April 23, 2007

God's own Englishman with a tube up his nose

The extraordinary PENNsound site has recently added an extraordinary trove of Ezra Pound recordings, along with an essay by Richard Sieburth, "The Sound of Pound: A Listerner's Guide". One of the most interesting aspects of Sieburth's essay, to me, was his description of Pound's encounter with the L'abbé Jean-Pierre Rousselot.

In the spring of 1913, Pound crossed the Channel to prepare a series of seven articles on modern French poetry, published later that fall in the New Age as “The Approach to Paris.” While in Paris he met the poets André Spire and Robert de Souza, both of whom were ardent proponents of the new vers libre—whose central principle had already been incorporated into Pound’s initial Imagist manifesto (“As regarding rhythm: to compose in the sequence of the musical phrase, not in sequence of a metronome”). Both Spire and de Souza had been attracted to the work of the abbé Pierre-Jean Rousselot (1846-1924), the founder of experimental phonetics in France, who occasionally invited poets to his laboratories at the Collège de France to conduct experiments on the phonological analysis of poetic diction. The ardent vers libristes were presumably eager to find out whether Rousselot’s modern recording devices (which produced what look like intricate seismographs of vowels, consonants, pitch, and tempo) could provide scientific proof that free verse was, in its own way, just as “regular” or “formal” (in terms of the patternings of accents or quantities) as, say, the traditional alexandrine.

Rousselot was one of the founders of phonetics as a branch of natural science -- John Ohala wrote that he "is widely regarded as the 'father of experimental phonetics'", who "with Rosapelly (1876) ... pioneered and refined the use of the kymograph for study of speech articulations". The kymograph was basically a recording oscilloscope, the ancestor of the polygraph, which used lamp-blacked paper wrapped around a revolving drum to preserve for later study the motions of a vibrating stylus. Several styli were typically recorded at once, and their motions might record the beating of the heart, the volume or pressure of air in the lungs, or the more rapid variation in air pressure that corresponds to sound. A clockwork time base was also recorded, so that exact durations (and therefore frequencies) could be measured with a ruler.

Sieburth quotes Pound's description from a 1935 essay:

There was in those days [1912-1913] still a Parisian research for technique. Spire wrangled as if vers libre were a political doctrine. De Souza had what the old Abbé called une oreille très fine, but he, the Abbé, wrapped up De Souza’s poems and asked me to do likewise in returning them lest his servante should see what I was carrying. The Abbé was M. Rousselot who had made a machine for measuring the duration of verbal components. A quill or tube held in the nostril, a less shaved quill or other tube in the mouth, and your consonants signed as you spoke them. They return, One and by one, With fear, As half awakened, each letter with a double registration of quavering.

And another discussion from 1920:

I admit that many people did "dismiss" l'Abbé Rousselot; it is, for example, impossible to imagine God's own Englishman with one tube pushed up his nose, reciting verse down another, and God's own Parisian, and God's own supporter of the traditional alexandrine made a good deal of fun of the phonoscope. . . [T] his little machine with its two fine horn-point recording needles, and the scrolls for registering the belles vibrations offers a very interesting field of research for professors of phonetics, and, I think, considerable support, for those simple discriminations which the better poets have made, without being able to support them by much more than "feel" and "intuition." For example, the "laws" of Greek quantitative prosody do not correspond with an English reality. No one has succeeded in writing satisfactory English quantitative verse, according to these "rules," though, on the other hand, no English poet has seriously tried to write quantitative verse without by this effort improving his cadence.

Given the phonoscope one finds definitely a reason why one cannot hear the in the in a phrase like in the wind, as a "long." It isn't long. Whatever the Greeks may have done, one does not hear the beginning consonants of a word as musically part of the syllable of the last vowel in the word preceding; neither does the phonoscope so record them. All of which with many other finer distinctions can now be examined with great saving of breath and paper, whenever the questions are considered of sufficient interest, either by professors, or by neophytes in the arts of versification.

L'abbé Rousselot was one of the first to attempt systematic measurement of the physical facts of speech in relation to their linguistic function. For example, he was the first to describe the (universal) phenomenon now known as "pre-boundary lengthening". As a result, he was also one of the first to encounter the central paradox of perceptual psychology in its application to speech. The relationship between our subjective impressions and the objective facts of sound -- between what we hear and what we measure -- is a complex and subtle one. The colors we perceive in a scene are determined by the distribution in space and time of wavelengths of light, but the mapping is not a simple one, and a flux of photons that seems bright white in one context can appear deep black in another. The same can be said for pitches and vowel qualities and what Pound called "the duration of verbal components".

There is still considerable debate about how much effect the language of color might have on color perception. (For some discussion in earlier LL posts, see "Sapir-Whorf alert", 3/25/2004; "What would Whorf say?", 1/3/2006; "What Whorf would have said", 1/15/2006.) But in the case of speech, there's no debate -- how we hear speech sounds is profoundly and systematically influenced by the language(s) we know, and the sounds' linguistic context. And the principles of poetic rhythm, whether in metered verse or free, are about the linguistic patterning of physical sounds, not about the physical sounds alone.

Pound's discussion implies a simple parallelism of subjective and objective realities -- "one does not hear the beginning consonants of a word as musically part of the syllable of the last vowel in the word preceding; neither does the phonoscope so record them." He's talking about the principle of length by position, which in Greek and Latin metrics causes short vowels followed by two consonants (even across a word boundary) to be scanned as long. But I very much doubt that a recording of Sappho or Horace would have shown "the beginning consonants of a word as musically part of the syllable of the last vowel in the word preceding" in any simple way that is different from recordings of modern English speakers.

The basic distinction here is between short vowels in closed syllables (scanned as two moras) and short vowels in open syllables (scanned as one mora) -- in Greek, and in Latin poetry based on Greek models, this distinction was metrically fundamental. In English, the fundamental metrical dimension has always been stress, not syllable weight.

As for the re-syllabification of connected speech, English shows much the same sort of indirect evidence that we know from classical Greek. Thus many American speakers will flap and voice the final [t] of "at" in "at every opportunity", but will produce the same [t] as a glottal stop in "at yesterday's meeting". This allophonic variation is determined by syllable structure, and word-initial and word-final consonants are often syllabically adopted by neighboring words, in English as in classical Greek. But English metered verse is based on stress, not syllable weight (well, if we leave out the curious experiments of Sidney, Spenser, Campion and others from time to time in the history of English poetry...).

I don't mean to suggest that experimental phonetics is irrelevant to poetry. There's a lot to be learned, both about poetic form and about poetic performance. And the instrumentation these days is much more sophisticated than it was in 1913, and much more available -- instead of the precision instruments that Charles Verdin built for Rousselot, all you need is a microphone (or internet access to PENNsound) and a laptop. But now as in 1913, the topic needs serious engagement by people who understand both poetry and phonetics. Pound with a quill up his nose for an afternoon at the Collège de France is a striking image, but not a serious basis for rational investigation.

Unfortunately, things have not changed a great deal since Rousselot wrote Principes de phonétique expérimentale (1897), at least if we substitute "humanists" for "linguists":

… les procédés des sciences expérimentale sont assez étrangers aux linguists. Une sorte de terreur superstitieuse s’empare d’eux dès qu’il s’agit de toucher au méchanisme le plus simple. Il fallait donc leur montrer que la difficulté est moindre qu’ils ne se la figurant et leur faire entrevoir le champ immense que l’expérimentation ouvre devant eux.

... the procedures of experimental science are rather foreign to linguists. A sort of superstitious terror takes hold of them as soon as the question comes up of touching the simplest mechanism. It is thus necessary to show them that the difficulty is smaller than they perceive it to be, and to make them see the enormous world that experimentation opens before them.

The current holder of the chair of phonetics at the Sorbonne, which I believe was originally created for Rousselot, is Jacqueline Vaissière, who was a postdoc in Ken Stevens' lab at MIT while I was a grad student, and who also spent some time at Bell Labs while I worked there. When she took up her current position, she told me that there is a real, physical chair involved, arranged in a ceremonial room of some sort among the physical chairs corresponding to the other professorial chairs in the same institution, and a brass plaque on each one is inscribed with the names of those who have held the position over time. (At least, that's what I remember from our conversation at the time -- apologies if my memory is wrong.) This sort of correspondence between conceptual and physical realities, though charming, is as unusual for professorships as it is for phonemes.

[Tip of the hat to Sam Hughes, who wrote so entertainingly about Pound's school days in Philadelphia with William Carlos Williams, in the innocent time before Yeats, Eliot, Vorticism, Paris, Italy and Fascism. ]

Posted by Mark Liberman at April 23, 2007 09:25 AM