already agreed with the theology of the person applying it. Lecture 12-2 works through his specific fallacies — the etymological fallacy, illegitimate totality transfer, and the rest — with New Testament examples. What matters here is the negative instruction it gives you for reading poetry: when a commentary supports a sense by appealing to a root, ask what the word actually does in its attested contexts, and if the answer is "it is not attested anywhere else," ask what is left of the argument.
One example to fix it, chosen because it is not obscure. The divine name (shaddai), rendered "Almighty" in most English Bibles and appearing in Genesis 17:1 as אֵל שַׁדַּי (el shaddai). Proposals for its origin include: Akkadian šadû, "mountain," giving "the One of the Mountain"; the Hebrew noun (shad), "breast," giving a sense of nourishing; the verb (shadad), "to deal violently, destroy," giving "the Destroyer"; and the rabbinic reading that divides it as שֶׁ + , "the one who is sufficient," which is an interpretation rather than an etymology and which the Greek translator Aquila reflected with (hikanos, "sufficient"). The Greek of the Septuagint itself is not consistent, using (pantokrator, "all-ruling") in some books and other renderings elsewhere.
Four etymologies and a translation history. Each has been argued by competent people; none is secure; and if one of them were proved tomorrow it would still not tell you what the name conveyed to a reader of Genesis. For that you would have to look at how the name is used — where it occurs, in whose mouth, in what kind of speech — which is a different investigation entirely, conducted on Hebrew evidence, and which the Dictionary of Classical Hebrew's method is designed to perform. That is the shape of every good word study, and it is the shape 12-2 will formalise.
9. What to do at the desk
A short practical sequence, for the moment when you have hit a word you cannot identify.
- Look for the root inside Hebrew first. Concordance before lexicon. Is there a verb, a noun, a name built on the same consonants? Most apparent hapax legomena dissolve here.
- Read the whole lexicon entry, hedges included — not only the bold gloss. Note whether the sense is supported by Hebrew occurrences or by a cognate paragraph.
- Check a second lexicon, and if possible one that excludes etymology. Where BDB is confident and DCH is thin, you have learned something about where the confidence came from.
- Ask what the ancient versions did. If they diverge, the difficulty is old and the modern debate is a continuation, not a discovery.
- Ask what the context requires — the parallel line especially. In poetry, line B is often the best lexical evidence you have, which is the point at which this lecture rejoins 5-1.
- Then decide how much weight the reading can bear, and say so out loud when you use it. "The word occurs once and its sense is uncertain; the versions read it as X" is a sentence a congregation can follow, and it is true.
Step six is the one people skip. It is also the only one that costs anything.
The first diagram is the anatomy of a lexicon entry: what each part of it is claiming, and how firmly.
The second diagram lays out the evidence at Psalm 22:16 (Hebrew 22:17), which §7 worked through. Every Hebrew and Greek form in it is repeated in transliteration on the line beneath, so the diagram remains readable if your device has no Hebrew or Greek font.
6. Further Reading
- James Barr, The Semantics of Biblical Language (Oxford University Press, 1961). The book behind §8 and behind lecture 12-2. Chapters 4 and 6 are the ones to read first.
- James Barr, Comparative Philology and the Text of the Old Testament (Oxford University Press, 1968; reprinted with additions, Eisenbrauns, 1987). The systematic critique of the method discussed in §5, by someone who accepted the method's legitimacy.
- Frederick E. Greenspahn, Hapax Legomena in Biblical Hebrew: A Study of the Phenomenon and Its Treatment Since Antiquity (Scholars Press, 1984). Read it for the distinction that deflates the category, and check its figures before you quote them.
- Mitchell Dahood, Psalms I: 1-50, Psalms II: 51-100, Psalms III: 101-150, Anchor Bible 16, 17, 17A (Doubleday, 1966-1970). Read critically, as §5 explains.
- Peter C. Craigie, Psalms 1-50, Word Biblical Commentary 19 (Word, 1983; 2nd edition with Marvin Tate, Nelson, 2004). The sober use of the same Ugaritic material.
- David J. A. Clines, ed., The Dictionary of Classical Hebrew, 8 vols. (Sheffield Academic Press, 1993-2011), with the introduction to volume 1 on the exclusion of etymology. A one-volume Concise Dictionary of Classical Hebrew (2009) exists for those without shelf space.
- Francis Brown, S. R. Driver and Charles A. Briggs, A Hebrew and English Lexicon of the Old Testament (BDB), and Ludwig Koehler, Walter Baumgartner et al., The Hebrew and Aramaic Lexicon of the Old Testament (HALOT), trans. and ed. M. E. J. Richardson (Brill). Have both open for §3, and read their front matter, which explains the conventions students guess at.
- Emanuel Tov, of the Hebrew Bible (3rd edition, Fortress, 2012), chapters on the ancient translations, for what §6 only sketches. Chapter 12-1 returns to this.
- John Barton, Reading the Old Testament: Method in Biblical Study (2nd edition, Westminster John Knox, 1996), for the wider methodological setting of Barr's argument.
- Wilfred G. E. Watson, Classical Hebrew Poetry: A Guide to Its Techniques (JSOT Press, 1984), on word pairs and the poetic register discussed in §2.
In short
- Greenspahn's Hapax Legomena in Biblical Hebrew shows that most Hebrew words occurring once still have a recoverable root elsewhere in the language, so the truly unsolvable residue with no Hebrew relative at all is a much smaller set than the commonly quoted figure of "over a thousand" hapax legomena suggests.
- The Proverbs 26:23 "silver dross" problem was resolved when the Ugaritic word spsg ("glaze," borrowed from Hittite) let Ginsberg re-divide כספסגים into "like glaze on an earthen vessel" — requiring no change to a single consonant, only a different word division, and explaining the older reading as a scribal mis-division.
- 1 Samuel 13:21's obscure פים, translated "a file" by WEB following the older versions, turned out from excavated inscribed stone weights to be a unit of weight (about two-thirds of a shekel), so the verse actually names a price for tool-sharpening rather than an object.
- Psalm 22:16 (Hebrew 22:17)'s כָּאֲרִי ("like the lion") has no verb as pointed, and the case is unresolved between the Masoretic reading, a variant consonantal כארו found in medieval manuscripts and a Naḥal Ḥever Psalms scroll, and the Greek's ὤρυξαν ("they dug," behind English "pierced") — with the lecture stressing this clause is not among the New Testament's quotations of Psalm 22, so no doctrine should hang on the disputed letter.
- James Barr's Comparative Philology and the Text of the Old Testament answered Mitchell Dahood's and G. R. Driver's heavy reliance on Arabic and Ugaritic cognates by showing the method had no controls — a large enough comparative lexicon and freedom to repoint an unvocalized text lets a scholar manufacture a plausible new sense for almost any difficult word.
If you remember nothing else from this lecture: A lexicon gloss is an argument, not a report, and the skill this lecture teaches is learning to see which kind of evidence holds it up. Most words that occur "only once" in Hebrew still have a recoverable root elsewhere and are not really mysterious; the real puzzles are the much smaller residue with no relative at all. Two cases show the range of outcomes: Proverbs 26:23's "silver dross" was cleanly solved by a Ugaritic word for glaze and a simple re-division of the consonants, while Psalm 22:16's כָּאֲרִי ("like the lion") remains genuinely unresolved between the Masoretic text, a variant כארו attested in an ancient Judean-desert scroll, and the Greek's "they dug" — and, crucially, that disputed clause is not one the New Testament ever quotes, so nothing doctrinal should be built on it. James Barr's critique of Dahood and Driver's freewheeling use of Arabic and Ugaritic cognates stands behind the whole lecture: comparative philology can genuinely solve a problem, but without controls it can also manufacture whatever meaning a scholar wants.
Check yourself
Not "did you read it" but "can you explain it". Ticks are kept on this device only.
Reflection and discussion
- Reading exercise. Take Psalm 23:5, which 5-2 left unfinished, and work on the single word רְוָיָה (*revayah*). Look it up in two lexicons. For each entry, write down separately: the headword chosen, whether a cognate paragraph appears, how many Hebrew occurrences are cited, and any hedging language. Then say in one sentence how much weight the English gloss can carry, and compare your sentence with what WEB prints.
- Reading exercise. Read Psalm 22:16-18 (Hebrew 17-19) in Hebrew, slowly. Verses 17 and 18 present no lexical difficulty at all; verse 16 presents the difficulty examined in §7. Write out, for a reader with no Hebrew, a three-sentence note explaining the situation at verse 16 which would be equally acceptable to a Jewish, a Catholic and a Protestant reader. Then ask what you had to leave out, and whether leaving it out cost anything true.
- §4 gives a five-item checklist for cognate arguments. Apply it to the Proverbs 26:23 "glaze" proposal, item by item, and give the proposal a score. Then apply it to any one proposal you can find in Dahood's commentary on a psalm you know. What is the difference, concretely?
- The *Dictionary of Classical Hebrew* prints no etymologies on principle. Suppose you had to defend the opposite policy to Clines. What is the strongest case for keeping cognates in a dictionary, and what safeguard would you add?
- §8 says that if an etymology of שַׁדַּי were proved tomorrow it would still not tell you what the name conveyed to a reader of Genesis. Is that too strong? Construct the best counter-argument you can, then answer it.
- This lecture repeatedly tells you to report uncertainty out loud. Write the sentence you would actually say from a lectern about Psalm 22:16 — one sentence, no jargon, to an audience with no Hebrew. Read it aloud. If it sounds evasive, rewrite it until it does not.