March 12, 2004

Fear not your toes, though they are strong

On my flight to Japan a week ago, I thought about beautiful feet in the Messiah. On my flight back yesterday, I beguiled a couple of hours by trying to remember other poetical feet. Not metrical feet, mind you. I mean things like "How beautiful are thy feet with shoes" (Song of Solomon 7), or "And did those feet in ancient time" (Blake, Jerusalem), or "The noble son on sinewy feet advancing" (Whitman, Leaves of Grass), or "Cuando no puedo mirar tu cara/ miro tus pies" (Neruda, "Tus pies"), or "He got feet down below his knees" (Lennon, "Come together"). Since I soon ran out of remembered feet, and it was a long flight, I allowed myself to count things related to toes: "And she said 'It's a fact the whole world knows/ That Pobbles are happier without their toes"' (Lear, "The Pobble who has no toes"); or "Mi fea, el mar no tiene tus uñas en su tienda" (Neruda, Cien sonetos de amor XX).

I had three hours between flights at SFO, so I paid $9.95 for tmobile wifi access to catch up on email, and when I was done, I checked on some lines that I hadn't been able to remember fully. In the process, I stumbled across some uses of "toes" that puzzled me, until I realized that they were were typographical errors for "foes." I enjoyed them so much that I'll share them with you here.

For the Canadian poet Henry Alline (1748-1784), the Lion corpus has 56 hits for the word foes, in catchy little couplets like

Ten thousand foes with all their rage
Against my naked soul engage

or

Come, mighty God, these foes subdue,
Form my benighted soul anew

and resounding stanzas like

O God for my poor soul appear,
And make my foes submit;
Unlock, unlock this prison door,
And bring me from the pit.

According to the Lion, the word toes occurs three times in Alline's poetry. With their context, these are:

Although ten thousand toes beset
Their souls on ev'ry side
Jesus securely guides their feet
On him they may confide.

and

Fear not your toes, though they are strong,
The conquest doth to you belong;

and

When I am try'd he bears my grief,
And doth my toes destroy;
When in distress he brings relief
With his immortal joy.

I surmise that all three of these are the result of OCR errors happily not caught in post-editing. T for F is not a very likely lapsus calami, despite the touch-typing affinity of T and F, but it's a very common OCR error.

The Lion's OCR system and proofreader have unfortunately missed several other opportunities to improve Alline's verse, notably:

GLAD news to men, the Prince of Peace
Has in his triumphs rose!
From death and hell he takes release,
And tramples on his foes.

Most of the Lion's 2023 poetical hits for toes are not typos. For example, Ambrose Bierce wrote

Through summer suns and winter snows
I sets observin' of my toes

and Shakespeare (in Romeo and Juliet) has

Ladies that haue their toes
Vnplagu'd with Cornes, will walke about with you:
Ah my Mistresses, which of you all
Will now deny to dance? She that makes dainty,
She Ile sweare hath Cornes

The toes typos in Alline's verse are a good example of the sort of thing that spelling-correction algorithms are still not clever enough to catch. I'm not sure whether we yet know how to build a statistical language model that would register "fear not your toes" as a sufficiently improbable 18th-century hymn sequence to set off an alarm, as it did for me when I read it. In the case of Alline's oeuvre, it might be better to set the algorithms to looking for places to introduce additional errors that could plausibly be blamed on the OCR process -- this is a technically easier task, as well as a more aesthetically defensible one. Unfortunately the Lion corpus is not (as far as I know) available for either sort of research.

Posted by Mark Liberman at March 12, 2004 01:50 PM