The Application of Distributed Morphology to the Lithuanian First Accent Noun Declension Paradigm
Click here to load reader
-
Upload
adrian-lin -
Category
Education
-
view
309 -
download
1
Transcript of The Application of Distributed Morphology to the Lithuanian First Accent Noun Declension Paradigm
Ling404: Morphological Theory Adrian Lin
1
The Application of Distributed Morphology to
the Lithuanian First Accent Noun Declension Paradigm
Distributed Morphology – Introduction and concepts:
Distributed Morphology (hereafter DM) as a morphological theory is based on three main concepts:
Late Insertion, Underspecification, and Syntactical Hierarchical Structure All the Way Down (Noyer).
Late Insertion envisions the morpheme as abstract bundles of semantic features that are realized with the
help of vocabulary item insertion. The vocabulary item insertion gives the morpheme a phonological
expression after the structure has been built syntactically. In DM, full specification of vocabulary items is
not needed. Instead, their phonological exponents can apply to many morphemes on words through
Underspecification, which often accounts for observed syncretisms1. Underspecification is also used
analogously in other linguistic fields such as phonology as the basis of generalized models of linguistic
phenomena. Finally, syntactical structures in DM are seen as extending beyond the sentential realm and
into the morphological realm, i.e the syntactical structure extends all the way down. As such, DM
assumes that processes and representations (such as trees consisting of individual “pieces”) in syntax can
and do often affect and are used in morphology.
This contrasts with other theories of morphology that often view morphology and syntax as operating
with different rules, such as the Lexical Integrity Hypothesis, thus suggesting a theory of theoretical
linguistic representation that is less unified than that of DM’, which unifies various fields of theoretical
linguistics with common processes. This paper will attempt to apply DM to a paradigm of words and
explain the various morphological phenomena observed in the paradigm.
1 A syncretism is “the situation where one morphological form corresponds to two or more morphosyntactic descriptions.” (Glottopedia)
Ling404: Morphological Theory Adrian Lin
2
The Lithuanian First Accent Noun Declension Paradigm:
I chose Lithuanian and extracted a segment of the first accent noun declension paradigm data from the
University of Texas at Austin’s Linguistics Research Center site. As a Balto-Slavic language, it has
fusional2 rather than agglutinative3 morphology. Therefore it has a complex case system with many
vocabulary items and syncretisms. The paradigm and its syncretisms are marked in (1). As the website
does not explain the bracketed parts, I included these as a part of the word. I ignored the consistent first-
syllable stress marks and wrote ą and ų as aː and uː, complying with the pronunciation.
(1) "Leg" “Needle” “Peel”
Singular Plural Singular Plural Singular Plural
Nominative koja kojos adata adatos lupena lupenos
Genitive kojos kojuː adatos adatuː lupenos lupemuː
Dative kojai kojom(s) adatai adatom(s) lupenai lupenom(s)
Accusative kojaː kojas adataː adatas lupenaː lupenas
Instrumental koja kojom(is) adata adatom(is) lupena lupenom(is)
Locative kojoj(e) kojose adatoj(e) adatose lupenoj(e) lupenose
Vocative koja kojos adata adatos lupena lupenos
Observed Syncretism:
At first glance, it appears the syncretisms in this paradigm can be summarized as in (2).
(2) Nominative singular = Instrumental singular = Vocative singular
Nominative plural = Genitive singular = Vocative plural
However, there appears to be several recurring elements, such as -o-, -m-, and -s in certain words. A
proposed segmentation scheme will be discussed in (4), (5), and (8) below.
2 In fusional languages, a single morpheme may contain many semantic features and grammatical functions. 3 In agglutinative languages, morphemes are more numerous but each contain fewer semantic features.
Ling404: Morphological Theory Adrian Lin
3
Posited Features:
I used semantic features for classification because they are shown to correlate with morphological
classification (Comrie, 1978). Frank (1995) provides a useful framework for case classification by
semantic features4 where he adapts Jakobson’s (1936) system of case system feature classification that
used three features: [± quantified], [±directional], and [±marginal]. Frank refers to Jakobson’s definition
of these ‘“semantic marks” as how significant the role of the entity in the message is, ““signalizing the
goal of an event,” and “assigning to the entity an accessory place in the message,” respectively’. Frank’s
addition and adaptation of the system results in:[±oblique], whether a noun can be the object of a verb or
preposition; [±marginal], as above; [±nonascriptive], a version of Jakobson’s (1958) ascriptiveness,
which Worth (1984) describes as “[ascribing] to the object a quality or state resulting from action directed
at the object;” and [±phrasal],which Frank describes as being or not being “associated with (not
necessarily maximal) phrasal projections.” In addition, [±singular] is also a part of this system.
I applied this system to Lithuanian because it is related to Russian and both language have many cases
suggesting shared features and classification. However, Jakobson and Frank list two genitives and two
locatives; whereas in the Lithuanian paradigm, it is unclear which would apply. Therefore, during the
course of the analysis, I tested both out to see which fit better. In addition, my Lithuanian paradigm
contained one extra case that was unaccounted for in the book – the Vocative. I eventually settled on the
feature matrices for the Lithuanian locative, genitive, and vocative exhibited along with others in (3).
4 Semantic features are qualities that indicate category membership and enable systematic linguistic classification.
Nom Gen Dat Acc Inst Loc Voc
± phrasal + - + - + - -
± nonascriptive + - - - + + +
± marginal - - + - + + -
± oblique - + + - + + -
(3)
Ling404: Morphological Theory Adrian Lin
4
€
ROOT − agr1[±sg]− agr2
±phrasal±nonascriptive±marg inal±oblique
However, because no underspecified set of these semantic features can exclusively account for the
syncretism pattern in (2) without referring to other cases, I conclude that there must be several
morphemes, as will be discussed below.
Posited Structure:
Because Lithuanian is a fusional language, I assumed fewer pieces in the morphosyntactical tree (4).
Part of my reasoning for having two agreement morphemes, despite the fusional nature of this language,
stemmed from my observation that certain post-root vowel distributions of the declensions were
correlated with plurality. As visible in (5), which shows the overall pattern I observed in the paradigm,
[-round] vowels (in blue) tended to be found in this position in the singular, while [+round] vowels (in
green) were found in the plural. This suggested two vocabulary items for this morpheme. I therefore
envisioned agr[+sg] to correspond to -a and agr[-sg] to correspond to -o and perhaps some
impoverishment rules5 to account for the appearance of unexpected morpheme.
5 An impoverishment rule is a rule that eliminates a feature altogether, allowing a less specified feature to surface.
(5)
Singular Plural
Nominative √ROOT-a-Ø √ROOT-o-s
Genitive √ROOT-o-s either √ROOTj-uː-Ø or √ROOT-u-u
Dative √ROOT-a-i √ROOT-o-m(s)
Accusative either √ROOT-aː-Ø or √ROOT-a-a √ROOT-a-s
Instrumental √ROOT-a-Ø √ROOT-o-m(is)
Locative √ROOT-o-j(e) √ROOT-o-se
Vocative √ROOT-a-Ø √ROOT-o-s
(4)
Ling404: Morphological Theory Adrian Lin
5
!"#$%&%'()*+,-*.*$"/0.(1-2*+3(
(45+"0#(!"#(
(6(
Me
tho
do
log
y:
Th
e In
th
e ch
art
in (
6),
th
e tw
o g
rey
sec
tio
n h
ead
ing
s la
bel
ed A
gr1
[±sg
] an
d A
gr2
[±p
hr
±n
asc
r ±
ma
rg ±
ob
l] r
epre
sen
t th
e tw
o m
orp
hem
es t
hat
are
atta
ched
to
th
e !
RO
OT
(h
ere,
!K
OJ)
. T
he
vo
cab
ula
ry i
tem
s ar
e li
sted
in
ord
er o
f sp
ecif
icit
y w
ith
th
e m
ost
sp
ecif
ic t
o l
east
fro
m t
op
to
bo
tto
m.
In
DM
, m
ore
sp
ecif
ic v
oca
bu
lary
ite
ms
pre
ven
t th
e p
ho
no
log
ical
ex
po
nen
ts o
f o
ther
les
s-sp
ecif
ic v
oca
bu
lary
ite
ms
fro
m b
ein
g g
ener
ated
du
e to
th
e
assu
mp
tio
n t
hat
eac
h s
yn
tact
ic t
erm
inal
no
de
can
ho
ld o
ne
ph
on
olo
gic
al e
xp
on
ent
and
th
at m
ore
sp
ecif
ic i
tem
s ar
e ap
pli
ed f
irst
. In
ch
art
(5)
bel
ow
,
the
emb
old
ened
rep
rese
nt
the
resp
ecti
ve
ph
on
olo
gic
al f
orm
s re
aliz
ed t
hro
ug
h t
his
pri
nci
ple
of
spec
ific
ity
. H
ow
ever
, fo
rms
in r
ed w
ith
cu
rly
bra
cket
s { }
rep
rese
nt
the
form
s th
at w
ere
no
t se
lect
ed,
des
pit
e b
ein
g m
ore
sp
ecif
ic,
du
e to
im
po
ver
ish
men
t ru
les,
wh
ich
are
lis
ted
un
der
nea
th.
Fin
ally
, se
ctio
ns
lab
eled
Sp
ell-
ou
t re
sult
bef
ore
ph
on
olo
gy a
nd
Ph
on
olo
gic
al
rule
s sh
ow
th
e p
roce
ss t
hat
cre
ates
th
e o
bse
rvab
le S
urf
ace
fo
rms.
(6)
Ling404: Morphological Theory Adrian Lin
6
Vocabulary Item List:
From my analysis, I identified the following vocabulary items which the chart (7) below highlights. Many
of these vocabulary items are conditional, depending on the presence and form of adjacent morphemes.
Resulting chart and segmentation:
This analysis results in the following segmentation chart (8).
(8) "Leg" “Needle” “Peel”
Singular Plural Singular Plural Singular Plural
Nominative √KOJ-a-Ø √KOJ-o-s √ADAT-a-Ø √ADAT-o-s √LUPEN-a-Ø √LUPEN-o-s
Genitive √KOJ-o-s √KOJ-Ø-uː √ADAT-o-s √ADAT-Ø-uː √LUPEN-o-s √LUPEN-Ø-uː
Dative √KOJ-a-i √KOJ-om-s √ADAT-a-i √ADAT-om-s √LUPEN-a-i √LUPEN-om-s
Accusative √KOJ-Ø-aː √KOJ-a-s √ADAT-Ø-aː √ADAT-a-s √LUPEN-Ø-aː √LUPEN-a-s
Instrumental √KOJ-a-Ø √KOJ-om-is √ADAT-a-Ø √ADAT-om-is √LUPEN-a-Ø √LUPEN-om-is
Locative √KOJ-o-je √KOJ-o-se √ADAT-o-je √ADAT-o-se √LUPEN-o-je √LUPEN-o-se
Vocative √KOJ-a-Ø √KOJ-o-s √ADAT-a-Ø √ADAT-o-s √LUPEN-a-Ø √LUPEN-o-s
Agr1[±sg]
[- sg] <-> -om / [__][+phr +mar]
[+sg] <-> -a
[ ] <-> -o
Agr2[±phr ±nascr ±mar ±obl]
[+phr +nascr +mar] <-> -is / [-sg][__]
[-phr +nascr +mar] <-> -je / [+sg][__]
[-phr +nascr +mar] <-> -se
[-phr -nascr] <-> -as / [-sg] [__]
[-marg +obl] <-> -uː / [-sg] [__]
[+phr -nascr]<-> -i / [+sg] [__]
[-phr -nascr] <-> -aː
[+nascr] <-> -Ø
[ ] <-> -s
(7)
Ling404: Morphological Theory Adrian Lin
7
Problems:
The phonological exponents of Agr1[±sg], -om and -o, presented a possible problem due to the similarity
in distribution – both are associated with the feature [-sg] and the presence of /m/ in both is the only
differentiator between having both be -o. Attaching the /m/ to Agr2[±phr ±nascr ±mar ±obl] requires the
creation of two new vocabulary items with the phonological exponent of /-ms/ and /-mis/. Also, another
problem arises with the ability of agr1[±sg] to reference other morphemes, because it is the first in the
sequence to be constructed. Normally, it is expected and empirically observed that in a sequence √ROOT-
X-Y-Z, morpheme Z can reference Y and X and morpheme Y can reference X, but morpheme X, being
the first to be built, can not reference any subsequently-built morphemes.
Possible Solutions:
If we posit a mystery morpheme X[±F] in a revised structure as in (9), then a modified set of vocabulary
items for the morphemes Agr1[±sg] and X[±F] in (10) emerges. Thus, [-sg] <-> -om/[__][+phr +mar] can
be removed unifying the exponents expressing [-sg]. This segmentation is expressed in chart (11).
Agr1[±sg]
[+sg] <-> -a
[ ] <-> -o
X[±F]
X[ ] <-> -m / agr1[-sg]X[__]agr2[+phr +mar]
X[ ] <-> Ø
(9)
(10) €
ROOT − agr1[±sg]− X[±F]− agr
±phrasal±nonascriptive±marg inal±oblique
Ling404: Morphological Theory Adrian Lin
8
(11) "Leg" “Needle” “Peel”
Singular Plural Singular Plural Singular Plural
Nominative √KOJ-a-Ø-Ø √KOJ-o-Ø-s √ADAT-a-Ø-Ø √ADAT-o-Ø-s √LUPEN-a-Ø-Ø √LUPEN-o-Ø-s
Genitive √KOJ-o-Ø-s √KOJ-Ø-Ø-uː √ADAT-o-Ø-s √ADAT-Ø-Ø-uː √LUPEN-o-Ø-s √LUPEN-Ø-Ø-uː
Dative √KOJ-a-Ø-i √KOJ-o-m-s √ADAT-a-Ø-i √ADAT-o-m-s √LUPEN-a-Ø-i √LUPEN-o-m-s
Accusative √KOJ-Ø-Ø-aː √KOJ-a-Ø-s √ADAT-Ø-Ø-aː √ADAT-a-Ø-s √LUPEN-Ø-Ø-aː √LUPEN-a-Ø-s
Instrumental √KOJ-a-Ø-Ø √KOJ-o-m-is √ADAT-a-Ø-Ø √ADAT-o-m-is √LUPEN-a-Ø-Ø √LUPEN-o-m-is
Locative √KOJ-o-Ø-je √KOJ-o-Ø-se √ADAT-o-Ø-je √ADAT-o-Ø-se √LUPEN-o-Ø-je √LUPEN-o-Ø-se
Vocative √KOJ-a-Ø-Ø √KOJ-o-Ø-s √ADAT-a-Ø-Ø √ADAT-o-Ø-s √LUPEN-a-Ø-Ø √LUPEN-o-Ø-s
Conclusion and Areas for Further Investigation:
Still, the mystery morpheme analysis presents its own problems. First of all, an extra morpheme requires
semantic features, which may require moving some from agr2[ ], but it is unclear which. Secondly, it is
somewhat suspicious that a morpheme would be realized as -Ø most of the time. Rather than three
morphemes with the exponents -m, -s, and -is, a third analysis may conceive two exponents, -mis, and
-ms. However this would not address the suspicious presence of /m/ in the two morphemes.
An area for further investigation could be extending and adapting this analysis to more noun declension
paradigms in Lithuanian, such as the entire first accent noun class and the second accent noun class as
listed on the University of Texas at Austin’s Linguistics Research Center site. Still, DM proves useful in
handling the syncretisms and similar elements of this paradigm through its vocabulary item
underspecification and order of specificity. In this manner, it unifies the system and conforms to linguistic
assumptions of storing as few mental representations as possible for the necessary meanings by creating
more generalized, underspecified pieces and processes.
Ling404: Morphological Theory Adrian Lin
9
Bibliography:
Comrie, B. (1978). Morphological Classification of Cases in the Slavonic Languages. The
Slavonic East European Review, 56(2), 177-190.
Franks, S. (1995). Parameters of Slavic Morphosyntax. New York, NY: Oxford University Press.
Syncretism (n.d.). Retrieved from Wiki: http://www.glottopedia.de/index.php/Syncretism
Noyer, R. (N.D.). Distributed Morphology: Frequently Asked Questions List. Retrieved, 10/12/10 from:
http://www.ling.upenn.edu/~rnoyer/dm/
Vasiliauskiene, V. & Slocum, J. (2009). Baltic Online: Lesson 3: Lithuanian. Retrieved, 17/12/10 from:
http://www.utexas.edu/cola/centers/lrc/eieol/litol-3-X.html
Worth, D. S. (1984). Russian Gen2, Loc2 Revisited. In van Baak, J.J. (Ed.), Signs of Friendship: To
Honour A.G.F. van Holk, Slavist, Linguist, Semiotician (295-306). Amsterdam, Netherlands: Rodopi B.V.