Reasoning Rhyme: Lightness that Drew Me: Rhyme in Gaelic

One question that arises for the new terminology is whether it can cover rhyme practice in languages other than English adequately. It ought to be able to, as it's based on phonetic/phonemic correspondence rather than any single tradition of what does or doesn't constitute a rhyme. In this post, I start to test it out against Scottish Gaelic rhyme practice.

In Gaelic poetry, rhyme is traditionally considered to focus on vowel sounds, with no reference to the coda consonants of the rhyme syllables. Because stress is on the first syllable of a word in Gaelic, the rhyme vowel in end rhyme often falls on the penultimate syllable of the line. However, this isn't feminine rhyme, as the final syllable doesn't participate in the rhyme correspondence. Final-syllable rhymes are nonetheless perfectly possible (there are plenty monosyllables in the language), and it's also common to find internal rhyme, in which a vowel sound rhymes with a stressed vowel at a given point in a preceding or following line. It's also worth noting that vowels do not necesarily have to be identical, but closely related: /e/ and /E/ can rhyme, for example.

Meg Bateman's beautiful short poem "Aotramachd" neatly demonstrates the rhyme on the penultimate vowel sounds:

Aotramachd
B’ e d’aotromachd a rinn mo thàladh,
aotromachd do chainnte ’s do ghàire,
aotromachd do lethchinn nam làmhan,
d’aotrmachd lurach ùr mhàlda:
agus ’s e aotromachd do phòige
a tha a’ cur trasg air mo bheòil-sa,
is ’s e aotromachd do ghlaic mum chuairt-sa
a leigeas seachad leis an t-sruth mi.

Lightness
It was your lightness that drew me,
the lightness of your talk and your laughter,
the lightness of your cheek in my hands,
your sweet gentle modest lightness:
and it is the lightness of your kiss
that is starving my mouth,
and the lightness of your embrace
that will let me go adrift.

There's so much going on in this densely wrought lyric, with its rich mix of semantic, grammatical and phonological parallelism, but the feature I want to focus on is the end rhyme scheme. According to the Gaelic criteria outlined above, the poem rhymes AAAABBCC. The A rhyme is on à (thàladh:ghàire:làmhan:mhàlda), the B on ò (phòige:bheòil-sa) and the C on u (chuairt-sa:t-sruth mi). The first four lines--those that rhyme on à--are concerned with what drew the speaker to the object of her desire; they unpack the attractive lightness of the first line. In the subsequent four lines, as the lightness repels instead of attracts and the lovers drift apart, the rhyme scheme changes and fractures into two couplets.

It will not have escaped the astute reader's notice that the scheme of "Aotramachd" is more or less nuclear rhyme. In fact, analysed in these terms, we can break it down further into a clear A1A1A2A2BBCC structure. Before we do that, a quick note on Gaelic phonology is necessary.

Gaelic understanding recognises two classes of vowels and consonants: the broad and the slender. Slender vowels are i and e in the orthography, which represent the high front vowels /i/, /e/ and /E/; the rest are broad vowels. The slender vowels are sounds that, across languages, cause palatalisation of consonants*. And that's what the slender consonants are: palatalised versions of the basic, broad consonants (except that, in some dialects, the /rj/** is pronounced as a [T] or [D]). This means that what might be a secondary articulation for a consonant English is a significant difference in Gaelic, significant enough to create a separate class. I'm sure I needn't point out the importance of that for the possibilities in rhyme.

The A1 rhyme thàladh:ghàire transcribes roughly as /hAl6G:GArj@/. The rhyming vowels are identical. In the codas***, we have /l/ and /r/, which linguists class together under the term liquids. The correspondence of /h/ and the voiced velar fricative /G/ is less obvious but, if we take into account the morphophonemics--the interface of grammar and phonology--we get a clearer relationship. In their unmodified form, the rhyme fellows are tàladh and gàire, so their first sounds are /t/ and /g/ respectively: two non-nasal stops. However, because they follow the possessive adjectives mo (my) and do (the singular and familiar form of your) in the poem, they are lenited. That is to say, the stops weaken to fricatives. In the change from the velar stop /g/ to the velar fricative /G/, the place of articulation stays the same. However, in Gaelic, the voiceless coronal stop /t/ always lenites to the voiceless glottal fricative /h/. Not a close correspondence phonetically or even necessarily phonemically, but a clear morphophonemic correspondence.

I don't even need to transcribe the A2 rhyme for it to be obvious that it's a texbook example of segmental swapping: làmhan:mhàlda. However, perhaps a comment on dialect is appropriate here. In all dialects of Gaelic, the grapheme mh represents /v/ when it occurs at the beginning of a word. The word-medial mh in làmhan would be realised as /v/ in some dialects but, in others, it would be a /w/. Happily for my analysis, a recording of Meg Bateman reading the poem is on hand in The Jewel Box: it reveals her pronunciation to be /lAv@n/. Even if it was a /w/, it would still share the labial place of articulation with /v/, so it would be a case of featural swapping rather than segmental.

The B rhyme, phòige:bheòil-sa, transcribes as /fO:kj@:vjO:ljsa/. As is clear from the orthography and transcription, the place and manner features of the onsets are identical; the only major difference is in voicing. The codas might seem less closely related, but the /kj/ is a palatalised velar stop, which brings the blade of the tongue into a position much closer to that for /lj/ than in an unpalatalised /k/ so, in Gaelic terms, both codas are slender consonants. Interestingly, in the second rhyme fellow, the onset consonant /vj/ is palatalised. There's a lot going on in this apparently simple rhyme.

Finally, there's the C rhyme, chuairt-sa:t-sruth mi. It transcribes as /xuartjsa:tru:#mi/. Again, there's some segmental swapping going on here, with a /rtj/coda becoming a /tr/ onset. The second rhyme fellow has an empty coda. We could borrow the /m/ from mi, which would involve the sonority and the stop features of the /m/ breaking into the /r/ (sonority) and /t/ (stop) from the coda in chuairt.

There's much much more I could say about the poem, but this analysis indicates two things: that rhyme in Gaelic can, in fact, go beyond the vowel and that the new terminology is well placed to describe a non-English rhyme tradition. I might, in future posts, test it against older Gaelic poetry, but this will do for starters!

*Note that the presence of a vowel in the spelling doesn't necessarily mean it's pronounced. Like English, Gaelic has plenty of digraphs (two-letter symbols that represent one sound). However, the orthography often indicates historical pronunciation and the presence of slender vowels around a consonant always means that that consonant is slender.

**I use the /j/to indicate palatisation. Hence, /rj/ is a palatalised form of /r/.

***In these words and several others in the poem, what I'm treating as codas would probably be better described as the onsets of the second syllable. Therefore, as there are clear relationships, we can say that they are borrowed from the second syllables, which doesn't really participate in the rhyme scheme, as is shown by the varying vowel sounds and the variation between syllables with and without codas.

Reasoning Rhyme: Tiel Aisha Ansari's Objections 2

I closed my first post on Tiel Aisha Ansari's criticism of my new rhyme terminology by saying that the mention of structure brought me to her most fundamental objection. She is bothered that my nomenclature

risks broadening the definition of "rhyme" to the point where it loses all usefulness.

This comment grows out of her adherence to the terms assonance and consonance, discussed in the previous post in this series. She says she is "not entirely prepared to agree" with my "assertion" that "these relationships are in fact rhymes".

To a certain extent, the assertion is not mine, at least not originally. It belongs primarily to those poets who have used and who use relative rhymes of any kind in their rhyme schemes. The examples I give are drawn from poems, a fact that Ansari overlooks at certain points. For example, when she calls sang:work "another Philips [sic] example", she neglects to state that it comes from Wordsworth's "The Solitary Reaper" and, instead of engaging with the relationship that I sketch in the relevant post, simply says:

If there's a relationship there, I'm not hearing it.

That's why it's a remote rhyme. There's no identity, but there is a relationship between the corresponding segments in the fellows, even if the /w/:/s/ relationship is fairly distant. But, perhaps most importantly, they're clearly in a position in which, given the rest of the poem, one would expect a rhyme. They're the a rhymes in the last stanza. Interestingly, the a fellows in the first stanza constitute a removed rhyme; in two middle stanzas, they are fraternal rhymes, as in all the other rhymes throughout the poem. Wordsworth's use is therefore clearly structural: there's a signal of opening and closing.

It strikes me that what Ansari misses in her insistence on maintaining a purity of twin rhyme is this structural nature. The Penguin Dictionary of Literary Terms and Literary Theory states:

Rhyme has two main functions: (a) it echoes sounds and is thus a source of aesthetic satisfaction. ... (b) Rhyme assists in the actual structure of verse ..., simultaneously opening up and concluding the sense.

The sang:work rhyme fulfils both these functions: it echoes sounds, though more weakly than most rhymes in the poem by echoing features rather than segments; through that weakened relationship, it performs a structural function.

But I can still see why Ansari might be reluctant to admit that as a rhyme. And she would probably respond by referring me to the following comment from her post:

these poems [Armitage's "Kid" and "Poem"] present an overall impression of rhyme, though any particular perceived rhyme may silverfish away when you grab for it.

Well, to my mind, if it patterns like a rhyme and functions as a rhyme, it's a rhyme. To an extent, I can appreciate her reticence about the Armitage pieces and my attempt to parse them (though less so for "Poem" because of obvious structural features that clearly invoke the sonnet), but I don't accept it.

What would Ansari say about poets intermingling twin and relative rhymes? Are those not rhyme schemes? If they are rhyme schemes, how are the relative rhymes not rhymes? And what does she think about Wilfred Owen? I note that she avoids all reference to him, although I devote most of a post to his rhyme practice. I simply can't see how she can say that his peripheral rhyme schemes aren't rhyme schemes.

Moreover, traditional terms such as half rhyme, which The Penguin Dictionary defines as:

The repetition in accented syllables of the final consonant sound but without the correspondence of the vowel sound. ... a form of consonance",*

demonstrate that I'm not the first to recognise the way these patterns have been used and theorise about such relationships as rhymes. It might not please Ansari to hear that the dictionary's entry for "head rhyme" is "See ALLITERATION."

That brings us back to the aim of my terminology: a coherent, comprehensive nomenclature that is based on and points up the different degrees of featural relationship. Ansari is free to use as much or as little of it as she finds useful. In fact, she's most welcome to do so and I'd be delighted if she adopted any of it! However, I hope I've made her reconsider her objections. I've certainly enjoyed the debate so far!

I'll tie up this defenceI by clearing up a couple of smaller misunderstandings. In discussing a rhyme from "Kid", Ansari says that her terminology and mine both

make no reference to the fact that the unstressed syllables correspond, which actually is to my ear the defining feature of the Armitage poem in question.

I agree with the defining-feature-of-the-poem bit, but she's wrong about the terminology: longer:larder is a compound removed:identical rhyme. Compound rhymes are discussed here.

Also, Ansari says of strong:stink:

Note that it's not a remote rhyme because there's no relationship between the nuclei. I think.

For her benefit, I'll repeat the definition of remote rhyme from the post on the basic terminology:

there is no identity, but all segments in the syllable exhibit some relationship to their partners in the other fellow.

Therefore, strong:stink can't be a remote rhyme because there are two identitical segments in the onsets (st and st), even if there is an additional segment in one fellow. Hope that's helpful.

*In other words, removed rhyme, if we're not talking about identical onsets, which the definition simply ignores, even though there is onset identity in peer:pare, one of the rhymes in the Emily Dickinson stanzas given as an example (the first two of this poem), which, of course, makes it a peripheral rhyme.

Reasoning Rhyme: Tiel Aisha Ansari's Objections 1

It's good to see Tiel Aisha Ansari's critique of my Reasoning Rhyme posts on her Knocking From Inside blog. This is the first time anybody has taken me to task on any elements of my rhyme terminology and analysis, and it's envigorating.

Ansari says there is a lot she likes about my terminology. However, she's a bit less keen on certain aspects. For example, she states:

I don't quite see why the traditional terms assonance and consonance should be discarded in favor of nuclear rhyme and peripheral rhyme.

It's a fair question, and hope I can persuade her or at least clarify my rationale. Part of it is that I'm aiming for a reasonably consistent, coherent, detailed, comprehensive nomenclature and taxonomy, based on phonetic/phonological correspondences. The rationale for the attempt at consistence is partly to begin to move away from the confused array of terms in traditional usage. Most traditional rhyme terminology is based on aesthetic judgements or is impressionistic. Assonance and consonance don't suffer from that fault, but neither do they fit into the schema of kinship vocabulary that I've used. And my use of that vocabulary is not arbitrary: as I think Ansari understands, I'm trying to illustrate through my choice of words the underlying linguistic connections between the different forms of rhyme.

There is a further reason for not using the term assonance in particular. As I stated when I introduced my basic nomencalture, the definition of assonance makes no reference to the onsets and codas in rhyme fellows. By contrast, nuclear rhyme refers to a rhyme with identical nucleii but onsets and codas that are related, not identical. Therefore, it does not mean quite the same as assonance, although it focuses clearly and transparently on the identity that is highlighted in the definition of assonance.

That objection is understandable, but I find Ansari's comments on alliteration somewhat curious:

Philips [sic] points out that the onset, the consonant or consonant cluster before the vowel of the stressed syllable, is irrelevant to traditional rhyme but becomes important when considering consonance of the onset or coda. What he overlooks is that there is already a traditional term for consonance of the onset: it's called alliteration, defined as the identity of the first sounds preceding the vowels in the syllables carrying the primary stresses of two words. I would like to have seen Philips [sic] develop his terminology to cover alliteration by analogy, rather than redefining it as a form of rhyme.

I haven't redefined alliteration at all; I simply don't think it's the right term to use in the context of rhyme. Ansari is, of course, wholly correct that it's an accepted term for onset identity. However, alliteration goes no further than the onset; rhyme necessarily includes some reference to the other constituent parts of the syllable in each fellow. As far as alliteration is concerned, the rest of the word can do what it likes; for there to be rhyme, there must be some other correspondence, even if non-identical, in nucleus and/or coda. I wonder, for example, would Ansari say that Wilfred Owen's rhyme is alliterative? That would surely be a less than satisfactory description, given how strongly coda identity features in Owen's distinctive rhyme practice.

Besides, alliteration doesn't cover non-identical but related onsets*, which my terminology does. And not all rhyme fellows are lexically stressed syllables. If we follow the definition Ansari uses, an unstressed fellow cannot be said to alliterate, even if its onset is identical to that of the other rhyme fellow. That might be a problem with the definition as much as anything else, but that's what she gave us to work with.

Alliteration is therefore an inadequate term to use when considering rhyme, but it is extremely useful when examining the overall sound structure--the broader consonantal and vocalic music--of a line, stanza or complete poem. It might even be a primary structural device, as in alliterative meter, but that wouldn't make it rhyme and I don't really understand why Ansari believes that I think it would. This isn't in Ansari's definition, but alliteration tends to work within a line, even in alliterative meter. It can work across lines, but that's not the general trend. By contrast, the default for rhyme is to work across lines, although internal rhyme is hardly unheard-of. So there's a further structural reason for not using the term alliteration.

That mention of structure brings me to Ansari's most fundamental objection. However, rather than extend this post, I'll hold my discussion of that over for a separate one.

*Although I'd say that features can alliterate just as well as segments and that any full account of a line's music should take that fact into account.

In Denial?!

On last week's edition of The Verb, Paul Farley opined that "we" are "in denial about rhyme" because, when "we" rhyme, "we" use relative rhyme*.

If you've read my Reasoning Rhyme posts, it won't surprise you to learn that this is, in my opinion, utter tosh. Far from being a denial of rhyme, relative rhyme is a more linguistically subtle and complex form of rhyme, a realisation of the potentials inherent in the rhyme system but repressed by half-baked aesthetically conservative statements like Farely's.

Note too the insidious nature of the statement: Farley doesn't actually say we shouldn't use relative rhyme, just implies that twin rhyme* is real rhyme, boys and girls. And who is this "we" anyway? I presume it is the mainstream poetry-writing fraternity, Farley being a self-confessed guardian of the mainstream.

I have nothing against rhyming, whether twin or relative. Yes, I will confess to doing it myself sometimes, but only in the privacy of a poem. Moreover, I like some of Farley's work, even if I dislike anyone positioning themselves as mainstream or bust. However, ill-thought-out statements that creep towards closing down possibilities in poetry rather than opening them up really rile me. Okay, so it was radio and maybe it was off the cuff, but it sounded like something that had been bothering him for a while. Thankfully, he was paired in this discussion with Eleanor Rees, who spoke more sense.

There is always a danger that I might make daft statements like that myself, though obviously not such reactionary ones on rhyme. If I do, betake yourself to your local arms dealer and purchase a small surface-to-air missile with which to shoot me down.

*My terms, not his. See here for an explanation.

Reasoning Rhyme: To Begin at the Beginning: the Role of the Onset 2

The previous post in this series examined the role of the onset in Wilfred Owen's poetry. In this post, I'll consider the subtle ways in which the onset participates in the rhyme scheme in a couple of poems by Simon Armitage.

"Poem"
Many contemporary poets exploit a wide variety of rhymes, often within a single poem. One piece with a fascinating relationship between onsets is Armitage's sonnet “Poem”, which mixes nuclear and fraternal rhymes:

And if it snowed and snow covered the drive
he took a spade and tossed it to one side.
And always tucked his daughter up at night.
And slippered her the one time that she lied.

And every week he tipped up half his wage.
And what he didn’t spend each week he saved.
And praised his wife for every meal she made.
And once, for laughing, punched her in the face.

And for his mum he hired a private nurse.
And every Sunday taxied her to church.
And he blubbed when she went from bad to worse.
And twice he lifted ten quid from her purse.

Here’s how they rated him when they looked back:
sometimes he did this, sometimes he did that.

If one ignores the onsets, the rhyme scheme appears to run aaaa bbbb cccc dd, an incredibly simplistic scheme which parallels the matter-of-fact tone of the poem. An attempt to parse the rhymes into a more complex scheme on the basis of the codas would produce something highly irregular, such as abab cddc efee gg. The first stanza is the easiest to parse, and abab the most sensible option, based on the identical codas of side:lied, but it does leave the somewhat distant relationship drive:night in which the only features that the codas share is that they are not sonorants.

The second stanza is more problematic: none of the codas is identical and each possible relationship has complications. The closest relationship is saved:made, on the basis of which the stanza is parsed as cddc, but this relationship would involve syllable-internal intrusion. The relationship wage:saved would involve swapping the continuancy feature in the final two segments, and the relationship saved:face involves breaking the continuancy and coronal features.

In the third stanza, determining fellowship on the basis of the codas would leave church without a fellow and produce a three-fellowed rhyme in a poem which otherwise uses only two-fellowed rhymes. However, there is no question about the final stanza, as this contains only two candidates for fellowship.

A standard approach to rhyme obviously has grave problems parsing the rhyme scheme of “Poem”. However, if one determines fellowship within stanzas partly on the basis of shared features of the onsets, a choice of schemes more complex than aaaa bbbb cccc dd and more regular than abab cddc efee gg emerges: abab cdcd efef gg--the rhyme scheme of the Shakespearean sonnet form--or possibly aabb cdcd eeff gg.

As all the onsets in the first stanza are coronal, determining fellowship boils down to a choice between drive:night; side:lied (abab) or drive:side; night:lied (aabb). Since the former choice involves a stronger relationship between the onsets of the a-fellows in that they both contain sonorants and obstruents, it seems reasonable to opt for this as the scheme. This also accords with, but provides a stronger relationship than, a choice based on the codas and provides a rhyme scheme which looks like a Shakespearean sonnet.

In the second stanza we are faced with three labial segments ([w]age, [m]ade, [f]ace) and one coronal ([s]aved). Obviously, the place feature is insufficient to define fellowship here, so we turn instead to the manner feature, which divides the stanza neatly into two sonorants ([w] and [m]) and two fricatives ([s] and [f]). Thus this stanza is parsed as wage:made; saved:face (cdcd).

The third stanza can be parsed as efef or eeff. As with the first stanza, the former choice accords with the sonnet rhyme scheme and is based on a stronger relationship: nurse:worse ; church:purse as opposed to nurse:church; worse:purse.

Interestingly, the least intimate relationship between onsets occurs in the gg stanza: back:that, two voiced obstruents. Although comparable weak relationships are found in other stanzas, they are partnered by stronger relationships. However, there is no functional need for a strong relationship here, since there are only two candidates and therefore no questions as to what rhymes with what.

These observations indicate that the onsets are strengthening in their role as compensation for a weakening in the differential role of the nuclei. This compensation is not so strong if the codas provide enough differential information to determine fellowship (e.g. side:lied versus drive:night) or if there are only two candidates. This strengthening is probably motivated by a preference for a complex but regular rhyme scheme, for which the nuclei normally provide sufficient differential information for fellowship within a single stanza.

"Kid"
Another poem by Armitage informative for the status of onsets is the single stanza, 24 line “Kid”, the title poem of the collection from which “Poem” also comes, and therefore roughly contemporaneous with the latter work:

Batman, big shot, when you gave the order
to grow up, then let me loose to wander
leeward, through the wild blue yonder
as you liked to say, or ditched me, rather,
in the gutter ... well, I turned the corner.
Now I’ve scotched that ‘he was like a father
to me’ rumour, sacked it, blown the cover
on that ‘he was like an elder brother’
story, let the cat out of the bag on that caper
with the married woman, how you took her
downtown on expenses in the motor.
Holy robin-redbreast-nest-egg-shocker!
Holy roll me over in the clover,
I’m not playing ball boy any longer
Batman, now I’ve doffed that off-the-shoulder
Sherwood-Forest-green and scarlet number
for a pair of jeans and crew-neck jumper;
now I’m taller, harder, stronger, older.
Batman, it makes a marvellous picture:
you without your shadow, stewing over
chicken giblets in the pressure cooker,
next to nothing in the walk-in larder,
punching the palm of your hand all winter,
you baby, now I’m the real boy wonder.

This poem uses exclusively feminine rhymes, including one mosaic rhyme (took her:shocker ll.10,12), all of which end in unstressed schwa, in a highly irregular scheme: abbc acdd efef ghij jikg khll. As expected, the stressed syllables participate in the rhyme scheme.

Fellowship was determined by finding as many identical vowels as possible then looking for the closest overall match in the remaining fellows. This was straightforward for longer:larder (identical onsets, and low, back vowels) and winter:wonder (identical onsets and near-identical codas). The choice for picture:cooker was determined by the lack of any other possible fellows within reasonable distance. The choice of caper:motor; took her:shocker was based on the backness of the vowels and identity of the codas in the latter pair.

Seven out of the twelve rhymes are fraternal (that is, have identical vowels); the remaining five do not have identical vowels. These two types are not arranged with any regularity. Five of the onsets of the seven fraternal rhymes (i.e., approximately 70%) share fewer than three features. In three out of these five, one fellow has an empty onset (order:corner ll.1,5; clover:over ll.13,20; shoulder:older ll.15,18). Four of the onsets of the five other rhymes (i.e., 80%) share three or more features. In two of these, the onsets are identical (longer:larder ll.14,22; winter:wonder ll.23,24). These figures indicate that, for this poem at least, onsets in a rhyme relationship are more likely to bear some similarity to each other if the nuclei of the syllables differ. It would therefore appear that the role of the onsets strengthens to compensate for weakening in the role of the nuclei.

Besides the features mentioned above, there is another notable aspect to the rhyme scheme of “Kid”, namely that the onsets of the second syllable are not always identical between fellows as would be expected with feminine rhymes. Examining these yields startling results. Of the twelve rhymes, seven exhibit this dissimilarity, the same proportion as exhibit vowel identity. However, onset dissimilarity is in no way connected to vowel identity, as only four out of these seven (i.e., approximately 56%) do not exhibit vowel identity. Strikingly, in all seven rhymes, both fellows share three major features. Two of these rhymes, d and k, may share four features, if, for cover:brother, the dental nature of the two fricatives counts, and if, for picture:cooker the secondary, palatal articulation of the /k/ counts as parallel to the frication in the affricate. There is a striking regularity to the shared features, as shown below:

proportion in which
- lack of sonority is shared: 6 out of 7, approximately 84%
- continuancy or lack of it is shared: all 7, i.e., 100%
- non-continuancy is shared: 6 out of 7, approximately 84%
- voicing or lack of it is shared: 5 out of 7, approximately 70%
- place is shared, excluding d and k: 3 out of 7, approximately 42%
- place is shared, including d and k: 5 out of 7, approximately 70%

It therefore appears that rhymes with dissimilar and similar onsets in the second syllable are involved in a separate rhyme scheme from the stressed syllables. This makes sense if one recalls that the rimes of all the unstressed syllables in the poem are identical, an unexpected feature of the work. Thus the onsets of these syllables appear, in terms of this separate subscheme, to be strengthening in their roles and simultaneously, in terms of the overall feminine rhyme scheme, to be weakening in their roles in order to compensate for weakening in the differential role of the nuclei. This is a very similar situation to that discussed above in relation to “Poem”.

Reasoning Rhyme: To Begin at the Beginning: the Role of the Onset 1

As I stated in the first main post in this series, the traditional conception of rhyme doesn't allow for syllable onsets to play any role in rhyming, except in identical twin rhyme. The problem is that this analysis can't account for much 20th century and contemporary rhyme practice, in which the onset muscles in on defining fellowship when vowel identity is violated or carried to an extreme. I'll examine Wilfred Owen's work with reference to the former circumstance and, in the next post, a couple of poems by Simon Armitage to demonstrate the latter, but first, a few words about how rhyme practice changed last century.

In the 20th century, constraints on rhyme in poetry were largely relaxed. Anne Ridler, in her 1951 introduction to the second edition of The Faber Book of Modern Verse (1951) states that “poets shunned full rhymes ... because of their scarcity in English, which made so many of them too stale for further use.” A major development of the First World War poet, Wilfred Owen, is the use of a peripheral rhyme scheme in a great number of poems. Louis MacNiece, a major poet of the 1930s, used nuclear rhyme in his “Bagpipe Music”. Vernon Watkins' poem“White Blossoms”, written in the 1940s, alternates between fraternal rhyme, peripheral rhyme, removed rhyme and, in one instance, remote rhyme. Alternation between different forms of rhyme, including twin rhyme, has become a pattern for numerous poets. To snatch a few names from the air, it is found in the work of John Berryman, Seamus Heaney and Simon Armitage.

Wilfred Owen
In much of Owen's finest and best-known work, the use of peripheral rhyme is consistent. Such rhyme schemes involve identical onsets and codas, with the relationships between stressed vowels being reduced to featural. An analysis of all peripheral rhyme schemes in The Collected Poems of Wilfred Owen, edited by C. Day Lewis (Chatto & Windus, London, 1964) reveals that almost all these poems include a varying number of departures from the strict peripheral norm.

These aberrant rhymes, which can be either relative or twin, make up approximately 31% of the rhymes in Owen’s relative rhyme schemes. Of the relative rhymes in these aberrations, around 35% are removed (e.g., cheese:joys “A Terre” ll.40,41; [w]once:France “Futility”, ll.2,4). Some fall between peripheral and removed (crisp:grasp “Exposure” ll.37, 38), permit protrusion (brambles:rumbles “Exposure”, ll.7,8; openings:deepening “The Show” ll.14,15), permit one empty onset (apple:supple “Arms and the Boy” ll.10,11) or an /h/ (heels:curls “Arms and the Boy” ll.11,12), which is close to empty. Occasionally, both onsets or codas may be empty (amber:ember “Miners” ll.26,28).

Most of these aberrant rhymes can therefore be thought of as lax peripherals, rather than wholesale departures from the scheme. The examples with one empty onset are slightly problematic in that it is difficult to decide whether they are peripheral or removed. It is possible to think of these as having two empty onsets and protrusion in one fellow. However, fraternal rhymes in which one fellow has an empty onset could therefore also being analysed as identicals with protrusion.

Owen very occasionally admits fraternal rhymes to his peripheral rhyme schemes. I have found six such rhymes, of which five (i.e., 83%) exhibit a close relationship between the onsets:

[g]lozed:[k]losed (velar stops) “Exposure” ll. 26,29
deci/[m]ation:imagi[n]ation (nasal stops) “Insensibility” ll.18,19
[d]rained:[t]rained (alveolar stops) “Insensibility” ll.33,34
[dZ]est:[tS]est (palatoalveolar affricates) “A Terre” ll.58,61
he[r]eafter:[l]aughter (liquids) “‘Has Your Soul Sipped’” ll.2,4

The one remaining fraternal (there:air “Miners” ll.14,16) pairs [D] with an empty onset. In his fraternal rhyme schemes, Owen takes no more notice of onset relationship than other rhyming poets.

Though his rhyme schemes are basically masculine, Owen also makes use of feminine and mosaic rhyme in approximately 21% of his relative rhymes:

(a) mystery:mastery “Strange Meeting” ll.30,31
(b) silent:salient “Exposure” ll.2,3
(c) ever:over “A Terre” ll.28.29
(d) knive us:nervous “Exposure” ll.1,4
(e) snow-dazed:sun-dozed “Exposure” ll.22,23

As examples (a) to (d) show, only the first syllable usually takes part in the relative rhyme scheme. Example (e) is the only instance I found in Owen of both syllables participating in the relative rhyme. The difference lies in the stress pattern of the rhyme fellows: in (a) to (d), only the initial syllable is lexically stressed; in (e), both syllables of the compounds are lexically stressed. This also holds for Owen’s rare trisyllabic rhymes, allowing us to generalise that only lexically stressed syllables participate in Owen’s relative rhyme schemes, as all his masculine rhymes are lexically stressed.

In the next post, I'll look at a couple of Simon Armitage poems with some surprising characteristics in the rhyme schemes.

Reasoning Rhyme: Supplement on the Features of Consonants

Throughout the Reasoning Rhyme posts, I refer to the features of segments. These are the aspects of each consonant or vowel that make it the sound it is. This post and its companion on vowels attempt to explain the basics of feature theory in fairly non-technical language, but I'm afraid they'll remain pretty dry! (Remember that the symbols I use to transcribe sounds can be found here.)

The consonantal features describe the place of articulation, the manner of articulation and whether the consonant is voiced or voiceless. Consonants can differ in place of articulation (e.g., /p/ /t/ /k/), manner (e.g., /t/ /s/ /l/) and voicing (e.g., /z/ /s/or /b/ /p/).

Voicing is perhaps the easiest to explain. A voiced sound is one for which the vocal cords vibrate and a voiceless sound is one for which they don't vibrate. All vowels are voiced, but not all consonants. The example above should make it clear enough: in the two pairs, the voiced consonant comes first.

A consonant's place of articulation is defined by the place in the mouth where the air flow is constricted or stopped and what articulators are making the constriction. The main ones we need to worry about are labial, dental, alveolar, palatal, velar and a couple of variations on them:

The labial consonants are articulated with your lips in the case of /p/, /b/ and /m/ or your lips and upper teeth in the case of /f/ and /v/.

The dental consonants /T/ and /D/ are articulated by the tip of your tongue at the teeth.

The sizeable class of alveolar consonants--/t/, /d/, /s/, /z/, /n/, /l/ and /r/--is pronounced by the tip or blade of the tongue at the little ridge just back from your top teeth, which is called the alveolar ridge.

For the palatal consonants /S/, /Z/ and /j/, your tongue sits up at the roof of your mouth--the hard palate.

The sounds /tS/ and /dZ/--the ones at the start of chip and judge respectively--are known as palatoalveolar consonants because the tip of the tongue sits at the alveolar ridge and the body of the tongue presses up against the roof of the mouth.

The velar consonants /k/, /g/ and /N/ (as well as /x/ for those of us who don't lock our lochs) are pronounced by placing the back of your tongue up against the back of your mouth, which is known as the soft palate or velum.

The sound /w/ (and its voiceless counterpart, the Scots pronunciation of wh) is called labiovelar because it is articulated by rounding the lips and moving the tongue into the position for a velar consonant.

Phew! So that's place of articulation. Manner of articulation is another important aspect of consonantal features. It divides into questions of continuancy and sonority.

Continuancy is all about whether or not there is a continued flow of air through the mouth. If the flow of air is stopped completely, the consonant is called a stop. A stop can be nasal (/m/, /n/ and /N/), simply voiced (/b/, /d/ and /g/) or voiceless (/p/, /t/ and /k/).

There are several kinds of continuants, consonants that have a continuous air flow through the mouth: fricatives, affricates and approximants. In fricatives, the air flow is constricted at the same place as in the corresponding stop. Fricatives can be voiced (/v/, /D/, /z/ and /Z/) or voiceless (/f/, /T/, /s/, /S/ and /x/). The affricates /tS/ and /dZ/ are like a combination of a stop and a fricative in quick succession. Approximants have less of a constriction than fricatives and are voiced. They're almost like vowels, if that makes sense. You might see what I mean when I list them: /w/, /l/, /r/ and /j/. The approximants /l/ and /r/ are very closely related and often called liquids.

Stops, fricatives and affricates can be brought together under the term obstruents to describe the fact that the air flow is obstructed. Another important class is the sonorants. As the name implies, they're rather sonorous. It covers nasal stops, approximants and vowels.

If you're still with me (award yourself a chocolate and/or whisky!) you'll probably have noticed that the classes of sonorants, obstruents and continuants overlap in different ways. In the main posts, you'll see those overlaps in action.

I hope that's helpful. Please comment and let me know whether it clears things up or leaves you confused. I'll do another supplement on vowels, but I think they'll be a bit more straightforward to explain.

Reasoning Rhyme: More Complex Phenomena

Last time, I introduced the basics of my new terminology for rhyme. This post applies that terminology to the examination of more complex phenomena in rhyme.

Remember that you can find a key to the symbols I use to indicate individual sounds here. You should be familiar with the use of slashes round a symbol (e.g. /I/) from the previous post. In this one, I use square brackets in some examples (e.g., ta[ks]i) to emphasise the point that I'm talking about the pronunciation of the sound(s) concerned. I also use a slash in examples to indicate where a word is being broken and use # to show a word boundary(e.g., at all is shown as a/t#all).

There's a bit more linguistic terminology in this post. I hope it doesn't bamboozle too much. I'm still working on the supplement (it might become two!) that should help to make it clear, but I'll post this bit of the discussion meantime nonetheless.

Compound Rhymes
It is possible for a feminine rhyme to be a compound of two of the types discussed in the previous post. One such example from Simon Armitage’s poem “You May Turn Over and Begin ...” is the rhyme cocktails:pigtails, which is a remote:identical compound. Interestingly, the remote element (/k/ and /p/ are both voiceless stops; /Q/ and /I/are both mid vowels; /k/ and /g/ are both velar stops) is the syllable with primary stress. It is also important to note that, although both fellows in the above example are compound nouns, compound rhymes need not be confined to such forms. For instance, waggle:fickle could also be thought of as a remote:identical compound.

Intrusion and Protrusion
Astrid Holtman uses the term “subsequence rhyme” to refer to any rhyme in which a segment present in one fellow is absent from the other. She places it under the heading of “imperfect rhyme”, but it is preferable separate it from twin and relative rhyme. Twin and relative rhyme are clearly concerned with different degrees of relationship between corresponding segments in two rhyme fellows. “Subsequence rhyme” is orthogonal to this relationship and can appear in twin and relative rhyme. For concision and clarity, the term intrusion is to be preferred to “internal subsequence rhyme”, in which a segment is inserted word-internally, and protrusion, to “subsequence rhyme”, in which a segment is added at the end or beginning of a word (e.g. fleet'st:sweets in Shakespeare's Sonnet 19 ll.5, 7).

Swapping
In swapping, two segments or features, usually but not necessarily contiguous, swap places between the fellows. Segmental swapping can be seen in ll.9,10 of Louis MacNeice’s “Bagpipe Music”, which rhyme Blavat[sk]y:ta[ks]i.

As classification of rhyme is based on degree of segmental correspondence, it seems reasonable that individual features as well as entire segments should be able to swap. Featural swapping can be seen in ll.1, 3 of Seamus Heaney’s “The Diviner”, which rhyme stick:pluck. In the initial consonant clusters /st/ and /pl/, one segment is a voiceless stop (/t/ and /p/) and the other is a continuant (/s/ and /l/). The stops have different places of articulation (alveolar and labial), but the continuants share their place of articulation.

Featural swapping can also be seen in the initial consonant clusters in the example above from Shakespeare's Sonnet 19. Here, the labiality of the segments swaps, but the manner of articulation and voicing stay in place.

Borrowing
In borrowing, one fellow “borrows” a consonant from the preceding word to account for its correspondent in the other fellow. The borrowed segment may be identical, as in toil:a/t#all (Wilfred Owen, “Futility”, ll.13,14), or related, as in foil:o/[v]#oil (Gerard Manley Hopkins, “God’s Grandeur”, ll.2,3).

The second fellow in the example from “Futility” could also arguably be identical twin rhyme, in that a/t#all may be rhyming with tall in l.12, rather than toil. An argument for this is the presence of an extended correspondence in between l.12 and l.14: ...cl[eI] grew tall:...sl[i:]/p#a/t#all.

“Futility” and “God’s Grandeur” provide strong arguments for the reality of borrowing: in “Futility”, this pair is otherwise the only one in the poem lacking a relationship between the onsets; in “God’s Grandeur”, the borrowing of the /v/ reduces the number of pairs lacking a solid onset relationship to one, namely God:rod. Borrowing is obviously related to mosaic rhyme in that both cross word boundaries.

Breaking
Breaking occurs when two or more features of a single segment in one fellow are reproduced in two contiguous segments in the other fellow, e.g., [w]once:[fr]ance (Wilfred Owen, “Futility”, ll.2,4) in which the labiality and sonority of the /w/ are broken into the labiality of the /f/ and the sonority of the /r/. Most instances of breaking appear to involve the place of articulation and either sonority or continuancy.

In the next posts, I'll consider the role of the syllable onset in more detail, examining how it plays a role in rhyme schemes from poems by Wilfred Owen and Simon Armitage.

Reasoning Rhyme: Making It New

In the previous post in this series, I briefly explored why rhyme terminology was ripe for revision. In this post, I set out the basics of a revised terminology. At one point, this entails using a wee bit more involved terminology from linguistics, which I'll explain in a supplementary post so that this one doesn't get too unwieldy. However, I'll post this one before the supplement is ready, because I think you'll get most of what I have to say without the background stuff.

In the new terminology, rich and perfect rhyme are brought together under the heading twin rhyme and termed identical and fraternal rhyme respectively. “Imperfect rhyme” is replaced with relative rhyme, which is subdivided into close and distant rhyme.

Two types of close rhyme are identified: nuclear and peripheral. In nuclear rhyme, the vowels are identical, while the onsets and codas are related, e.g., made:face. This would traditionally be termed assonance, although the definition of assonance includes no reference to the correspondences in the onsets and codas. In peripheral rhyme, which would traditionally be called consonance, the onsets and codas are identical but the vowels are not, e.g., brutes:brats.

Distant relative rhyme subdivides into removed and remote. In removed rhyme, which is traditionally labelled pararhyme, either the onset or the coda is identical, but other constituents of the syllable are related, e.g., stick:pluck. In remote rhymes, there is no identity, but all segments in the syllable exhibit some relationship to their partners in the other fellow.

All these classes apply to masculine and feminine rhymes, although the only nuclear rhymes I've found so far are masculine. Some feminine rhymes may be compounds of more than one class, and I'll talk about that in the next post.

The new terms have a number of advantages:

• they unify the terms for segmental correspondence in rhyme, as it bases them all based on kinship terms, which furnish an obvious metaphor for such correspondence;
• although the terms are still not fully transparent, they are clearer than their traditional counterparts;
• they classify rhyme in terms of a clear continuum of segmental correspondence, from the greatest (in identical twin rhyme) to the weakest (in remote rhyme), as opposed to the rather haphazard fashion of the old terms;
• they contain no implicit value judgements, thus allowing for a better characterisation of the rhyming practice of poets who favour relative rhyme; and
• they conceive rhyme as including the onset, which, as shall be demonstrated in subsequent posts, is necessary for a full understanding of rhyme phenomena.

I've found few instances of remote rhyme, but sang:work is a clear one. It occurs in the fourth stanza of Wordsworth’s “The Solitary Reaper”, which uses an ababccdd rhyme scheme in each stanza. The remote a-rhyme sang:work, appears in ll.25, 27. Apart from the a-rhyme in the first stanza, which is removed (field:self, ll.1,3), the rest of the rhyme scheme is fraternal. Thus, by the time the reader reaches the fourth stanza, it is well established that this position rhymes, but whilst a fraternal rhyme is expected, some sort of relative is possible.

Given the concept of rhyme Wordsworth was probably working with, it is likely that he didn't perceive these words as rhyming, but the segments share features nonetheless: the /s/and /w/ are both continuants and articulated forward of the hard palate; the /a/ and /@:/ are both low, unrounded vowels; the /N/and the /k/ are both velar stops. (For an explanation of these symbols, see this page.) If Wordsworth didn't perceive sang:work as a rhyme, the featural similarity of the fellows is all the more striking.

In the next post, I'll discuss some slightly more complex phenomena.

Reasoning Rhyme: The Intolerable Wrestle With Words

This post concerns itself with brief definitions of the basic rhyme phenomena and a quick critque of the traditional terminology. Throughout this series, it'll be necessary to use some technical terms from linguistics and literary studies, but I'll explain them as I go along.

Rhyme involves a correspondence between two or more elements, usually, but not necessarily, in different lines of a poem. Each word or group of words involved in a given rhyme relationship is termed a fellow and the relationship is accordingly referred to as fellowship. Three correspondences are used to define the basic forms of rhyme: syllabic, morphological, and segmental. Traditional rhyme terminology covers each of these, which I will explain.

Syllabic correspondence is pretty self-explanatory. Monosyllabic rhymes (e.g., fame:name; begin:sin), which are almost all stressed, are referred to as masculine rhymes. Disyllabic rhymes (e.g., pleasure:treasure; rolling:controlling), in which the stress usually falls on the initial syllable, are referred to as feminine rhymes.

Morphological correspondence refers to the grammatical form of the rhyme fellows. Most rhymes are morphologically identical. That is, both fellows are normally one word or, less frequently, more than one word. Rhymes that are morphologically non-identical (e.g., jealous:tell us; intellectual: peck’d you all) are called mosaic rhymes. For obvious reasons, the vast majority of mosaic rhymes are feminine, although a masculine mosaic rhyme in English is not beyond possibility (e.g., cheese:she’s).

In linguistics, individual speech sounds are referred to as segments. Thus the term segmental correspondence covers the degree of similarity between the consonants and vowels in the rhyme fellows. Linguists divide syllables into the onset (the initial consonant(s), if the syllable has any) and the rime (the vowel and any following consonants). The rime is further divided into the nucleus (the vowel) and coda (the following consonants). Traditionally, rhyme is considered to begin at the nucleus of a stressed syllable and to require the nucleus and any coda consonants to be identical. For feminine rhymes, any syllables that follow the stressed syllable should also be identical. This discounting of the onset is a crucial point, which we'll return to. The one exception to the skipping of syllable onsets is so-called rich rhyme, which involves homophones (e.g., felt [past tense of feel] and felt [material]) or the repetition of the same word.

Normal masculine and feminine rhymes such as those quoted above are referred to as perfect or full rhyme. Other forms of rhyme, which will be explored properly later, are variously called half rhyme, near rhyme, slant rhyme and imperfect rhyme. These phenomena include assonance, consonance and pararhyme. Those terms and the reasons for labelling the phenomena that they describe as rhyme will be explored in a subsequent post.

The traditional terms for the segmental correspondences are inadequate for various reasons. Firstly, none of them are at all transparent. Secondly, because they exclude the onset, they fail to take account of various subdivisions of “imperfect” rhyme or the relationship between “perfect” and “rich” rhyme. Thirdly, they are misleading as to poetic practice: the term “rich” rhyme implies a highly developed form of rhyme, but such rhymes are considered poor practice; the terms “half-rhyme” or “imperfect” rhyme have been used to imply that such rhymes are shoddy, when in point of fact they are linguistically--and, in the opinion of some poets and readers, aesthetically--more subtle and interesting than their “perfect” cousins.

In the next post in this series, I'll explain my suggestions for a revised terminology.

Reasoning Rhyme: Prologue

Back in the dim and distant past, as a Linguistics student, I wrote an Honours dissertation on rhyme. It grew out of my reflection on rhyme practice as a reader and nascent writer of poetry and developed into a critque of a PhD thesis on rhyme by a Dutch linguist, Astrid Holtman.

It struck me that this blog would be a good place to revisit some of the material, leaving aside the more linguistic theoretical aspects of the dissertation and focussing on the stylistics. It's therefore my intention to create a thread of posts exploring this and reflecting further on rhyme. Don't know how long it'll take, but I hope it will generate some comment and discussion that will feed into the exploration as it develops. I'll start soon with a critque of traditional rhyme terminology.