Content uploaded by Daniel W. Hieber
Author content
All content in this area was uploaded by Daniel W. Hieber on May 04, 2022
Content may be subject to copyright.
Content uploaded by Daniel W. Hieber
Author content
All content in this area was uploaded by Daniel W. Hieber on Nov 28, 2019
Content may be subject to copyright.
Hieber, Daniel W. 2014. Degrees and dimensions of grammaticalization in Chitimacha
preverbs. Term paper, ‘Advanced Language Change’, Prof. Marianne Mithun, Fall 2014, UC
Santa Barbara.
1
Degrees and dimensions of grammaticalization in Chitimacha preverbs
Daniel W. Hieber
University of California, Santa Barbara
1. Introduction
It is well known that grammaticalization is a composite phenomenon, consisting of a
number of micro-level changes that give rise to broader patterns (Lehmann 2002:108–153;
Norde 2009:120). Some commonly-cited dimensions of grammaticalization include frequency,
productivity (actuation through the linguistic system), syntactic independence, lexicalization,
and semantic compositionality, co-occurrence with competing forms in a process of renewal,
and phonetic reduction, among others. But any one of these features may be absent, or present
more strongly than others, for any given case of grammaticalization. While a form might
exhibit a high degree of grammaticalization in terms of semantic shift, for example, it may also
have undergone little to no grammaticalization in terms of syntactic reanalysis or phonetic
reduction. Thus in explaining synchronic patterns through grammaticalization, it is not enough
to show how a thing became grammaticalized (i.e. demonstrate the grammaticalization
pathway). One must also examine the degree to which an element has been grammaticalized,
and the dimensions along which that grammaticalization has occurred. This paper therefore
adopts a multidimensional approach to the study of the grammaticalization of Chitimacha
preverbs, which assumes that there is more than one way for a construction to exhibit
grammaticalization, and that these dimensions of grammaticalization can therefore be analyzed
independently of each other. I conclude that the grammaticalization of Chitimacha preverbs is
not due to any one process or set of features, but rather the strong covariance among a number
of features canonically attributed to grammaticalization phenomena.
Though preverbs in Chitimacha clearly constitute a unified class with certain core
functions, their individual behaviors are diverse, and several members exhibit related but non-
preverbal functions. While synchronic polyfunctionality and distributional differences are often
explicable as the result of functional divergence and semantic persistence in grammaticalization
(Hopper and Traugott 2003), such processes would be insufficient to explain the state of affairs
for Chitimacha. Even preverbs which share the same grammaticalization pathways exhibit
different synchronic behavior. This paper shows that only when the degree and dimensions of
grammaticalization are accounted for does the synchronic behavior of Chitimacha preverbs
become explainable.
2
Section 2 introduces the data used for this study, and Section 3 gives an overview of the
system of preverbs and their possible historical origins. Section 4 examines the preverbs along
each dimension of grammaticalization listed above, showing that each method tells us different
things about the degree of grammaticalization for each preverb. Section 5 summarizes these
findings, and argues that it is only when the dimensions of grammaticalization sufficiently
covary that we recognize an instance of grammaticalization at work.
2. Data & Methodology
The Chitimacha language (ctm) was spoken in Louisiana until the death of its last native
speaker in 1940. While the recent availability of archival materials in digital form has
facilitated renewed academic interest in the language (Iannucci 2009; Hieber 2013; Brown,
Wichmann, and Beck 2014; Mithun forthcoming), there remain few published descriptions of
the language. The data for this study come from a collection of 88 texts dictated by one of the
last fluent speakers, Chief Benjamin Paul, to linguist Morris Swadesh between 1930 and the
time of Ben’s death in 1934. These texts consist mostly of tribal legends and personal
narratives, but also a few expository and procedural texts. Swadesh later compiled these into a
typed manuscript which he deposited at the American Philosophical Society Library in
Philadelphia (Swadesh 1953). He also elicited 22 texts from Ben’s niece, Delphine Ducloux,
but since she was somewhat less fluent than Ben, her texts were excluded from this study. For
this research, Ben’s texts were retyped using the linguistic analysis software Fieldworks
Language Explorer (FLEx) (Summer Institute of Linguistics 2013) and glossed at the sentence
level using Swadesh’s free translations. Some general statistics regarding the resulting corpus
are given in Table 1.
Table 1. Statistics for the corpus of Chitimacha texts
# of texts
88
# of paragraphs
418
# of sentences
3496
# of lexical entries (lexemes)
~3724
# of unique words (types)
4467
# of words (tokens)
29028
3. Background
A PREVERB is definitionally a category in flux. A common feature of all definitions is the
synchronic layering indicative of grammaticalization in process, where certain (older) preverbs
are more tightly syntactically bound to the verb and add only a vague contribution to its
3
meaning, and other (newer) preverbs are more clearly lexical and may be separated from the
verb while still remaining a cohesive unit with it—a phenomenon known as TMESIS (Booij &
van Marle 2003:1, 88; Diessel 1999:141; Lehmann 2002:86–92; Matthews 2007:318).
Individual preverbs may also exhibit divergence, so that lexical and grammatical uses of the
same form coexist synchronically. Matthews (2007:318) even notes that, “It is perhaps for this
last case that the term [preverb] is most useful.” Preverbs are therefore ideal candidates for
grammaticalization studies like the present one.
Chitimacha is primarily verb-final and morphologically suffixing. Almost the sole exception
to this preference is the system of preverbs, which in most cases appear immediately before the
verb and comprise a lexical and syntactic unit with the verb. The Chitimacha preverbs and their
meanings are presented in Table 2.
4
Table 2. Chitimacha preverbs and their functions
ʔap
1. (to) here
2. coming
ʔapʃ
1. returning here
2. together
3. randomly
4. reflexive
5. reciprocal
hi
1. (to) there
2. going
his
1. undoing
2. returning
3. doing in response
4. doing again
kap
1. up
2. suddenly
3. inceptive
4. inchoative
5. stative
kaːpˀs
1. up
kas
1. returning there
2. reverse
3. apart
ni
1. down
2. thing
3. detransitivizer
4. compound nominalizer
5. imperative
The preverbs have very different historical origins.
ʔapʃ
and
kaːpˀs
derive from
ʔap
and
kap
plus an internally-reconstructable reversative *
-ʃ
.
his
too may follow this pattern with
hi
.
ʔap
and
hi
themselves are of uncertain origin. One known origin for preverbs crosslinguistically is
the grammaticalization of preverbal adverbs into the verb complex (Lehmann 2002:87). This
would be surprising for Chitimacha, which has no strong class of adverbs. Any words with
5
adverbial function are bimorphemic and clearly very recent, if not still semantically
decompositional.
Forgacs (2004) documents a different grammaticalization pathway for Hungarian, where
some preverbs arose from postpositions that were reanalyzed as being part of the verb rather
than the preceding noun. Chitimacha shows evidence that at least some of its preverbs followed
a similar trajectory, by the way that preverbs affect the semantics and transitivity of the verb.
Consider the meaning of the verb
kow-
‘call’ when used with and without the preverb
hi
‘to,
toward’:
1
(1)
ni tiːkmiʃ hi koːnaka
Governor to we.called
‘we called (to) the Governor’ (A03e.2)
(2)
ʔakʃuʃ hetʃˀin koːʃnaʔa.
cypress holy they.call
‘They call (i.e. name) them holy cypresses.’ (A09f.2)
In (1)
hi
is indistinguishable from a postposition. At the same time,
hi
does also frequently co-
occur with the lexical postposition
hup
‘to, towards’, perhaps casting doubt on the
postpositional origin of
hi
:
(3)
ʃeːnink hup hi nitʃwiʔi
pond to to he.came.to.water
‘he came to the edge of a pond’ (A01a.2)
The preverb
kas
‘returning back there, reversing, apart’ likely comes from the internally-
reconstructable theme *
kaʔ
‘extending across, touching’ + the reversative *
-ʃ
.
kap
‘up,
starting’ is even more uncertain, but may relate to the root *
kaːp
that appears in
kaapte-
‘sprout’. In both cases the exact pathway of grammaticalization is unclear. Since both seem to
have been verbal roots, their origin as preverbs may stem from a serial verb construction, but
such an analysis is clearly speculative.
Finally, the preverb
ni
has its origin in a noun meaning ‘thing’, and this gave rise to both its
detransitivizing function as
ni
became incorporated into the verb (4)–(5), and its nominalizing
use in lexicalized compounds (6)–(7).
1
In referencing examples from the corpus, I follow Swadesh’s organization system of the speaker label (A =
Ben Paul), text number, paragraph letter, and sentence letter. Thus A02c.5 refers to a text dictated by Ben Paul,
which is the second text in that collection, in the third paragraph, fifth sentence.
6
(4)
ʔun kun ʔutʃiːkˀ ni kˀuʃtʃuji
some thing making DTRZR he.would.eat
‘making [himself] something, he would eat’ [intransitive use] (A07a.9)
(5)
kipi ʃamaʃ kˀuʃmpujna
meat fresh they.usually.ate
‘they ate fresh meat’ (A74f.1)
(6)
ni naki dempi
thing chicken.hawk killing
‘the story’ [< idiomatic
ni naki dempa-
‘kill chicken hawks’] (A05e.11)
(7)
tʃˀah ni ʃa ʔoːniʃ
bird thing voice many
‘mockingbird’ [lit. ‘many-voiced bird’]
Despite their diverse origins, there are a number of reasons for treating these eight words as
a unified preverb class, and not as verbal prefixes or types of adverbs:
Chitimacha preverbs share similar phonotactics: all are monosyllabic with short
vowels except for
ka:pˀs
.
While the preverbs are always prosodically part of the verb that follows, they do not
become phonetically attached, except in some possible historical instances of
ni
as
will be seen below. So while in other places /ʃ##hV/ → /ʃV/, one never sees this
pattern with
ʔapʃ
.
Preverbs have in common a directional sense as one of their meanings, suggesting a
semantic basis to the category.
There are structural and semantic patterns within the class. Consider the following
pairs, selectively arranged so as to highlight their similarities:
hi
‘to there’ :
his
‘returning’;
ʔap
‘to here’:
ʔapʃ
‘returning here’;
kap
‘up’:
kaːpˀs
‘(returning) back
up’. The general pattern is DIRECTION : RETURNING DIRECTION, with a final /s/ or /ʃ/
conveying the meaning RETURNING.
2
2
This is a simplification of the data.
hi
‘to there’ also corresponds semantically to
kas
‘returning there’ in
terms of the DIRECTION : RETURNING DIRECTION analogy.
kas
occurs with over twice the frequency of
his
in my
corpus (291 occurrences to 114), suggesting that
hi : his
is a less salient pair than
hi : kas
. The two forms may be
in competition over which is most analogous to
ʔapʃ
in the
ʔap : ʔapʃ
pair.
his
is the most structurally analogous,
but
kas
is the most semantically analogous. The frequency data suggest that
his
is the newer form, and may be
slowly assuming the meaning of
kas
.
7
With a few exceptions, the preverbs always occur in the same syntactic slot and are
mutually exclusive. Swadesh (1939:147–148; 1946:329–330) even describes a set of
rules which he calls “preverb displacement” that determine which of two preverbs
will appear when more than one would be expected. Use of the preverbs in places
other than the preverbal slot is not productive, but is restricted to a small, specific
set of lexemes. These facts suggest a class which is somewhat paradigmatic and thus
unified, and too restricted in its distribution to be considered adverbs.
With the exceptions of
ʔap
and
hi
, Swadesh says that the preverbs frequently form
“an essential part of the verbal lexeme” (1939:147), by which he means they are
semantically non-compositional in many (but not all) cases. This comports well with
the shared definition of preverbs as displaying a range of more-or-less tightly bound
uses, from the very lexical
ʔap
and
hi
to the non-compositional uses that Swadesh
mentions.
There is no distinct class of adverbs in Chitimacha to lump the preverbs into, if one
were so inclined. Almost without exception, adverbial words are minimally
bimorphemic, synchronically analyzable, and do not occur in the same syntactic
slots as preverbs. Adverbials occur only clause-initially or postverbally, and
frequently co-occur with the preverbs, often with a direct object intervening between
them.
Having given evidence of how these words constitute a sufficiently homogenous unit of
study, I now turn to their differences, which constitute the
explananda
for this research.
4. Degrees and dimensions of grammaticalization
The first and most notable difference among preverbs is their absolute frequency of
occurrence and overall frequency rank, shown in Table 3.
While Swadesh only gives ‘up’ as the definition for
kaːpˀs
, all six of its occurrences in the texts convey a
strong sense of ‘back up’, i.e. returning to the up position: ‘rise up (from the dead)’ (A11c), ‘rise up from his bed
again’ (A16c), ‘as he got back up’ (A69c) (references are to text numbers and paragraphs in Swadesh 1949b). This
lends further support to the DIRECTION : RETURNING DIRECTION pattern among the preverbs.
ni
does not participate in this analogy for reasons discussed further on.
8
Table 3. Absolute frequency and rank of preverbs in corpus
hi
1298
33.50%
#2
kap
775
20.00%
#3
ni
646
16.67%
#5
ʔapʃ
462
11.92%
#8
ʔap
335
8.65%
#14
kas
279
7.20%
#17
his
74
1.91%
#51
kaːpˀs
6
0.15%
#534
If it is true that “the more frequently a form occurs in texts, the more grammatical it is assumed
to be” (Hopper and Traugott 2003), then the preverbs at the top of Table 3 must be highly
grammaticalized while those at the bottom are still significantly lexical. This is precisely what
we find. The most frequent preverbs,
hi
,
kap
,
ni
, and
ʔapʃ
, each have specific grammatical
functions on top of their lexical, directional ones, while the less frequent preverbs do not (cf.
Table 2). Along just the frequency dimension alone, then, certain preverbs are found to be
much more highly grammaticalized than others.
Related to frequency differences are differences in productivity. One way to measure this is
by the number of verbs that each preverb can occur with, since the more productive a preverb
is across lexemes, the more grammaticalized it has become. This was done using a
representative sample, and the result of this count is shown in Table 4.
3
Table 4. Productivity of preverbs
ni
195
25.42%
hi
190
24.77%
kap
154
20.08%
ʔapʃ
82
10.69%
kas
67
8.74%
ʔap
57
7.43%
his
20
2.61%
kaːpˀs
2
0.26%
3
This count was obtained by using a subset of all the verbs in the corpus. Because the texts were not yet fully
glossed, and so information about lexical category was unavailable, regular expressions were used to search for
affixes known to be unique to verbs, resulting in the 2,844 results on which the count was based. It was then easy
to sort by stem and combine equivalent verbs.
9
Table 4 shows a pattern similar to the frequency counts in Table 1, with one noticeable
difference: the preverb
ni
appears with a significantly greater proportion of verb roots than its
overall frequency would suggest. While
ni
is only the third most frequent preverb overall, it is
by far the most productive, occurring with more verb roots than any other. Given its use as a
detransitivizer, this is unsurprising. While
ni
is somewhat less grammaticalized in terms of its
overall frequency, it should be considered the most grammaticalized in terms of its
productivity. The two dimensions of grammaticalization yield different results.
A third dimension of grammaticalization is the relative syntactic independence of each
preverb, or the extent to which they may participate in tmesis. More tightly bound and thus
grammaticalized preverbs will not allow for tmesis, while freer preverbs will. In Chitimacha,
noun phrases only occur between the preverb and the verb in certain highly lexicalized and
invariable instances. The resulting timetic constructions are likely fossilized reflexes of an
earlier stage of the language where preverbs were syntactically independent from the verb.
Examples are given in (8)–(9).
(8)
hus waʃi kiːtstiʃ we piji ʃih ki hi nam tʃˀah-t-ˀiʃ-i
her thumb those cane joints LOC to brand hew-TRZR-CONT-NFsg
‘Her thumb print is embossed in those cane joints.’ (A13e.2)
(9)
hi kaji ʔutʃi-tʃuj
to life do-FUT(sg)
‘you will live’ (A24a.9)
In (8), the verb takes a historical *
-t
transitivizing suffix and the nominal object intervenes
between
hi
and the verb. At one time the phrase literally meant ‘to hew a brand (as in a
distinctive mark) into’, but now it simply means ‘to emboss’. Since this is taken as a lexical
unit, the preverb
hi
now occurs before
nam
, the historical object. The same is true of the phrase
hi
kaji ʔutʃi-
‘to live’, literally ‘to make/do a life’.
Only
hi
,
ni
, and
ʔapʃ
, and
kas
participate in timetic constructions. Timetic
hi
occurs most
frequently (n=23), then
ʔapʃ
(n=4) and
ni
(n=2), and finally one occurrence of
kas
. Examples
of each are given below:
(10)
hi nej nutʃminaʔa
there land they.worked
‘they buried him’ (A35a.2)
10
(11)
ʔapʃ ni nektmaːʃnaʔa
together DTRZR they.leagued
‘they leagued together (as with an animal spirit guide)’ (A67a.11)
(12)
ni katʃ hamtʃiːkˀ
thing fortune having
‘having (good) luck’ (A10k.2)
(13)
kas panʃ teykˀʃ
back person turning
‘turning back into a person’ (A35d.15)
While other metrics for assessing the grammaticalization of
hi
have so far shown it to be one of
the most grammaticalized of the preverbs, it is actually the least grammaticalized in terms of its
syntactic independence. It is the most adverb-like of the preverbs in this respect, and Swadesh
often translates it with the English adverb ‘there’.
At the opposite end on the scale of syntactic independence is the preverb
ni
, which shows
evidence of having been historically incorporated into a number of verb stems, as the verb pairs
in Table 5 illustrate.
Table 5. Verbs with incorporated preverb
ni
Verb
Unincorporated form
naːkʃte-, naːkʃt- ‘write’
haːkʃt-, haːkʃte- ‘write, draw’
napˀe- ‘raw’, napʃtʃˀe- ‘be black’
hapt ‘vermillion’, *hapʃ ‘black’
nakst- ‘sell’
ʔaks-, ʔakst- ‘buy’
naʃi-, naʃma- ‘hunt’
haʃi-, haxma- ‘stalk, hunt’
neːmi- ‘be scared’
ʔem- ‘fear’
netʃˀin ‘temple’
hetʃˀi- ‘be holy’
nekt- ‘skin, peel s.t.’
haki- ‘peel s.t.’
nitˀi-, nijaʔ- ‘believe’
(both forms have
ni
in common)
nih- ‘turn over’
ʔij- ‘turn’, ʔiht- ‘circle’
niːnʃt- ‘turn upwards’
ʔiːnʃt- ‘upturn s.t.’
niki ‘real’
ʔiki- ‘hide’
nokt- ‘permit, release’
hokt- ‘leave, permit, release’
nokun ‘shoulder’, mokun ‘knee’
ʔokun ‘shoulder’
nowa ‘hominy’
huwo ‘crop’
nuki- ‘pray’
huːka ‘prayer’
11
No other preverb shows signs of historical or synchronic incorporation, and so in this respect
ni
is much more grammaticalized than its preverbal counterparts.
ni
is also the only preverb which
may be preceded by a further preverb, suggesting that it is more tightly bound to the verb stem,
though this happens only five times in the corpus, three of which occur in the same paragraph.
An example is given in (14).
(14)
wetk hus wa:kˀipaʃ ʔapʃ ni hejʃmiːkˀ
then his pillow together DTRZR gathering
‘then, gathering up his pillow’ (A26h.5)
Another useful metric for assessing degree of grammaticalization is the extent to which a
given preverb + verb combination has become semantically noncompositional (lexicalized).
Swadesh believes this phenomenon to be quite frequent:
The presence of preverbs as fixed parts of verbal lexemes comes into conflict with the free
use of preverbs in their literal senses, for the rule is that two preverbs may not precede the same
verb stem. Verbs with bound preverbs either resist the use of free preverbs or are subject to
what might be called preverb displacement; the free preverb is used and the bound one omitted.
(Swadesh 1939:147–148)
Semantic compositionality is of course a difficult and subjective criterion to assess. Still, one
way to approximate the number of semantically non-compositional forms is to see how many
preverb + verb combinations appear in the Chitimacha lexicon that is based on these texts.
Swadesh only listed preverb + verb combinations when their meaning was (in his view) not
predictable based on the meaning of their composite elements. While is this certainly a
subjective criterion, subjective does not mean arbitrary, and Swadesh’s intuitions on these
matters still provide us with potentially useful insights in the aggregate. Table 6 shows the
number of preverb + verb entries in the lexicon for each preverb.
Table 6. Number of semantically non-compositional uses of each preverb
kap
183
ʔapʃ
127
ni
114
kas
64
hi
30
his
15
kaːpˀs
3
ʔap
2
12
The resulting count is drastically different from the overall counts for preverb frequencies. If
more semantically integrated and non-compositional forms are taken to be indicative of
preverbs further along the path of grammaticalization, then
hi
is a significantly less
grammaticalized form (despite having the greatest overall frequency), and
ʔapʃ
is a significantly
more grammaticalized form (whereas it is only middling in terms of overall frequency).
The final dimension of grammaticalization to be examined here is the co-occurrence of
competing forms with the same meaning as the preverbs, which is suggestive of renewal.
Chitimacha has a nominal marker
=(n)k
which among other uses functions as an adverbial ‘to’
or ‘at’:
(15)
we=nk ne tˀuti kaːhan
DEM=LOC even go(pl) cannot
‘they can’t [even] go there’ (A28d.7)
(16)
wetk we ʔaʃintʃˀatˀaʃ hus hiʔi=nk kas tʃuji.
then DEM.DET old.man his be(NFsg)=LOC back he.went
‘[Then] the old man went back home.’ (A11b.2)
(17)
ni naki dempi kuti=nk daatk
NZR chicken.hawk killing end=LOC now
‘(We have reached) the end of the story now.’ [‘to tell a story’ is literally ‘to kill
chicken hawks’] (A05e.11)
(18)
pakta=nk pekup hi tʃujiʔi
sky=LOC upward to he.went
‘he went away up to the sky’ (A09a.3)
It also appears on almost every noun phrase that serves as the object of a verb with
hi
or
ʔap
, and also with directional postpositions like
hup
‘to, towards’.
(19)
we ʃeːni=nk hi nitʃwinkiʃ
the pond=LOC to when.he.came.to.water
‘when he got to the edge of the pond’ (A01a.3)
(20)
we ʃeːni waʔa=nk hi peʃiʔi
the pond other=LOC there he.flew
‘he flew toward the opposite side of the pond’ (A01c.1)
13
(21)
namu=nk namu=nk hi tʃuːʃiʔi
town=LOC town=LOC to he.goes
‘He goes from town to town.’ (A07a.14)
(22)
ha=nk ʔap nemnaʔa
this=LOC here they.crossed.water
‘they crossed over [water] to here’ (A02c.1)
(23)
wetk we siksink hiʃ we ʔasink hup hi neːtʃˀimi.
then the eagle ERG the man=LOC to to he.said
‘[Then] the eagle spoke to the man.’ (A02b.2)
I interpret this to be a case of renewal, where
=(n)k
is taking over the former function of
directional preverbs like
hi
and
ʔap
. Repeating this process for each of the preverbs gives the
results in Table 7.
4
It shows the frequency of co-occurrence of the two forms as a percentage of
the total uses of each preverb.
Table 7. Co-occurrence of preverbs with locative
=(n)k
kas
21.15%
hi
11.02%
his
10.81%
ʔap
6.87%
ni
6.19%
kap
5.29%
ʔapʃ
3.90%
kaːpˀs
0.00%
4
Only cases in which
=(n)k
immediately preceded the preverb are included in this table.
14
=(n)k
other
kas
28
107
135
other
72
100
You can use the above to get the expected, actual, and observed frequencies
According to this table, 21.15% of the time
kas
occurs in the corpus, it co-occurs with
=(n)k
.
This metric is most useful for preverbs with primarily directional meanings, namely
hi
,
ʔap
,
kas
,
his
, and
kaːpˀs
. Along this dimension of grammaticalization, then,
kas
is the most
grammaticalized since it shows the most evidence of functional renewal.
5. Summary & Conclusions
What makes a canonical case of grammaticalization? Section 4 examined six features that
contribute to the process:
Absolute frequency of occurrence (frequency)
Productivity with different verb roots (actuation)
Degree of syntactic independence (subdivided into tmesis and incorporation;
syntactic integration)
Semantic non-compositionality (lexicalization)
Co-occurrence with competing forms (renewal)
To this we can also add phonetic reduction, which was not examined here.
It was seen that preverbs differ, occasionally drastically, in the extent to which they have
become grammaticalized along each of these dimensions. Table 8 compares the ordinal ranking
of the preverbs, from most grammaticalized at the top to least grammaticalized at the bottom,
in terms of each of the six criteria.
15
Table 8. Comparison of relative degrees of grammaticalization
Frequency
Productivity
Tmesis
Incorporation
Compositionality
Renewal
More grammaticalized
Less grammaticalized
hi
ni
other
ni
kap
kas
kap
hi
other
ʔapʃ
hi
ni
kap
ni
his
ʔapʃ
ʔapʃ
kas
ʔap
ʔap
kas
kas
hi
ni
kas
ʔap
ni
his
kap
his
his
ʔapʃ
kaːpˀs
ʔapʃ
kaːpˀs
kaːpˀs
hi
ʔap
kaːpˀs
Given the nature of grammaticalization as a composite phenomenon, these differences
should not be surprising. But they do caution against relying on just frequency data as a
representative measure of how far grammaticalized a form is. Though grammaticalized forms
do tend to continue increasing in frequency long after the grammaticalization process is
complete (or at least undisputably recognizable; Mair 2011:245), grammaticalization theory
does not say this must be the case. A fully grammaticalized form could simply fall out of use or
come to be replaced.
In fact, none of these features are sufficient for grammaticalization in themselves, and it
may be that no one of these features is even necessary. And yet there are some general trends
to the data.
hi
and
ni
consistently rank among the top three most grammaticalized forms, and
kaːpˀs
among the least, creating a rough cline of more-to-less grammaticalized preverbs. What
this shows is that it is still meaningful to speak of general degrees of grammaticalization, as
long as it is appreciated that such statements can only be made in the aggregate, and always
risk conflating important differences. To say that
kas
is a generally less-grammaticalized form
may be true in the aggregate, but ignores the highly grammaticalized nature of
kas
when
viewed from the perspective of functional renewal in process.
What can we conclude from these seemingly disparate sets of data? Grammaticalization is
not a monolithic process proceeding along a single dimension. Constructions grammaticalize in
different ways to varying degrees, as we have seen for Chitimacha preverbs. It is not
unidirectionality that defines grammaticaliztion, but
unity in change
. One or two of these
16
changes alone is not sufficient to constitute a case of grammaticalization. It is only when
enough of these processes converge that we recognize it as a canonical case of
grammaticalization. The multidimensional approach presented here has revealed that the
apparent homogeneity of a small category of words is the result of a convergence of behaviors
among forms that are actually quite diverse, each showing reflexes of its particular pathway of
grammaticalization and persistence in its original meaning. This fortuitous alignment of
independent phenomena is what gives rise to the larger phenomenon we call
grammaticalization.
17
References
Brown, Cecil H., Søren Wichmann, and David Beck. 2014. Chitimacha: A Mesoamerican
language in the Lower Mississippi Valley.
International Journal of American Linguistics
80.
Forgács, Tamás. 2004. Grammaticalisation and preverbs.
Acta Linguistica Hungarica
51.45–84.
Hieber, Daniel W. 2013.
A dictionary of Chitimacha (Scholar’s Edition)
. Charenton, LA:
Chitimacha Tribe of Louisiana.
Hopper, Paul J., and Elizabeth Closs Traugott. 2003.
Grammaticalization
. 2nd ed. Cambridge
Textbooks in Linguistics. Cambridge: Cambridge University Press.
Iannucci, David J. 2009. Aspects of Chitimacha phonology. University of Utah.
Lehmann, Christian. 2002.
Thoughts on grammaticalization
. Oxford: Oxford University Press.
Mair, Christian. 2011. Grammaticalization and corpus linguistics.
The Oxford handbook of
grammaticalization
.
Mithun, Marianne. Linguistic areas of Native North America.
Handbook of areal linguistics
, ed.
by Raymond Hickey. Cambridge: Cambridge University Press.
Norde, Muriel. 2009.
Degrammaticalization
. Oxford: Oxford University Press.
Summer Institute of Linguistics. 2013. Fieldworks Language Explorer.
Swadesh, Morris. 1939. Chitimacha grammar (Copy 1).
Chitimacha grammar, texts and
vocabulary
. (American Council of Learned Societies Committee on Native American
Languages, Franz Boas Collection of Materials for American Linguistics, Mss.497.3.B63c
G6.5) Mss.497.3.B63c G6.5. Philadelphia, PA: American Philosophical Society.
--- 1946. Chitimacha.
Linguistic structures of Native America
, ed. by Harry Hoijer, 312–336.
(Viking Fund Publications in Anthropology 6) 6. New York: Viking.
--- 1953. Chitimacha texts. Philadelphia, PA.
18