Monday, March 20, 2017

Cadillacs and ragged Fords


To continue with the automobile analogy advanced in the post on recommended fixes, I'm here with the tow truck for that broken-down Ford. Again. Three wheels are off and MacDonald P. Jackson wants more gas. What's that about "a follow-up study"? In the can, apparently:
... a follow-up study that finds high-frequency words and phoneme pairs that combine to distinguish Livingston’s verse from a similarly-sized corpus of verse by contributors to the newspapers and journals in which he published, and that places “The Night Before Christmas” with the Livingston poems. It remains true, however, that when capitalized and uncapitalized words are differentiated, the “Livingston-favored words” results show “The Night Before Christmas,” along with the majority of Livingston’s poems, falling beyond Moore’s range (Who Wrote, 133). --Jackson on Norsworthy
Mister, no disrespect but you might ought to think about trading it in.

Jackson justifies his withholding of "Charles Elphinstone" and other manuscript poems by Clement C. Moore in order "to test the efficacy of the tests" as "best practice." Best practice of what? Answer: authorship attribution studies in Jackson's field of globally acknowledged expertise which is Shakespeare and early modern English theater. Good, except in that game scholars play with bigger balls. Not literally, of course. I mean bigger data sets. In Shakespeare-World, it's all about Big Data.

How big? For quick and easy reference I checked Open Source Shakespeare for numbers of words in Shakespeare's plays. Zounds! Hamlet has 30,557 words; Midsummer Night's Dream 16,511. Average number of words per play for Shakespeare is 22,595. Your average Shakespeare play contains more words than all of Clement C. Moore's published poems combined (20,585 according to the Words in Moore page at Mary S. Van Deusen's marvelous Henry Livingston site). Hamlet at over 30,000 words plus half of MND is roughly the size of the entire Moore corpus including unpublished pieces, for example Moore's ambitious Charles Elphinstone and the revealing Biography of the heart.

The number of words in poetry by Henry Livingston, Jr. is difficult, maybe impossible to nail down. A few enthusiastic family members have sought to credit Livingston with poems he did not write. Where's a licensed English Professor when you really need one? We could use one now, preferably from the old school of Samuel Schoenbaum. Going old-school, the first order of scholarly business would be to determine if possible which pieces in Henry Livingston's manuscript book are actually his own compositions. Some of them might have been copied from the magazines and newspapers in which Livingston evidently published. (I do not claim that any were. I'm pointing out the undone dirty work of scholarship. Livingston submitted prose work by other writers without giving them credit, as the published essay on Eskimos demonstrates. Today we call that plagiarism, but when Livingston did it American literature was mostly just recycled British literature anyway, thoroughly and unavoidably derivative.)

Published in two different places, A Frontier Song in Livingston's manuscript book deserves particularly close attention for evidence of transmission. "The Frontier Song" appeared February 1791 in The New-York magazine; or, Literary repository, unsigned, with the refrain "My wife, my dog, and gun" closing each of the Song's four stanzas. The "Song" is sandwiched between two poems that also appear in Henry Livingston's manuscript book, The vine and oak and Epithalamium. Neither is signed. "Vine and Oak" is dated "Banks of the Hudson, Feb. 8, 1791," although its manuscript context implies an earlier date of c. 1786. Then on March 5, 1791, "A Frontier Song" appears over the signature "R—" in Henry Livingston's local newspaper the Poughkeepsie Journal. The Journal printing follows Livingston's manuscript reading "My puppy, squaw, and gun," as also reprinted in the American Museum - July 1792. Again, I'm not arguing that Livingston did not write these pieces. I do mean to point out how the authorship "controversy" about "The Night Before Christmas" too easily leapfrogs necessary grunt work of scholarship. On the other hand, I appreciate how questions of authorship draw attention to overlooked or neglected writings, and along the way create interesting and fun opportunities for learning. Hence my obsession with authorship questions.

Second job, weeding out newspaper and magazine contributions by other writers who also signed their contributions "R." If you think all "R's" for the period are Livingston, you might feel bound to add the following piece, published January 30, 1790 in the New-York Weekly Museum:
A TRUE DREAM.
I Dreamt,--my dear; (quoth Ralph to Joan,
One morning as they lay alone)
I can't help laughing, faith!--I dreamt,
Our Neighbour Charles was impotent,--
There is no truth in dreams (says Joan;)
And whilst I live I'll credit none.--
But afterwards, (quoth Ralph) I dreamt,
For all that he was impotent,
He got in bed my dear to thee!
And made a shift to cuckold me--
Good lack! (cry'd Joan) I never knew
A dream before that happen'd true.
R.
--New-York Weekly Museum January 30, 1790.
But that dont sound like Henry Livingston, Jr. and indeed can't be his original composition, since it formerly appeared in The Gentleman's Magazine for January 1733. True, Livingston could have submitted "Dream" over his cryptic initial. Still, not all "R's" represent Livingston in disguise. Does "R." in 18th century periodical literature ever signify merely "Reader"? Should we distinguish "R"; "R." with a period after it; and "R—." followed by a dash or dash and end punctuation? What else besides the signature "R." identifies the Song - Shrew'd remarkers (for example) as an original piece by Henry Livingston, Jr.? Another candidate for banishment from the official, approved Livingston collection would be Adventures of an American Eagle. Published over the signature "R." in 1822, "American Eagle" exhibits a curiously un-Livingston-like vocabulary (Capac, Canton Crape, quaff, chemic fire) and ends in a clunky moral about the "golden calf" as "Israel's sin and Israel's shame." Before "American Eagle," when was Livingston's last confirmed publication over the signature "R"? So far as I can tell, nothing else had been published over the signature of "R." in the Poughkeepsie Journal for twenty years and more before the appearance of "American Eagle" in 1822, making it doubtful that Livingston was actively writing and submitting poetry then.

Reality Check, and Double-Standard Alert


Imagine the worldwide delight (not to mention the raucous academic dinner parties) that would follow the finding of a published letter from Stratford upon Avon in which William Shakespeare exclaimed, "Kyd be hanged! I writ Arden of Faversham to amuse my kin." Or think of the joy in Poughkeepsie over the discovery of a published claim for "A Visit from St. Nicholas" by Henry Livingston, Jr. News flash! Clement C. Moore openly and vigorously claimed authorship of "A Visit from St. Nicholas" in his letter to Charles King, published in the New York American on March 1, 1844. Smart people like Samuel H. Turner and Evert A. Duyckinck who really knew Moore praised his integrity, decency, and humility. Nobody in Moore's lifetime ever called him a scoundrel, but Jackson now has to. Unless he changes his mind and stops driving that ragged Ford.


Subtracting the 1803 and 1819 Carrier Addresses (735 + 677 = 1412), for a start, would (as Jackson has reminded me) bring down the gradually shrinking corpus of known poems by Henry Livingston from 12,599 to 11,187 words--and that's still too many for Livingston, probably. And we would need far larger data sets for Jackson's common single words to have any chance of identifying distinctive patterns of usage, even in combination with other common, single words. The earlier post on recommended fixes attempted to fine-tune Jackson's methodology for the purpose of correctly discriminating between Livingston and Moore. Seems now I underestimated the difficulty of insufficient data. Regardless, I'm continually fascinated by results and potential results of Jackson's approach for purposes of literary criticism and scholarship. One very fine finding by Jackson is the revelatory presence of the noun heart in the list of "Moore-favored" high-frequency words, pulsating there amidst the usual run of essential but uninspiring function words. What a brilliant discovery! and most suggestive for the appreciation of the thing that distinguishes Moore's poetry from that of dabblers and recyclers like Livingston. It's heart.

And let's not neglect the medium-high frequency words for Moore that also jump out of Jackson's footnotes: joy, joys, life and light.

For identifying authorship, however, the data sets may be too small for Jackson's method to work its magic. Best practice requires big data. Shakespeare brainiacs prefer testing with blocks of multiple thousands of words, though for practical reasons they will settle for 1500. Considering the tiny and tinier bodies of work involved here, best practice when evaluating the poetry of Moore and Livingston would be to Use Extreme Caution. Primum non nocere, as my old friend Galen used to say. Without big data we should not bother overmuch about common pronouns, prepositions, conjunctions and such. Well, of course we can count them and play around with the numbers, but our concern for "best practice" will prohibit us from trying to use partial data of dubious value in the attempt to, say, deprive a fellow scholar and gentleman of his good reputation.

If we have to fool with common words, let's at least count them properly, all of them, before deciding which ones are "Livingston-favored" and which ones are "Moore-favored." Jackson points out that he deals with "Charles Elphinstone" and other withheld manuscript poems on page 92 of Who Wrote "The Night Before Christmas"? But look how his numbers replace and eventually displace words, and consider the cumulative effect of that numerical displacement. Moreover, in accordance with Jackson's method, his numbers often do not stand for words but combinations of words, and the percentages those seemingly magical word-combinations generate. That can't be best practice over such small samples. Sorry, we're sold out of Big Data. Have you tried the Searchable Database of Longfellow Poems?

Ironically, Jackson's successive tables of stats do a great job of hiding the actual words behind the numbers. The presentation of table after table serves to flatten essential differences which occasionally scream at you when you read and compare poems by Moore and Livingston. Besides masking the literary insignificance of much raw data, this process of flattening also works to preserve the false equivalency of Moore and Livingston as equally worthy contestants for the grand prize of The Night Before Christmas. That kind of hypothetical construction might pass for best practice in Early Modern studies where in some cases little to nothing has survived in the way of external documentary evidence. Sometimes there's no water in their swimming pool. When they give lessons over there they have to imagine how real swimmers look and act in real water. Over here it's Laguna Beach, baby. Nothing but surfing and splashing all day, an ocean of biographical and historical evidence before us. Best practice in 19th century studies will be to integrate external and internal evidence of authorship. As Harold Love advises,
"Neither kind should be given priority by fiat." --Attributing Authorship: An Introduction
Internal evidence like shared trigrams and near trigrams may not establish authorship, but in this particular case the numerous parallels confirm qualitatively, in aesthetically satisfying terms, what has already been firmly established by the biographical and historical evidence. With so much physical evidence on his side already, numerical ties go to Moore.

Thanks to the generosity and bravery of Mary S. Van Deusen, the word-frequency counts on her fantastic Henry Livingston website permit independent verification of Jackson's statistical tests. 

At their best, Jackson's tables of stats establish that one poet's verses can look like those of another poet when examined for usages of common words in isolation from their grammatical and literary contexts. His statistics "prove" just nothing in connection with the authorship of Moore's "A Visit from St. Nicholas." I checked the Manual, it's still best practice to acknowledge the limits of your model:
The results of computational stylistics are always a matter of probability, not of certainty. Writers are free agents, and language is an endlessly flexible instrument. Writers tend to remain within a defined band of style, but this is a propensity, not an iron law. In the past, quantitative work in literary studies has sometimes suffered from exaggerating the reliability of its findings.... Hugh Craig and Arthur F. Kinney
The good news is, best practice evolves. Around the same time that Jackson began moonlighting in the new-fallen snow, tracking the footsteps of Don Foster, their peers were formalizing the received opinion of Foster's methodology as bad practice. In Shakespeare, Computers, and the Mystery of Authorship, Hugh Craig and Arthur F. Kinney cite Foster as their prime example of how not to go about the tricky business of authorship attribution. Paradoxically, Foster's work is so bad that it's good. Foster's celebrated Funeral Elegy fiasco is now regarded as instructively bad--which makes it good for something, after all. To the same effect, Patrick Juola summarizes Foster's unusually public record as an exemplary "failure":
"Another noted episode (and noted failure) in the history of authorship attribution is the work in the late 1990s, of Don Foster [41, 44]."
--Authorship Attribution by Patrick Juola
As Joe Nickell and Seth Kaller have shown already, Foster's distorted treatment of Clement C. Moore and "'Twas The Night Before Christmas" offers no improvement over his notoriously flawed work on Shakespeare and Ford's Funeral Elegy. Best practice nowadays is to acknowledge Foster's mistakes, and avoid repeating them. Jackson in his book Who Wrote "The Night Before Christmas"? only gets halfway there.

Before determining "high frequency" and "medium-high frequency" words, Jackson excluded Moore's verse translations and numerous unpublished poems including the allegorical "Biography of the heart" and epic verse narrative "Charles Elphinstone." Besides being inappropriate for the small sample sizes involved, Jackson's exclusions pre-judged and mis-judged Moore's ability to creatively manipulate the "endlessly flexible instrument" of language. As "Charles Elphinstone" will reveal to the patient and attentive reader, our old boy had a wild imagination and was surprisingly ambitious for an amateur writer of light domestic verse. Hurrah!!! once more for Mary S. Van Deusen who heroically transcribed Charles Elphinstone, and made it available online with the rest of Moore's manuscript poems.

In future, best practice should make it easier for engaged readers to see words behind the screen of numbers. More transparency next time, please. Transparency seems especially desirable when presenting crucial numbers that relate to words in "The Night Before Christmas." That may sound obvious, but in chapters 16 and 17 of his book, Jackson makes us work to determine the precise distribution of supposedly Livingston-favored and Moore-favored words. He sticks the inventory of high frequency and medium-high frequency words in the footnotes, and in the main body only gives numbers and percentages, not the distribution of actual words. Let's peep behind the curtain.

In chapter 16, Jackson's Table 16.2 counts 53 "Livingston-favored" words in The Night Before Christmas, vs. 34 "Moore-favored" words. Here as in most of Jackson's tables, the numbers look more impressive as numbers than as the common words they are. Breaking down the 53 so-called "Livingston-favored" words in The Night Before Christmas:
I = 10
his - 18
my = 3
her = 1
on = 6
as = 2
was = 8
When = 3
me = 1
While =1
GRAND TOTAL = 53

his

By Jackson's method, Livingston gets credit for eighteen occurrences of the pronoun his, all with reference to (who else?) Santa Claus. 18 hits for one word represents a huge chunk, practically one third, of the total 53. To see why Jackson deems the common pronoun his a "Livingston-favored" word, we can to go to Mary S. Van Deusen's online charts of the word frequencies. 
Henry has 109 instances of his = 0.865% of his 12,599 words.
Moore has 112 instances of his = 0.544% of his 20,585 words, excluding his unpublished poems.

(Here and below, I am using word counts in Jackson's book and the Livingston website whenever possible.) For the sake of argument I will pretend to accept small differences as meaningful when Jackson says they are. Poetry is more than data, however, and percentages may have statistical "significance" without having any real poetical or literary importance.

Playing the numbers game for fun (no judgment now), let's see what happens when we incorporate more of the manuscript poetry including "Charles Elphinstone" and "Biography of the heart." True, the total sample for Moore might swell to the size of two Julius Caesars after adding everything Jackson kept out. I and my humble calculator app will try to adjust accordingly.

First we need to add occurrences of "his" in light blue on Mary Van Deusen's color-coded page of All Words, to capture the excluded poems. Then we can add instances from Moore's "Biography of the heart."

237 in Charles Elphinstone plus 5 in other manuscript poems = 242 + 16 in Biography of the Heart of Clement C. Moore = 258. (Not counting instances of "His," with "H" capitalized, considered a separate category). And I'm keeping translations out as requested, only adding "Biography of the heart," bringing the tally of Moore words to 38,714. I got that total number by adding 12 manuscript notebook poems with 3,479 words by Jackson's count + 13,670 words in "Charles Elphinstone" by Jackson's count + 980 in "Biography of the Heart" according to my ordinary Microsoft Word software  = 18,129 + 20,585 from Mary Van Deusen's Words in Moore page = 38,714 total words in Moore. Not counting the translations from Aeschylus, Metastasio, and (in manuscript) Petrarch. 

So then, add 258 to Moore's old total of 112 = 370 in 38,714 words = 0.956%. Now Moore's percentage turns out to be greater than Livingston's percentage, though not 1.2 times greater, which is what Jackson's method would require for "his" to count as a Moore-favored word. The main thing is, the word his can no longer be counted with "Henry-favored" words. Subtract 18 from Livingston's total of 53.

was

Henry = 47 instances of was = 0.373%
Moore = 62 instances of was = 0.301%

To Moore's total for the word "was" we need to add 89 hits in the unpublished manuscript poems, plus 6 in "Biography of the heart" = 95 + 62 = 157 which is 0.4055% of 38,714.

Thus, was must not be considered a "Livingston-favored" word, so we have to subtract 8 additional hits. 53-18 = 35; 35-8 = 27

on, etc.

Livingston has 68 = 0.540%
Moore has 81 = 0.393%

62 + 1 = 63 + 81 = 144 which makes 0.372% of 38,714. With the larger sample size, Moore's percentage for "on" actually drops a bit: "on" stays a "Livingston-favored" word. Livingston likewise keeps "I" and "my," with "her" and the rest of them.

Corrected GRAND TOTAL for the key stat in Jackson's Table 16.2 of "High-frequency words" should be 34 Moore-favored vs. 27 Livingston-favored.

The corrected percentage of Jackson's figure L/L + M on page 83 would be 27/34+27 = 44.26.

In chapter 17 Jackson examines "Medium-high frequency" words and counts 26 Moore-Favored words; vs. 29 Livingston-Favored words in "The Night Before Christmas."
Let's break down Livingston's 29:
an = 1
around = 1
As =2
good = 1
He = 5
him = 2
His = 3
little = 4
long = 1
up = 4
rose = 1
what = 2
name = 1
meet = 1
Here, too, observe the predominating cluster of masculine pronouns He/him/His. These collectively are responsible for 10 of the 29 hits for Livingston--again a big chunk, roughly one third of them.

He

Moore has 15 = 0.073%
Livingston has 18 = 0.143%

For Moore's usages of "He" add 45 in "Elphinstone" + 2 in "Biography of the heart" = 47 + 15 = 62 of 38,714 = 0.160% beating out Livingston. No longer Livingston-favored, subtract 5.

His

Moore 13 = 0.63 %
Livingston 13 = 0.103%

Add 24 + 5 = 29 + 13 = 42 which is 0.108% of 38,714. Dead heat now, so we have to subtract 3 more from the total of Livingston-favored words.

him

For "him" we need to add 57 from "Charles Elphinstone" plus 14 in "Biography" = 71 + the measly 16 in Moore 's original count 87 is 0.225% of 38,714 which doubles Livingston's 14 = 0.111 %
Ha! The word "him" turns out to be insanely Moore-favored. I think we're legally obligated to subtract 2 from Livingston's total, and also to add 2 to Moore's.

The rest of them can stand for now.

Corrected GRAND TOTAL for Table 17.2 of "Medium-High-frequency words in "The Night Before Christmas" 29 - 10 (5+3+2) = 19 for Livingston; vs. 25 + 2 = 27 for Moore, after correcting Jackson's partial stats. And the percentage of 19/19 + 27 = 41.30%

To be clear, I do understand that enlarging Moore's database to its right and proper size, and appropriately reducing Livingston's as well, means that all figures, everything in every category would need to be recounted and all percentages re-calculated. Further scrutiny seems desirable and potentially enlightening. I may try to do more of that down the road. Hopefully Jackson, too, will re-count and re-calculate along the lines I am recommending. I will be interested to learn exactly how the expansion of Moore's corpus would affect the words now categorized as Moore-favored; and how the reductions to Livingston's will affect the frequency rates of words currently treated as Livingston-favored. I guess that "heart" would remain a high-frequency word, distinctive to Moore. [Sigh.] Oh alright, let me check that one, starting with Words in Moore: Moore's heart rate now holding steady at 52 = 0.253%  To which, add 55 occurrences in "Elphinstone" and other manuscript poems plus 6 in "Biography of the heart." 52 + 61 = 113 which means that Moore's heart rate remains stronger than ever at 0.290% Woo-hoo!!!!

Randomly, let me try another one from "The Night Before Christmas." What happens to "not"?
Current total for Moore, 73 = 0.355%. Adding 90 from the manuscript poems gets us to 163 which is 0.421%. The word not, like heart, remains distinctive of Moore even with the greatly expanded sample size.

With my calculator still humming, I might as well check to see what how elimination of two Carrier Addresses would affect one or two Livingston-favored words. Having triumphantly dropped the all-important "his" from Jackson's tally of Livingston-favored words, I'd better take a minute to see how "his" would fare with Livingston's smaller sample.

his in Livingston according to Words in Henry currently stands at 109 = 0.865%
1803 Carrier Address instances of his,  4
1819 Carrier Address instances of his, 8
Well then, 109 - 12 = 97 which, after subtracting 735 + 677 = 1412 lines from Livingston's previous number of total words (12,599 - 1412 = 11,187) would be 0.867% About the same, still short of Moore's 0.956%

How about the word was?

was in Livingston poems currently stands at 47 = 0.373%
1803 Carrier Address instances of was, 0
1819 Carrier Address instances of was, 0
OK, still 47 which with the reduced corpus of 11,187 words increases the percentage to 0.420% , beating Moore's  0.403% by a nose. Nevertheless, by Jackson's rule the usage must be 1.2 times greater to count as Livingston-favored. To keep was, Livingston needed to reach 0.484%. Since it's basically a tie now, Jackson still must remove was from his list of Livingston-favored high frequency words in "The Night Before Christmas."  Ho ho ho.

One more before I go: him, from the list of medium-high frequency words. I'd better check that, too, having reassigned it from Livingston to Moore.

Over at Words in Henry, him in Livingston poems currently stands at 14 = 0.111 %
1803 Carrier Address instances of him, 1
1819 Carrier Address instances of him, 1
Now 14 - 2 = 12 in 11,187 words, which reduces Livingston's percentage to 0.107% vs. Moore's
0.225% Ha! Jackson still has to give him to Moore.

The evident swing from Livingston to Moore in "The Night Before Christmas" seems mainly though not exclusively a matter of pronouns. In our relatively small samples, some numbers and percentages may be volatile and change with every addition or subtraction to the data set. But numbers for "heart" and "not" remained pretty stable, which looks encouraging for the explanatory potential of even a smaller database. And percentages for the very small Livingston sample will probably be more volatile and therefore unreliable than Moore's. Especially with respect to the Livingston corpus, it seems unwise for now to rely on the commonest words for anything but fun and games with respect to Clement C. Moore's authorship of "The Night Before Christmas."

You can't escape from subjectivity, and best practice requires qualitative as well as quantitative analysis.

Best practice is integrative. The Cadillac model integrates qualitative with quantitative, external with internal evidence. Being integrative, best practice would never discount, as Jackson persistently does, excellent documentary evidence of authorship, especially with zero documentary evidence for any other claimant. (Sorry, the Livingston "witness letters" count as rich and fascinating family history, not evidence of authorship, as I will have to review in a separate post. And since Jackson opened the door, I'll walk him through the reindeer names so he won't keep tripping on Blixem.) In his Appendix III, Jackson demonstrates convincingly that the 1830 Sentinel broadsheet served as the printer's copy text for Moore's 1844 Poems. Good! Therefore, what Jackson has in front of him is solid physical evidence of Clement C. Moore's authorship of "The Night Before Christmas." For Christmas' sake, hang on to that "doctored copy of the 1830 Sentinel broadsheet," as Jackson describes it in a footnote to Who Wrote "The Night Before Christmas"? (page 184, note 4).

Doctored? I guess that means, "doctored" with revisions to the printed text in Dr. Moore's handwriting. In other words, physical evidence of authorship.  Double-standard alert!

Moore's use of the Sentinel broadsheet reminds me of Melville's practice when he submitted his more substantially revised version of "The Admiral of the White" for re-publication as "The Haglets" in John Marr and Other Sailors. Like Moore's "Visit" (though never so sensationally popular, of course), Melville's poem had already been printed in more than one newspaper. Melville gave the printer a marked-up and cut-up clipping from the New York Tribune.
The printer's copy for John Marr, in fact, consists of the Tribune clipping cut apart and fitted in with the manuscript pages.
--Hennig Cohen, Selected Poems of Herman Melville
Relevant images showing how Melville marked and scissored a printed newspaper text of his own poem may be found in Douglas Robillard's facsimile edition of John Marr and Other Sailors, published in 2006 by The Kent State University Press.

Did I mention the poem Clement C. Moore wrote for one of his daughters titled "From Saint Nicholas"? The manuscript is owned by the Museum of the City of New York. You can see a fine image of the manuscript poem online via Emily Chapin's beautifully illustrated article on Clement Clarke Moore and Santa in the City. Smoking-Gun Alert!   Double-standard alert!

Aside from hypothetical and largely counter-factual (which is to say, imaginary) constructions, no real problem exists with respect to the authorship of "The Night Before Christmas." There exists no counter-claim by Henry Livingston, Jr., and there is no troubling conflict between internal and external evidence. Granted, many of us would benefit from remedial classes in prosody and source-study. Our old-school English professor could favor us with a guest lecture on how anapests work in limericks and verses for kids, and why every poem in anapestic tetrameter kind of sounds the same. Any present-day disciple of Andrew B. Myers could guide us to Moore's poem by the easy and direct route from Washington Irving's house. Jackson's detour through Tristram Shandy is entertaining for sure, but far out of our way.

Possibly I have mentioned this once or twice already, but Jackson unreasonably slights good internal evidence for Moore. Here's my latest favorite example, not cited in the Melvilliana blog-review of his book. Jackson necessarily admits the word like as a Moore-favored word, but his method of combining multiple common words effectively masks how strong of a Moore-marker "like" really is. Introducing very persuasive evidence of similes, Joe Nickell has Jackson in a pickle. Like after like after like after like after like after like after like after like. What to do with so many Moore-markers? Jackson has to count them, of course, but he wishes they would behave and go away:
"Without those eight instances of "like"--six of them crowded into a mere nine lines, in a manner unique within either poet's work--the percentage of Livingston-favored test words in Table 17.1 would be 64.043, even further outside Moore's range." 
--MacDonald P. Jackson, Who Wrote the Night Before Christmas? p. 89
If all those Moore-words weren't there, the test results for Livingston would be stunning. I've heard that doing computational statistics is basically like sorting jellybeans. Hmm. Here I guess we've got jellybeans of two different colors: green for Livingston's high and medium-high frequency words, blue for Moore's. They're mixed up in the jellybean jar labelled "NIGHT BEFORE CHRISTMAS." Our task: sort by color and count them up. Earlier, the trouble was that some jellybeans are green AND blue. Problem now is, somebody keeps eating the blue ones.
 

Total Words

Directly confronting the inadequate sample sizes, Livingston's in particular, leads me to make a suggestion in closing. For a truly helpful follow-up study, MacDonald P. Jackson might test-drive the Cadillac model of best practice by investigating or re-investigating the "Total words" column in his Table 13.1. In all seriousness, the number of Total Words could turn out to be the easiest and most economical discriminator of all between Livingston and Moore.

In presenting "Definite and indefinite articles in Livingston and Moore," Jackson gives us potentially significant counts of the "Total words" in most (not all) poems by Moore and Livingston. "A Visit from St Nicholas" aka The Night Before Christmas contains 542 words. No stretch for Moore, but clearly beyond Livingston's comfort zone. Livingston liked to work in smaller fields, fenced in by clever puzzles and sturdy governing conceits. Only a handful of his pieces exceed 300 lines. Three poems over 400 lines are probably not by Livingston anyway (namely those 1803 and 1819 Carrier Addresses, and that strangely late, curiously worded 1822 parable, "Adventures of an American Eagle." Drop poems of doubtful authorship and we're left with a couple of plain fables and Livingston's playful verse-letter to his brother Beekman. None of which is as long as Moore's "A Visit from St. Nicholas."

Livingston almost never attempted longer poems. The average length of the 54 poems evaluated by Jackson after excluding very short poems of less than 100 words is 228 words (12,303/54). Delete the 1803 Carrier Address in Isaac Mitchell's Political Barometer, and the 1819 Address of doubtful authorship, and Livingston's average drops to 215 (11,187/52).  The 100-word minimum requirement excluded eleven poems by Livingston, but only one of Moore's published poems (Lines / On Seeing My Name Written By a Young Lady in the Sand of the Sea-Shore). Moore's average here, even without adding in "Charles Elphinstone," is 20,556/32 = 642 words. Adding "Sand" back in plus the manuscript poems brings the average to 38,714/47 = 824.

This here is just where I need Jackson's expert help, since averages won't properly convey what the Total Words category demonstrates about the essential difference between Livingston's poems and Moore's. Besides the jet off-take, railroad air horns and military spot, I want this Cadillac equipped with arithmetical means and all the standard deviations. And t-tests. It's a yellow convertible, so we definitely want good t-tests. Factory built. While we're rolling on down the road, tell me again... What's a t-test?
"I want four carburetors
And two straight exhausts
I'm burnin' aviation fuel
No matter what the cost
I want railroad air horns
And a military spot
And I want a five-year guarantee
On everything I got
I want ten-dollar deductible
I want twenty dollar notes
I want thirty thousand liability
That's all she wrote

I got me a car
And I'm headed on down the road
(No money down)
I don't have to worry
About that broken-down, ragged Ford."
--Chuck Berry - No Money Down

No comments:

Post a Comment