Here is the list of "weak" verbs I used:
{ Every form of the verb IS (to be)
"get", "gets", "got",
"move", "moves", "moved",
"walk", "walks", "walked",
"put", "puts",
"go", "goes", "went",
"smile", "smiles", "smiled",
"hear", "hears", "heard",
"see", "sees", "saw",
"look", "looks", "looked",
"watch", "watches", "watched",
"witness", "witnesses", "witnessed",
"stare", "stares", "stared",
"notice", "notices", "noticed",
"begin", "begins", "began",
"start", "starts", "started",
"feel", "feels", "felt",
"help", "helps", "helped",
"let", "lets",
"love", "loves", "loved",
"hate", "hates", "hated",
"knew", "know", "knows",
"set",
"stay", "stays", "stayed",
"meet", "meets", "met",
"keep", "keeps", "kept",
"appear", "appears", "appeared",
"sound", "sounds", "sounded",
"need", "needs", "needed",
"grow", "grows", "grew", }
Autocrit also highlighted forms of have, touch, and could, but that was overkill.
Next, I had to lay some ground rules to keep the comparison fair. First, my program detects the part of speech from context because most of these words could be either a noun or a verb. We only care about the noun usages. Second, I ignore dialog, because people use these weak terms in conversation all the time. I had to reduce the noise and focus on narrative only. Third, I had to exclude first-person point-of-view novels, since that is effectively conversation.
To test the range of weak content, I first ran the tool on all 25 of my books. I came up with an average weight of 18.25% weak, plus or minus a 3 percent range. Then I did the same thing for 26 common-domain masterpieces of literature. They averaged a whopping 28.47 percent (from 20.7 to 34.6). About one sentence in three violates this rule editing. Hmm. By this metric, these hacks like Tolstoy, Wells, Kipling, Twain, Burroughs, Dickens, Chopin, Joyce, and the like must not be very popular.
But Scott, these were examples from the 1920s. English has improved so much since then. I'll bite. Getting a word or text copy of any current novel is difficult. Luckily, Brandon Sanderson has a website where he shares and workshops his novels with fan support. I downloaded version 6.1 (final version before handing over to TOR) of the novel "Warbreaker." It scored 25.2, in the same range as the 1920s stuff. A fluke? I downloaded three more novels from some friends at Fiction Vortex, and they averaged 25.6 (+-7 percent range).
Therefore, good and popular writers use "weak" writing a quarter to a third of the time. By Autocrit logic, to be successful, I should inject my work with more weak verbs like a butcher grinds more fat into the hamburger. If we only flagged those books significantly above the classical metric, half the sentences in a document would be highlighted for change--and the recommendation would be wrong most of the time. My conclusion is that only a few weak verbs matter for pre-editing.
I have felt all along that my "Senescence" editor was right saying that narrators should rarely "start" or "begin" to do something; rather, they should just DO it, clearly and cleanly. My editor "Katie" was also correct that "get/got/gets/gotten" are a crutch that I should eliminate. Beyond these, I will add on a case-by-case basis. To one who knows how to mix a palette of words, no word is truly weak when it's direct and intentional.
