So how is the coda-onset change -kt(-) > -pt(-) a case of lenition?
Abstract
The paper discusses how changes like -kt(-) > -pt(-) can be described as cases of lenition in Element
Theory approaches that assume that velars and labials do not have equal complexity. In Rumanian,
reflexes of Latin -kt(-) regularly show -pt(-) as the following data show:
(1) pie[pt] ‘chest’ < Latin pe[kt]-
la[pt]e ‘milk’ < Latin la[kt]-
lu[pt]ă ‘fight’ < Latin lu[kt]-
noa[pt]e ‘night’ < Latin no[kt]-
li[mn] ‘wood’ < Latin li[ŋn]u-
The change occurred in a phonological context which is regarded as a typical lenition site: a coda
obstruent, /k/, stands before another obstruent, /t/, and this phonological context is favorable to the
lenition of the coda /k/. Such lenition can be manifest in the loss of place or manner features, in loss
of structural complexity. Such lenitions are indeed attested in the other Romance languages: for
instance, Latin la[kt]- ‘milk’ gives Galician leite ‘id.’, Spanish leche ‘id.’, French lait [lɛ/le] ‘id.’;
and Latin lu[kt]- ‘fight’ gives Galician loita ‘id.’, Spanish lucha ‘id.’, French lutte‘id.’.
A similar case is found in Middle English when /x/ changed to /f/ word-finally (this context
is different from the Rumanian pattern) or before a consonant. Many such words are found in
contemporary English where is pronounced /f/:
(2) cough cf. Du kuchen
enough cf. G genug; Du genoeg
laughter cf. Du lachter
What poses problems for theories using privative monovalent elements is that it is matter for
debate whether velars and labials have identical elements for place of articulation or one of them is
more complex than the other, lacking such a place-defining element. For approaches assuming they
share the same melodic element, the change is a relatively simple case of change in headedness.
Backley (2011) is one such approach and the Rumanian data set is used to demostrate a crucial
point in his argumentation. Backley (2011: 77-84) assumes that both labials and velars have U
because in some languages they can be shown to pattern together to the exclusion of coronals.
Grouping velars and labials together, he acknowledges, goes back to at least Jakobson’s [+grave]
specification as opposed to [+acute] in coronals. In Backley’s analysis, the Rumanian data are
treated as a case of acoustic reinterpretation: speaker-hearers misinterpret the velar resonance as
labial since the acoustic information about the place of articulation of /k/ is masked by the following
consonant and might be reinterpreted as labial resonance, which is acoustically said to be similar.
Even though Backley presents cases where reinterpretation works in the opposite direction as well,
*p can be reinterpreted as /k/, such an explanation seems much more difficult to apply to the Middle
English cases since there need to be no following consonant at all and the masking effect of a preconsonantal
position was the crucial acoustic condition on reinterpetation. The contrast between
labials and velars, in his approach, is taken care of by headedness: velars have non-head U while
labials have U. The Rumanian data are not trivial under this analysis since going from unheaded U
(/k/) to headed U (/p/) in a weak phonological position requires theory internal motivation. The
Rumanian data are clearly contrary to the Caribbean Spanish data (2011: 84) where in identical
phonological context conce/p/to ‘concept’ is pronounced conce[k]to, and which are analyzed as
neutralisation suggesting that the velar place is weaker.
The present paper proposes that changes like -kt(-) > -pt(-) in Rumanian and /x/ > /f/ in
Middle English can be described as cases of genuine lenition in approaches assuming that velars
and labials do not have equal complexity and that labials are structurally more complex. Carvalho
(2013) proposes an analysis that is based on states of resonance cavity where both coronals and
velars are universally underspecified for place of articulation: they both have the element Ɨ, which
defines the open state of the pharyngeal cavity, in other words ATRness, as opposed to pharyngeals,
which lack this element and are produced with a constricted pharynx, that is with A instead ([ʕ/ʢ]
and [ɑ] when a vowel). The element Ɨ on its own is phonetically interpreted as [ɯ] when under a Vslot,
and the approximant [ɰ] when linked to a C-slot. Coronals are more complex than velars
because, in addition to Ɨ, they have the element T, for tongue, which defines the constricted state of
the supra-pharyngeal (oral) cavity as opposed to nasals that have N, defining the open state of that
oral cavity (and velars lacking it altogether). It is therefore true that neither velars nor coronals are
defined by a melodic element per se, and are therefore unspecified for place. Although Carvalho
does not give the specific representation for /p/ or /m/ in his approach, he does claim that “the sole
major consonantal articulator is [labial]ˮ and adds that [labial] “associates with SPREAD-containing
elements, that is Ɨ and Nˮ. This makes sense since the labial gesture is most audible when neither
the pharyngeal nor the supra-pharyngeal cavity is constricted. This also accounts for the patterning
together of velars and labials: they lack the constricted element T on the supra-pharyngeal tier.
Assuming [labial] is defined by U, this vocalic element, just like I or A, can associate with velars,
thus accounting for why velars are highly “colourable segmentsˮ. Colouring, by element spreading
from a neighbouring segment, can be conceived of as lenition because, although it does not destroy
structure as in vocalisation or debuccalisation, this process fills in, through double linking of U, the
melodic tier that is already open, and this is a loss of the luxury of leaving it empty.
The Rumanian data can be accounted for by the following representations: []
All that needs to be established is the origin of the labial gesture. The Rumanian
phenomenon is part of a larger areal pattern where indeed there is evidence that a preceding
rounded vowel seems to be a likely origin of labiality: Latin octu gave Dalmatian, a now extinct
language, guapto ‘eight’, cognatu gave comnut ‘male relative’. Also, Albanian lu/f/të ‘fight’ is
borrowed from Latin lu/k/ta. In the Middle English change /x/ > /f/, the labial environment is still
prevalent. Once labiality has spread from the vowel to /k/ or /x/, it is a simple analogical change that
sweeps through the lexicon, bringing it up to date with the new phonotactics.
References
Backley, Philip. 2011. An Introduction to Element Theory. Edinburgh University Press, Edinburgh.
Carvalho, Joaquim De Brandão. 2013. Why there is no backness: the case for dismissing both
[coronal] and [dorsal]. J.-L. Léonard & S. Naïm. Backness and backing. Lincom, pp.45-58.
Domains
Linguistics
Origin : Files produced by the author(s)
Loading...