Date: Tue, 19 Aug 2003 14:05:03 -0400
From: Elizabeth J. Pyatt <ejp10@psu.edu>
Subject: Syntactic Effects of Morphological Change
Lightfoot, David W., ed. (2002) Syntactic Effects of Morphological
Change, Oxford University Press.
Elizabeth J. Pyatt, Penn State University
This volume is a collection of papers originally presented at the May
2000 Diachronic Generative Syntax Series (DIGS IV) held at the
University of Maryland. The overriding theme of this book is the
analysis of documented syntactic changes within modern
morpho-syntactic frameworks, usually Minimalism. Lightfoot has
divided the book into four sections - "Morphologically Driven
Changes," "Indirect Links Between Morphology and Syntax,"
"Independent Changes in Movement Operations" and "Computer
Simulations." Most of the languages covered are Western European,
primarily Germanic and romance, but there is one paper on Old
Japanese and another on Middle Welsh.
Overall, this is a very valuable resource for anyone interested in
the theoretical implications of diachronic syntax. Each article
presents the linguistic data and an extensive discussion of how
changes may affect our understanding syntactic theory and vice-versa.
Most papers are done within some version of the Minimalist framework,
but other possibilities are presented in some of the articles; for
the most part, the theoretical assumptions are well presented in each
article. Even if you disagree with a particular analysis, you may
find the questions raised important to consider.
Summaries of each of the articles are presented below.
The first article is the "Introduction" by David Lightfoot in which
he provides a broad overview of the importance of studying historical
changes in syntax then explains how the volume is divided into the
different subsections. Interestingly, Lightfoot inserts a note of
caution about how universally applicable a generalization might be.
For instance although there is widespread empirical support for the
proposed Null Subject Parameter (phonologically null subjects
licensed by rich verb agreement) and Bobaljik's (2000) Rich Agreement
Hypothesis (rich verb agreement licenses overt V to I movement),
counterexamples can be found for both principles. Later, he notes
that Bobaljik and Thrainsson (1998) propose that functional
categories may vary from one grammar to another. This theme is
repeated in the later articlres and adds an interesting dimension to
both historical reconstruction and syntactic theory.
Article 2 is "A History of the Future" by Ian Roberts and Anna
Roussou. This is the first article in the "Morphologically Driven
Changes" section in which authors propose that syntactic changes
result from changes in a languages morphology. Here Roberts and
Roussou compare the development of three future tenses - the English
"will" future, the Romance future tense which evolved from a reduced
form of Latin "habere" 'to have' and the Modern Greek "tha" future
particle which evolved from "thelo" 'want'. In all three cases, the
new monoclausal future tense evolved from a biclausal control
structure; in addition, the tense markers began as lexical verbs
which went through Verb to INFL (V to I) raising. Once these
auxiliaries became future tense markers, their category changed from
a lexical verb to that of a higher projection such as T (Tense).
Roberts and Roussou argue that the common reason for both changes was
the loss of a key morphological distinction, such as infinitival
morphology in English or Greek and a reduced forms of "habere" in the
specialized inifinitive+habere idiom in Romance. In this analysis,
the loss morphological distinctions meant that children acquiring
these language lost overt cues for biclausal control, forcing them to
reanalyze control structures as monoclausal sentences.
Article 3 is "Case and the Middle English Genitive Phrases" by
Cynthia L. Allen which is followed by response by Zeliko Boskovic. As
has been noted before, Old English had an overt genitive case
typically marked by the suffix -es which could appear on multiple
items in a Determinier Phrase (DP). By Modern English, this case
ending had evolved into the possessive clitic 's which attaches once
to entire possessive DP. Old English allowed split or "scrambled"
genitive DPs in which constituent elements were split, but all were
case marked (1a). However, Middle English developed an intermediate
stage in which the constituents were split, but only one of the items
were case marked for genitive (1b).
(1) a. on aethelred-es daege cyning-es (Old English)
in Ethelred-Gen day king-gen
'In King Ethelred's day.'
b. the king-es moder Henri (Middle English, ca 1325)
the king-gen mother Henry
'King Henry's mother'
Lightfoot (1999) proposed that split genitive were structured so that
the left genitive element was in Spec-DP of the head possessum DP,
and the right element was the complement of that possessum DP. To
account for the genitive morphology, Lightfoot assumes that the head
DP ("daege" 'day' in (1a)) assigns case and a theta role to both the
specifier position and the complement position. Lightfoot tied the
loss of overt genitive marking for the right element to the change of
genitive case to a clitic. That is, Lightfoot assumes that Middle
English had reanalyzed the genitive case ending as a possessive
clitic.
Allen, on the other hand, argues that Middle English still had
morphological genitive case and that change to the forms in (1b) is
due to another factor. One piece of evidence she proposes is that
some irregular genitives survive into Middle English and were used in
the split possessive for the left element only. If the genitive case
ending had been reanalyzed as a clitic, one would expect that all
irregular genitives would be lost and that only the regular clitic
form would be used. To Allen, this indicates that genitive case is
still present in Middle English grammar, but that it is no longer
used to mark the right-hand element. To account for this, Allen
proposes that that genitive case was present in Middle English, but
became optional for the right hand element. This is tied to the
breakdown of the Old English case system in general through
phonological mergers. In Allen's analysis, it is because genitive
case on the right-hand becomes optional that future generations
analyzed the genitive ending as a clitic, thus reversing Lightfoot's
chronology. This idea may be similar to the idea of "competing
gramamrs" proposed by several authors later in this volume.
In article four, "Split Constituents within NP (Noun Phrase) in the
History of English: Commentary on Allen" Zeliko Boskovic reviews the
data and provides a third account for both Old English and Middle
English. He analyzes both constructions as instances of scrambling,
similar to what occurs in Japanese. Following a previous article
Boskovic co-wrote with Takahashi (1998), he proposes that both
elements are base generated in their surface position, but that LF
movement covertly raises the right element so that it is in the Spec
DP with the left element. In this analysis, theta role assignment
happens once to the specifier during LF. Boskovic argues that the
raising in LF accounts for the patterns of semantic interpretation in
the split possessive constructions.
Article five is Eric Haeberli's "Inflection Morphology and the Loss
of Verb Second in English." In the first portion of this article
Haeberli points out that although Middle English had verb second (V2)
order, it was not the same as V2 order in other Germanic languages.
In both Modern English and Middle English, when an operator, such as
a Wh-word is raised to Spec-CP (Complementizer Phrase), verb
inversion is triggered. But in Middle English, non operators, such as
a DP, could be fronted and trigger verb inversion, hence V2 order.
However, if the subject was pronominal, even if a non-operator was
fronted, verb inversion did not occur; this is the so-called verb
third (V3) order. Around 1400 though, fronting with inversion was
lost and hence English "lost" V2 ordering. To account for this,
Haeberli first assumes that there was a distinction between
pronominal DPs which were raised to Spec-AgrSP (Agreement-Subject)
versus non-pronominal DP's which could remain in situ in a lower
position; these were licensed by a null expletive in Spec-AgrSP
(Haeberli cites independent evidence for the Middle English null
expletive in this article). For non-operator V2, the verb remains in
AgrS (unlike operator V2 where verbs raise to C) although the
operator raises to Spec-CP. If the subject is pronominal, it raises
to Spec-AgrSP before the verb (2a); when the subject is pronominal,
it remains in situ and the null expletive is in Spec-AgrSP (2b). A
fronted non-operator would come before either pronominal subject (V3)
or a bare verb (V2). Thus
(2) a. [CP [XP hiora untrymnesse] [AgrSP he [Agr sceal] dhrowian ]]
their-weakness he shall atone
'Their weakness, he shall atone.'
b. [CP [XP dhas gifte] [AgrSP pro-i [Agr sealde] seo ceasterwaru-i ]]
this gift gave the citizens
'This gift, gave the citizens'
For Haeberli then, when Middle English loses the null expletive, it
also loses V2 ordering because the loss of the null expletive forces
all DP's to raise to Spec-AgrSP. Effectively, all sentences become
"V3". In accordance to the Rich Agreement Hypothesis (Bobaljik 2000),
Haeberli further ties the loss of the null expletive to the loss of
distinctive verbal agreement. One interesting fact noted by Haeberli
is that West Flemish and Dutch have very similarly rich verbal
paradigms, yet West Flemish does not license null expletives, while
Dutch does. Haeberli and Haegeman (personal communication to
Haeberli) propose that the key may be that in Dutch, the infinitive
is a distinctive form while in West Flemish and English, the
infinitive merged with the first person singular. Haeberli proposes
that if a singular form merges with the infinitive, null expletives
can no longer be licensed. In a future paper, Haeberli will present
evidence that V3 order (no inversion) begins to predominate at the
stage when infinitival morphology is lost.
The next two articles deal with the Middle English transition from
use of an overt dative case ending to the use of the preposition "to"
to mark dative case as in Modern English. Article six, "The Rise of
the To Dative in Middle English" by Thomas McFadden analyzes the
Penn-Helsinki Parsed Corpus to track the development of this
construction. First, he shows that although the "to" dative appears
sporadically in the Middle English M1 period (1150-1250), it does not
gain a strong foothold until the M2 period (1250-1350). McFadden also
compares the possibility of word ordering for ditransitive sentences
at this time (e.g. "I gave Bill the book"). For texts which had no
"to" dative or very rare instances of it, texts showed both V DO IO
(that is verb, bare noun direct object, bare noun indirect object)
and V IO DO sentences, with the second variation slightly
predominating. Once the "to" dative becomes more common though, the
strong preference is to interpret a ditransitive sentence as V IO DO.
Based on these facts and binding facts, McFadden argues that in fact
the "to" is a continuation of the older V DO IO order, but that in
this case the indirect object is obligatorily marked with a
preposition because it cannot receive structural case from the verb
itself; this assumes that only DP's in the complement of the head
verb receive a structural case.
The next "to" dative article is Chiara Polo's "Double Objects and
Morphological Triggers for Syntactic Case." Contra to Weerman (1997)
who argues that word order of ditransitive sentences becomes fixed
when overt accusative and dative case marking is lost for full NPs,
Polo argues that the fix in word order does not happen until the
distinction is also lost in the pronominal system. Polo argues that
the use of overt dative within the pronominal system still gave
children overt cues for abstract dative case; thus case marking with
"to" was not needed in those grammars. Like McFadden. Polo uses data
from Middle English texts to show a correlation in chronology for the
loss of datives in the pronominal system and the large-scale use of
the "to" dative.
Articles eight an nine focus on Portuguese grammar. The article "Cue
Based Change: Inflection and Subjects in the History of Portuguese"
by Arcisio Pires discusses the origin of Portuguese inflected
infinitives (infinitival which take person number agreement to agree
with the subject of an embedded infinitival clause). Pires argues in
favor of deriving inflected infinitives from the Latin imperfect
subjunctive, versus deriving it from a bare infinitive. Specifically
Pires proposes that when embedded clauses with imperfect subjunctives
lost both past tense meaning and while at the same time, key
presetencial complementizers (specifically Latin "ut") were deleted,
the cues for mood and finiteness were lost leading to the reanalysis
of these subjunctives as infinitives with agreement morphology. The
second half of the article discusses the loss of inflected
infinitives in Brazilian Portuguese in favor of bare infinitives
only. Here, Pires ties this change to the simplification of the
Brazilian Portuguese tense system in which many "inflected
infinitives" are homophonous with bare infinitives.
In article nine, "Morphology and Null Subjects in Brazilian
Portuguese" by Cilene Rodrigues, Rodrigues traces the decline in the
use of Null Subjects or pro in Brazilian Portuguese as compared with
European Portuguese. Like Pires, Rodrigues correlates the decline in
Null Subject with the impoverishment of the Brazilian Portuguese verb
system. Although overt subjects are now obligatory in many cases in
Brazilian Portuguese, there are some constructions which use Null
Subjects; however the conditions on binding and coreference have
become much more rigid. For instance, a Null Subject in an embedded
clause must refer to a c-commanding antecedent; in other words, it
has become an anaphor. Similarly, in matrix clauses, a third person
Null Subject must receive an impersonal subject interpretation. To
account for both the decline in Null Subjects and the change of the
Null Subject from a [+pronoun] pro with free reference to a
[+anaphor] form with restricted reference, Rodrigues proposes that
the collapse of the verb morphology means that Agr has lost phi
features needed to license robust Null Subjects, but maintains the D
properties.
The subject of article ten is Old Japanese, the only article
discussing a non-Western language. Akira Watanabe's "Loss of Overt Wh
Movement in Old Japanese" focuses on the Old Japanese 'kakarisumbi'
phenomena, a system where the use of different classes of focus
particles in Japanese triggered changes in the verbal agreement
endings. In the earliest Nara period (8th century AD), the "ka" focus
particle actually marked constituents which underwent
Wh-Movement/Focus Movement; thus "ka"-phrases must proceed nominative
marked phrases in the Nara period. Therefore, Watanabe proposes that
kakarisumbi system is actually a form of Wh-Agreement. Unlike
European languages though, the Wh phrase did not raise to Spec CP,
but to a lower Spec-FocusP position; an additional higher Spec-TopicP
was available at this time. As Watanabe notes, overt Wh-movement is
marked in SOV (subject-object-verb) languages, so it would not be
surprising that this movement would be lost. In fact, Watanabe
proposes that this has already happened in Heian period (9th
century). In addition, the function of the "ka" particle changed from
a Wh/focus marker to a polarity marker of rhetorical questions during
this period. To explain these changes, Watanabe connects the loss of
Wh-Movement marked by "ka" to the increase in topicalization where
subject DPs are more likely to raise to Spec-TopicP. At that point
the word order cues that "ka" phrases phrase to proceed subject
phrases would be lost, and in situ Wh-phrases would evolve. Watanabe
also includes a discussion of how the details of this change support
the Chomsky (2000) proposal that the Wh feature is [-Interpretable]
and uses this analysis to explain the semantic evolution of "ka."
The last article in the "Morphologically Driven Changes" section is
article eleven, "Changed in Subject Case-Marking in Icelandic" by
Thorhallur Eythorsson. Both Old Icelandic and Modern Icelandic have
quirky case marking on subjects so that subjects can be marked with
either dative case or accusative case depending on the verb. However,
Icelandic has two phenomena, called "Nominative Sickness" and "Dative
Sickness", where oblique subjects may surface nominative or dative
respectively. In Dative Sickness, subjects which are supposed to be
accusative case in the prescriptive grammar become dative case
instead. By looking at Old Icelandic texts as well as a survey of
junior college students in Iceland, Eythorsson is able to distinguish
key properties of both Nominative Sickness and Dative Sickness.
Nominative Sickness has been present in Icelandic since the Old
Icelandic period, and Eythorsson analyzes this as an instance where
subjects in Spec-IP are receiving the default structural case of
nominative instead of the oblique case assigned by the verb. Dative
Sickness is different in several respects. First, Dative Sickness
specifically targets verbs whose theta role is that of experiencer;
therefore Eythorsson proposes that Dative Sickness is a
generalization of quirky case for the experiencer. Second, this
phenomenon is more recent than Nominative Sickness, and Eythorsson
reports that Dative Sickness is more stigmatized than Nominative
Sickness. Interestingly though, Eythorsson argues the rise of Dative
Sickness is reversing the trend of Nominative Sickness; that is,
instances of using nominative case to substitute oblique case is not
as acceptable or as common in Modern Icelandic. This article is
especially interesting because it shows a trend towards
"regularization" (i.e. Nominative Sickness) can exist in a language
for several centuries yet still not cause a full restructuring of the
grammar - Icelandic still maintains quirky subject marking.
Intriguingly, the effects of Nominative Sickness appears to be
declining and being overtaken by the more recent Dative Sickness. It
would be interesting to see whether accusative marked subjects remain
in Icelandic, are lost althogether in favor if the nominative or
dative or are generalized on some other theta role.
The second section of this book, "Indirect Links Between Morphology
and Syntax" contains just two articles
Article twelve is Dick Bury's "A Reinterpretation of the Loss of Verb
Second in Welsh". Unlike some Germanic V2 languages in which embedded
clauses may be SOV, Middle Welsh embedded clauses and negative
sentences are VSO (verb subject object). As Bury notes, V2 and VSO
are interesting "minimal pairs" because in both the inflected verb
proceeds the subject; the only difference is whether an XP must
proceed the verb as in V2 or there is just a functional words and a
bare inflected verb as in VSO. Bury follows a proposal of Neeleman
and Weerman (1999) and assumes that the specifier position for
fronted XPs in V2 languages are created by self-attached verbs and is
not the specifier of a higher category. Both Middle Welsh and Modern
Welsh share a number of affirmation and [+Wh] agreement particles
including affirmative particles "a" marking movement of fronted
subjects and objects and "yd" marking movement of fronted adverbs and
the negative particle "ny". To account for the distinction between
affirmative sentences which are V2 and negative sentences which are
VSO, Bury proposes that the affirmative particles are adjoined to the
highest self-attached V as focus agreement markers while the negative
particle "ny" heads its own functional NegP (Negative Phrase) thus
blocking VSO. In Middle Welsh, Bury proposes that the affirmative
focus agreement markers were reanalyzed as functional heads and that
full V2 was blocked. In particular, the subject pronouns "fe" 'he'
and "mi" 'I' were reanalyzed as affirmative markers located in their
own functional projection. Finally, Bury also proposes a general
correlation that VSO languages must have preverbal particles. This is
an interesting observation, but I do wonder whether that falls out of
word order; however, it may be the case that VSO languages may have a
richer set perverbal particles, such as overt affirmative markers
found in Modern Welsh and overt past tense markers found in Irish and
Middle Welsh.
Since I am familiar with some of the Middle Welsh data, I would like
to make additional comments to this article. First, V2 in Middle
Welsh is particularly anomalous because both Old Welsh and Literary
and Modern Welsh show VSO order - thus V2 order evolved in Middle
Welsh and was lost by Literary Welsh. One interesting aspect to
Bury's analysis is that he crucially assumes that the negative
particle "ny" occupies a distinct head position from the affirmative
particles which are adjoined to the verb. However, I was not able to
find any any external evidence cited by Bury that this was the case.
In fact, when considering the behavior of object clitics in Middle
Welsh, Evans (1964) reports that object clitics attach a preverbal
particle, either the negative particle or to the affirmative
particle. If the preverbal particles were in a different position,
some asymmetries might be expected. In fact, I am not aware of any
differences in word ordering between affirmative particles and
negative particles other than the V2 vs. VSO ordering. Although
Bury's analysis may be correct, it would be nice to have these facts
explained. In addition, an explanation for the original origin of
Middle Welsh V2 from a VSO language should also be explored in the
context of his analysis.
The thirteenth article is "The Loss of IP Scrambling in Portuguese:
Clause Structure, Word-Order Variation and Change" by Ana Maria
Martins. Old Portuguese had a number of OV structures in which direct
object proceeds the verb in embedded clauses. This contrasts with the
general SVO (subject-verb-object) order of Portuguese. By using a
series of diagnostic tests involving clitic placement, Martins
proposes that many instances of the OV ordering are due to clause
internal IP scrambling or "medial scrambling" where constituents are
fronted to multiple specifier positions.
The next section is "Independent Changes in Movement Operations." For
most articles I this section, the theme is not just changes
independent of morphology, but rather arguments that analyses arguing
a direct correlation in changes in morphology leading to changes in
syntactic structure, particularly movement operations, may not be
tenable. Several articles present data for corpus analysis and field
work showing that "archaic" structures persist even after the
proposed morphological cues have already been lost.
In article fourteen, "Independent Changes in Movement Operations" by
Dianne Jonas, Jonas discusses the V to I movement parameter in
Germanic languages. It has been well-established that there is a
contrast in Scandinavian languages between Icelandic which shows both
V to I movement and has rich verbal agreement versus Mainland
Scandinavian languages (Danish, Swedish, Norwegian) which has no V to
I movement and where verbs only inflect for tense. Within the
Minimalist framework, it has been assumed that rich verbal agreement
cued V to I movement and that the loss of agreement morphology cued
the lack of V to I movement. In this analysis, as the Mainland
Scandinavian languages lost subject agreement morphology, the
parameter for overt V-to-I movement switched and the overt movement
was lost. However, Jonas presents data from Faeroese, Old Scots and
Shetland which challenge these assumptions. Perhaps the most
interesting data comes from Old Scots in which the order of main
clause verb before negative particles show that this language had
V-to-I movement, yet even at this stage, Old Scots had lost all overt
agreement morphology. Similarly, Shetland has V to I movement, even
though its verbal agreement is extremely defective, that is showing a
lot of syncretism. Additional data comes Faeroese in which there is a
southern dialect with V to I movement and a northern dialect which no
V to I movement. I should note here that Jonas is looking at embedded
clauses for the Faeroese date, because Faeroese is otherwise V2.
Although Faeroese has lost some agreement features; some person
distinctions can be seen in the singular forms. Contrary to what
might be expected though, the dialect with the richer agreement
system is the northern dialect which has lost V to I movement. Thus
there is both a language with V to I movement and no agreement
morphology and a language with some agreement morphology which has
lost V to I movement (in comparison with the dialect with more
defective agreement). Jonas proposes that the reason V to I movement
is maintained in these languages is that these language still have V
to COMP movement either through V2 ordering (Faeroese) or through
question inversion or other similar structures (Old Scots, Shetland).
However, this does raise the question of how many V to C constructions
are needed before V to I movement is lost.
In article fifteen "Syntax and Morphology are Different: Commentary
on Jonas" Stephen R. Anderson argues that Jonas' data can be taken
further to show that there really is no connection between the
presence of V to I movement and richness in agreement morphology.
First he argues that the proposed "lines" between enough agreement to
trigger V to I agreement and to little for V to I movement is, as
Anderson puts it, "too legalistic" to be a likely general principle
in UG (Universal Grammar). For example, Roberts (1993) proposed that
V to I was lost if all plural forms were the same, while Falk (1993)
proposed that V to I is obligatory when both person and number were
distinguished; otherwise it is optional. Another argument that
Anderson presents is that the main diagnostic for V to I movement
used by linguists is word order (e.g. order of the main verb versus
negative particle), not verbal agreement. Therefore "if that is good
enough for syntacticians, it ought be good enough for the child as
well." Although Anderson maintains that morphological information can
be a useful diagnostic tool, he sends a general caution that
syntactic structure may not be identical to morphological structure.
Article seventeen, "Verb-Object Order in Old English" by Susan
Pintzuk also questions a relationship between morphology and syntax.
In this case, Pintzuk question an assumption that the loss of the
case system in English forced word order to become rigid and even
switch from OV to VO (Roberts 1997). In this paper, Pintzuk analyzes
an Early Middle English corpus in which morphological case was being
lost and found that word order was almost evenly split between VO and
OV orders even after the reduction of the case system. Therefore,
Pintzuk proposes that the rise of VO order in later English is not
directly connected with the loss of the case system, but some other
factor. Pintzuk also argues that Kayne's(1994) proposal that OV order
is derived from objects raising to a Spec-AgrOP from a base generated
VO order cannot explain certain distribution facts. Instead, Pintzuk
proposes that English went through a stage where the VO and OV
grammars "competed" although the nature of the competition is not
elaborated here. However the notion of grammars in competition is
echoed in almost all the remaining articles in this volume.
The following article by Jairo Nunes is article eighteen, "VO or OV?
That's the Underlying Question: Commentary on Pintzuk." Here Nunes
re-examines Pintzuk's data on verb-object adjacency and scrambling
that Pintzuk had used to argue against Kayne (1994) and proposes an
initial re-analysis which is compatible with Kayne (1994).
Article nineteen, Susana Bejar's "Movement, Morphology, and
Learnability" focuses on older English psych-verb verb construction
(e.g. "me thinks"). For this article Bejar returns to a Principles
and Parameters approach and proposes that the interaction between two
parameters, [+/- AM} (A-Movement) and [+/- UM] (Unpredictable
Morphology) can explain the distribution and evolution of these
structures. Like Pintzuk and Jonas, Bejar also rejects a direct
relationship between the loss of overt morphology and the loss of
psych verbs.
The last article in the section on "Independent Changes in Movement
Operations" is article nineteen, "Object Shift and Holmberg's
Generalization in the History of Norwegian" by John D. Sundquist.
Previous accounts had compared Icelandic which allows object shift on
pronouns and full DPs and includes a robust case system to Mainland
Scandinavian languages where object shift is more restricted and the
case system had been lost or reduced and proposed that overt
accusative morphology licensed overt object shift. This article
examines a corpus of Middle Norwegian diplomatic letters to show that
object shift has always been restricted to pronouns even when there
was overt accusative marking. Middle Norwegian was also unusual in
that it allowed both Object shift and scrambling despite the fact
that VO order was already present in the language; normally a
movement of an object over a head-initial non-finite verb would cause
a movement violation and be blocked. To account for this discrepancy,
Sundquist proposes that this stage of Norwegian had two competing
verb-object orders, VO and OV. Again this echoes the analysis
proposed by Pintzuk earlier in this volume.
In the final section of this volume, "Computer Simulations", two
articles explore possible mathematical properties of a linguistic
system and how children could evolve different grammars over time.
In article twenty, "The Computational Study of Diachronic Linguistics"
by Partha Niyogi, Niyogi develops a preliminary computational model of Lightfoot's (1991) cue-based approach using two grammars with two
populations using set operations and probability models. Article
twenty-one "Grammar Competition and Language Change" by Charles D. Yang,
on the other hand, assumes that children may develop two competing
grammars during the acquisition process. One reason Yang argues
against a trigger-based approach is that many structures are lost
gradually during the acquisition process, not suddenly as a
trigger-based account might predict. To show how his model works,
Yang then uses the model to provide an account for the loss of V2
ordering in French. Unfortunately, there is not enough space in this
review to adequate explore both, but as both Niyogi and Yang note,
developing a viable computational model for language change will help
delineate some properties of the process.
Interestingly one factor that neither the computational models nor
any of the other articles mention is the concept of areal changes.
Except for one article on Old Japanese and one discussion on Byzantine
Greek, the languages of this volume are exclusively Western European.
Since the Roman Era, a number of changes have occurred in Western
Europe which are shared among the Germanic, Romance and Celtic
languages. These include the tendency for the case system to be
eliminated or vastly reduced (Western Romance, Mainland Scandinavian,
English, Middle Irish, British Celtic, etc); the change from OV order
to VO order (Western Romance, English, Scandinavian, possibly Archaic
Old Irish) and the loss of V2 ordering (French, Middle English,
Middle Welsh, etc). Obviously, this cannot be a coincidence, yet
almost all accounts in this volume - with the exception of Jonas who
discusses possible interactions of Old Scots and Old Norse in
Northern Scotland - assume a completely language-internal account.
This is possibly problematic for a number of reasons. The main one is
that, as Anderson notes earlier in this volume, it is prudent to be
cautious generalizing patterns of syntactic and morphological case
based just on studying languages from one region, or one language
group, when so many general changes are shared. Also, if we are to
take the notion of population dynamics seriously in a computational
model as Niyogi does, then areal phenomenon will probably be a
factor. While it is not explanatory enough to just say one feature is
"borrowed" from a neighboring language, it may be the case that
multilingual children or children exposed to some features of a
neighboring language are more likely to construct a grammar with a
particular areal feature than monolingual or isolated children might
be. Western Europe, and other regions, can provide excellent case
studies of how and when shared feature changes occur and what
"preconditions" are needed for creating an "areal region." In this
case, all the languages in question are Indo-European even though the
split occurred centuries before the Roman Era, so perhaps certain key
similarities already existed.
Another theme of this volume that I wish had been more clarified more
is the idea of competing grammars. Although several authors cite
convincing evidence that texts for the same period may contain
several variants of a given structure, it is not clear how a
"competing grammar" could be formalized in the syntactic theories
assumed in this volume. Boskovic does note that Optimality Theory
(OT) is one possibility (although most OT analyses assume a single
grammar) and Yang doe propose a set theoretic model of competing
grammars in his computational model, but in terms of this volume
hitch focuses on the Minimalist or Principles and Parameters
framework, this is not "mainstream" so to speak. Clearly this is an
area for future research.
Despite the comments on a restricted language set and the notion of
completing grammars, I feel that this is a valuable volume for anyone
interested in the nature of syntactic change. I know that these
articles have provided this linguist many issues to consider.
REFERENCES
Bobaljik, Jonathan D. (2000) The Rich Agreement Hypothesis in Review.
MS, McGill University.
Bobaljik, Jonathan D. and Hoskulder Thrainsson (1998) Two Heads
Aren't Always Better than One. Syntax 1: 37-71.
Boskovic, Zeliko. and D. Takahasi. (1998) Scrambling and Last resort.
Linguistic Inquiry 29: 347-66.
Chomsky, Noam (2000) Minimalist Inquiries: The Framework in Step By
Step, Essays in Minimalist Syntax in Honor of Howard Lasnik, ed. by R.
Martin, D. Michaels and Juan Uriagereka, pp 89-155. Cambridge, MA: MIT
Press.
Evans, D. Simon (1964) A Grammar of Middle Welsh. Dublin: Dublin
Institute for Advanced Studies.
Falk, C. (1993) Non-Referential Subjects in the History of Swedish.
Ph.D., University of Lund.
Kayne, Richard S. (1994) The Asymmetry of Syntax. Cambridge, MA: MIT
Press.
Lightfoot, David W. (1991) How to Set Parameters: Arguments from
Language Change. Cambridge, MA: MIT Press.
Lightfoot, David W. (1999) The Development of Language: Acquisition,
Change and Evolution. Oxford: Blackwell.
Neeleman, A. and F. Weerman (1999) Flexible Syntax: A Theory of Case
and Arguments. Dordrecht: Kluwer.
Roberts, Ian (1993) Verbs and Diachronic Syntax: A Comparative
History of English and French. Dordrecht: Kluwer.
Roberts, Ian (1997) Directionality and Word Order Change in the
History of English in Syntactic Case and Morphological Case in the
History of English., van Kemenede, A. and N. Vincent, eds., pp.
397-426. Dordrecht: Foris.
Weerman, F. (1997) On the Relation Between Morphological and
Syntactic Case in Syntactic Case and Morphological Case in the
History of English., van Kemenede, A. and N. Vincent, eds., pp.
427-459. Dordrecht: Foris.
Willis, David (1998). Syntactic Change in Welsh: A Study in the Loss
of Verb Second. Oxford: Clarendon (Oxford University Press)
|
