Review of Explaining Language Structure through Systems Interaction
| Date: Wed, 21 Jul 2004 10:19:23 +0700
From: Viatcheslav Iatsko <firstname.lastname@example.org>
Subject: Explaining Language Structure through Systems interaction
AUTHORS: Frajzyngier, Zygmunt; Shay, Erin
TITLE: Explaining Language Structure through Systems Interaction
SERIES: Typological Studies in Language 55
PUBLISHER: John Benjamins
Viatcheslav Iatsko, Katanov State University of Khakasia
Since the authors of the book under review don't define its genre, I
will take the liberty to state that this book is a monograph that can be
of interest to a wide variety of persons engaged in linguistics:
undergraduate, graduate, and postgraduate students, researchers, and
The monograph is focused on theoretical foundations and applications of
Systems Interactions approach developed by the authors. This approach is
a methodology for the description of relationships among main language
systems, such as lexicon, morphology, syntax, and phonology. Within this
approach language is considered to be a complex organism consisting of
forms that interact with one another in the coding of various functions.
In order to investigate different types of lexical, morphological, and
syntactic systems, the authors drew data from languages from different
families: English, Polish, Russian, French, Welsh, German, Spanish
(Indo-European); Hausa, Gidar, Lele, Hdi, Mandara, Mina, Mupun, Hona,
East Dangla (Chadic); Arabic, Hebrew, Amharic (Semitic); Songhay
(Nilo-Saharan); Lango (Nilotic); Kiyaka (Bantu); Krongo (Korodofanian);
Dhalo (Kushitic); Malay (Indonesian); Tagalog (Philippinian); Japanese
(Altaic); Mandarin. Most extensively are used data from Polish, Mina,
Gidar, and Hdi.
The book comprises 2 parts, 12 chapters, a References section, and three
indices: of languages, of authors, of subjects.
Part I consists of chapters 2-8 that deal with various coding means;
part II includes chapters 9-12 that discuss specific features of several
functional domains. Of these chapters the biggest are chapters 1, 3, 5,
11; the shortest are chapters 6, 7, and 12. Chapter sizes will be taken
into account in my review: longer chapters will get a more detailed
This chapter is entitled ''Introduction: Theoretical and methodological
foundations'' provides important information about theoretical principles
and notions underlying the research. They are: the principle of
functional transparency, the notion of formal coding means, the notion
of functional domains, and the notion of systems interaction in the
coding of functional domains.
The principle of functional transparency implies that 1) every utterance
in discourse must be transparent with respect to its relationship to
preceding and following discourse; 2) the role of every constituent in
the utterance must be transparent within the applicable functional
domain. Functional transparency applies only to those domains that are
grammaticalized in a given language. Consequently, the scope of
functional transparency is not the whole language, but only the
functional domain of the utterance. For example, Mandarin hasn't
grammaticalized agent-patient dichotomy, and these roles do not have to
be transparent to the hearer, though they may be transparent in other
Means of coding functional domains, grammaticalized in the language
comprise lexicon, inflectional and derivational morphology, linear
order, and phonology. These coding means must meet the following
requirements: they must be independent, that is, the use of a given
means must not be triggered by any other element occurring in the
clause; two different features belonging to the same domain can not be
coded on constituents of the same construction.
A functional domain is defined as ''a class of mutually exclusive types
of expressions constructed by a set of specific means of coding'' (p.26).
The authors reasonably note that contemporary syntactic theories do not
distinguish between the functional domain and the coding of the domain
reflecting a lack of distinction between means and function. The book
discusses the following functional domains: locative predication coding,
semantic relations coding, and reference systems.
The principle of systems interaction in the coding of functional domains
excludes the possibility of having one function coded by different means
within the same construction. If several coding means have functions
within one functional domain, these means are in complementary
distributions such that each means codes a different subdomain.
The notions and principles mentioned above constitute a basis for a new
approach in cross linguistic analysis developed by the authors. As they
correctly state, contemporary typological studies make use of the notion
of prototypical grammatical categories that are based on languages, with
which the linguist is most familiar. Pointing out drawbacks of extending
the notion of prototype to cross linguistic typology, the authors
suggest an innovative methodology that implies the description of
functional domains that have been grammaticalized in the languages in
question, and then comparing the internal structure of a given
functional domain across languages. I can suggest calling this
methodology ''functional comparative analysis''. Though I agree with the
authors I'd like to point out that a linguist conducting research, no
matter how he tries, is anyway influenced by the knowledge of his mother
tongue and other languages. To the best of my knowledge, the degree if
this influence and its distinctions, haven't been studied yet. I can
only conjecture that two linguists would get different results if they
apply the same methodology to one and the same object of study if they
have different language backgrounds. As it will be shown later the
authors themselves couldn't avoid adopting some prototypes.
I have paid so much attention to the first chapter of the book because
it well reflects its advantages and disadvantages. They will be
summarizes later; meanwhile I'd like to note the following.
1) Analyzing linguistic structures the authors make emphasis on form,
rather than on meaning: ''Within the proposed approach there exist no
functional domains, individual functions or individual meanings, except
as they are coded by the formal means of a given language... . In our
view, meaning ... exists only within a given language and only to the
extent that the language codes it through some formal means'' (p. 3). I
can't help agreeing with this thesis. But I can't help noticing that
there is no isomorphic correlation between linguistic form and meaning;
it is common knowledge that one and the same form can code different
meanings, and, on the contrary, one meaning can be represented by
different meanings. In such cases, as it will be shown later, systems
interaction methodology proves to be insufficient and must be
supplemented with semantic analyses.
2) On page 19 the authors state that ''the presence of inflectional
coding affects the forms of discourse'' substantiating this statement by
the example from Polish ''Moi starzy mowia'' (I can not use here
diacritics), in which, according to the authors, possessive pronoun
occurs without its head. I can't help agreeing that in Polish as well as
in Russian and other languages with inflectional coding possessive
pronouns and adjectives can occur without their heads, but as for this
specific example, the authors are wrong. Possessive pronoun in the
sentence given above modifies ''starzy'', which is sure to be a
substantivized adjective, with which the possessive pronoun agrees in
It's interesting to notice that due to gender, number, and inflection
coding there are much more substantivized adjectives and participles in
Russian, and Polish than in English. Unlike English Russian can
substantivize almost any participle producing much shorter discourse
unites. For example (unfortunately I can not use here Cyrillic
characters) participle ''ukushennyi'' should be rendered into English by
at least 6 words - ''a man bit by some animal''. Perhaps this holds for
other Slavic languages and all languages with the same type of coding.
This fact hasn't been given due attention yet in contrastive analyses
and the book under review doesn't consider it also, though it provides a
convincing explanation for it.
3) The authors sometimes resort to categorical statements that can
hardly be accepted in language analysis. For example one of the sections
in Chapter 1 is entitled ''Agreement and government do not exist, but
disagreement and insubordination are not tolerated''. Denial of
agreement's existence is too strong a statement, especially if we take
into account that the authors use the term ''agreement'' in their notation
system (see, for example sentences' parsing on p.14) and write: ''We take
agreement, like all other types of inflectional coding, to be an
independent coding means'' (p.21). Perhaps it's a rhetorical device used
to attract readers' attention.
4) The chapter, as well as the whole book, abounds in misprints that may
baffle the reader. For example on p. 20, example (17) a question mark is
placed after the second Polish sentence, and in the English translation
- after the first English sentence. Actually both Polish sentences may
be interrogative. A question mark is missing in Polish sentence in
example (27). I am not going to list all misprints scattered all over
the book. I am just stating that the book could have been much better
''Interaction of the lexicon with other coding means'', shows
that the defining characteristics of lexical categories and
subcategories are based on grammatical systems of individual languages.
Accordingly, languages with similar grammatical systems will have
similar lexical categories.
This chapter can be divided into two parts. The first part proposes a
model of lexicalization that provides criteria for determining what
lexical categories exist in a language. The model is based on the
principle that if a function has been lexicalized no grammatical means
is used to code it when its lexical form coding is used. If a lexical
item is used in a function other than that for which it has been
lexicalized within a given functional domain, special morphological
and/or syntactic means must be used to code the function of that item.
Such an approach allows each lexical category in a language to be
described independently of other categories in a language. Any lexical
category may contain subcategories that code specific semantic functions
or that allow for the formation and functioning of other coding means.
The authors illustrate the proposed model with an examination of the
category ''Adjective'' in Mupun.
The second part of the chapter illustrates how the proposed model may
lead to the discovery of a lexical subcategory, in this case the
subcategory of inherently locative nouns in Gidar.
On the whole, authors' conclusions in Chapter 2 seem quite sound;
nevertheless, we again com across doubtful interpretations of some
examples. Analyzing Polish example ''Profesor a glupi''
(Professor+conjunction 'but'+silly) the authors state that one can have
a noun in the subject function and an adjective or verb in the predicate
function, but one cannot reverse the order of constituents (*Glupi a
professor) since the subject would then be more predicative that the
predicate. If the authors had punctuated properly they would have found
out that both variants are grammatical and the degree of predicativness
is the same: ''Profesor, a glupi'', ''Glupi, a profesor''. Denouncing the
idea of prototypical categories drawn from English the authors have
overlooked the fact that punctuation in such languages as Polish and
Russian plays much more important role than in English. In my country
high school teachers illustrate to the students importance of
punctuation by the sentence ''Kaznit' nel'zia pomilovat'
(Infinitive-'execute' + negation-'can't' + infinitive - 'pardon'). If
the comma is placed after the first infinitive the sentence acquires the
meaning: ''Execute him/her, he/she cannot be pardoned''. If the comma is
placed after the negation the sentence acquires the meaning: ''Pardon
him/her, he/she cannot be executed''. Actually, punctuation in Russian
and Polish is a separate coding means. This fact has been completely
ignored by the authors.
This chapter is entitled ''Coding through linear order''. This term means
the coding of a function through the position of an element with respect
to some reference point. The authors give evidence that only the verb is
a potential reference point for coding through linear order. Considering
the function of third person agreement the come to a conclusion that in
a clause with a nominal subject verb and noun can not be distinguished
by phonological means and coding 3d person singular through the verbal
suffix marks the lexical categoriality of the verb. Analyzing English
constructions with the verb ''give'' the authors provide convincing
evidence that the English construction with two nouns following the verb
has the function of coding the first noun as beneficiary and the second
noun as an object. The authors also discuss some specific features of
coding through linear order in languages of different types:
dual-branching, left branching, right-branching.
''Coding through nominal inflexion'' is focused on case marking.
It demonstrates that case marking is an independent coding means and
that the presence of nominal inflection increases the number of linear
orders available in a language and thereby the number of functions that
may be coded by linear order. The authors discuss some functions of the
accusative and genitive cases and discourse implications of case coding.
This chapter, along with some interesting conclusions, such as coding
non affected object by the genitive case in Polish, default role of the
accusative, contains some not well substantiated and somewhat obvious
statements. On p. 99 the authors, without providing any evidence, state
that the accusative case cannot be used when the clause is negative.
This statement seems strange. Consider the following example from
Russian: ''Ia ne vizhu knigu'' (I + not see +book, I can't see the book).
This sentence is sure to be negative and the noun ''knigu'' is in the
accusative case. Similar examples can be found in Polish and in other
Slavic languages. It's interesting to note that in some declination
types (Russian has 4 main types of declination, each subdivided into
three subtypes) forms of the accusative coincide with the forms of
nominative and genitive cases. For example ''Ia ne vizhu tebia'' (I + not
see + you, I can't see you). Here the accusative form of the pronoun
''tebia'' (you) coincides with the form of the genitive case (second type
of pronominal declination). It should be noted that the authors discuss
negative accusative constructions later, in chapter 11 (see below).
Analyzing discourse differences between Polish and English the authors
examined 100 pages of Polish written discourse to come to the
conclusion: ''The differences between English and Polish can be explained
by the fact that a noun in Polish is marked by the inflexion for its
role in the clause while a noun in English is not. Adding a predicate or
auxiliary in English allows the role of the noun to be coded through
linear order'' (p.111). This statement is rather obvious and even
trivial. These differences between analytical and synthetic languages
have been known in linguistics since the 19th century and are reflected
in numerous textbooks. Cf. Blokh's (1983, p.32) words in a textbook for
undergraduate students: ''Synthetical grammatical forms are realized by
the inner morphemic composition of the word, while analytical
grammatical forms are built up by a combination of at least two words,
one of which is a grammatical auxiliary ... . Synthetical grammatical
forms are based on inner inflection, outer inflection, and
''Interaction of phonology with other coding means'' deals with the
problem of the relationship between phonology and other coding
means, specifically linear order and morphology. The authors propose
that one function of the phonological component in some languages is to
code internal structure of the utterance. They managed to prove that the
function of the form with the underlying final vowel is to mark a
phrasal boundary while the function of the form with reduced vowel codes
phrase-internal position. For example, in Polish and Russian nasal
variant of third person pronouns codes the pronoun as belonging to the
same phrase as the preceding word, while the palatal variant indicates
that the pronoun doesn't belong to the preceding phrase and so should be
interpreted as belonging to a separate phrase. In some Chadic languages
a demonstrative pronoun that undergoes assimilation with the preceding
syllable is coded as belonging to the preceding syntactic init. In Welsh
soft mutation of consonants after the article is the marker of the
feminine gender of the noun.
In ''Agreement or coding on other constituents'' the authors
consider different approaches to agreement, prove that coding on the
verb and agreement are an independent coding means, and analyze
functional domains coded by agreement on the verb, such as: information
status of the participants (topic, focus), grammatical roles of
participants (subject, object), semantic roles of participants, and
referentiality of participants.
This chapter convincingly and logically argues against the widely held
belief that the function of nominal classification is to fulfill some
inherent human need to categorize the world, and describes interaction
between systems of nominal classification and other coding means in a
language. Nominal classification is considered to be a coding means
whose overall function is to enable operation of other coding means. The
authors prove that, systems of nominal classification enable the
existence of such coding means as agreement systems and systems of
deictic and anaphoric reference.
This chapter is concentrated upon the structure and functions of 'raising'
constructions in which the subject of the embedded clause appears as
object of the matrix clause. The authors consider matrix clause coding
as an independent grammatical means that interacts with and is
complementary to other lexical, syntactic, and morphological means. They
show that it may have various functions: with the verbs of perception
and volitional verbs it often code epistemic modality. With the verbs of
perception matrix clause coding has the modal function of coding direct,
as opposed to indirect perception. With the volitional verbs matrix
clause coding indicates that the speaker considers the event to be
possible or real.
It should be noted that in Russian 'raising' constructions similar to
English ones are not used and to learn them is difficult for Russian
speaking students. This chapter provides materials that have some
practical applications in foreign language teaching.
In ''Determining the function of a linguistic form: the indirectly
affected argument and the external possessor'', the authors
provide ample evidence on the one hand against external possessor
hypothesis, according to which a construction, whose formal properties
include the presence of two noun phrases, one marked by the dative case
and the other by some other case, has the semantic function of coding
possession; on the other hand they prove that this construction, in
fact, codes the relationship of indirect affectedness between the verb
and the dative argument. Having studied data from Polish, the authors
show that the function of external possession doesn't exist in Polish, a
language for which this construction has been claimed, and that Polish
and other languages have grammaticalized the function of indirect
''Systems interaction in the coding of locative predication'' focuses
on locative predication in Mina, a Central Chadic SVO language.
It demonstrates interaction of the various coding means in the domain of
locative predication. These means include lexical items occurring as
predicates and complements; linear order; the prepositions, alone or in
combination; and some combinations of these means. The authors managed
to formulate rules that determine combinations of various coding means
to express locative predication. It turns out that in Mina. The
fundamental criteria for forming the locative predication are the
semantic class of the predicate and the semantic class of the noun
serving as the locative complement. If the predicate is inherently
locative nothing else is required to code the locative nature of the
predicate. If the predicate is not inherently locative a predicator must
follow the predicate.
If the noun that serves as a locative complement is inherently locative,
nothing else is required to mark the noun as a locative complement. If
the noun is not inherently locative a preposition must be used to mark
it as a locative complement. These rules formulated for Mina are a good
example of authors' methodology that can be extended to the analysis of
''Systems interaction in the coding of reference'' outlines specific
features of a number of subdomains in the domain of reference.
The authors provide for the existence of the following subdomains:
deixis; previous mention; 'known' referent'; de dicto and de re;
definiteness; 'deduced' referent.
It is in this chapter where constraints on systems interaction approach
become obvious. It turns out that when applied to the domain of
reference this methodology must be supplied with a more profound
semantic analysis. For example, on p.271 the authors state that in
Russian the accusative case code a referent that is known, specific,
previously mentioned, or present in the environment of speech, while the
genitive case codes a referent that is not identifiable in any of these
domains. This statement is substantiated by examples of Russian negative
clauses, in which nouns without determiners are marked by the genitive
case, and nouns with determiners are marked by the accusative case.
Specifically, the authors give the following examples:
(57) A (conj) ischezli (disappeared) sovershenno (completely)
ne (negation) ostaviv (leave) sledov (traces - genitive)) dazhe (even)
v (in) literature (literature)
(58) A (conj) ischezli (disappeared) sovershenno (completely) ne
(negation) ostaviv (leave) eti (these) sledy (traces - accusative) dazhe
(even) v (in) literature (literature).
The authors provide the following translations: (57) And they have
completely disappeared without a trace, even in the literature;
(58) And they have completely disappeared, without those traces, even in
If the authors had given correct translation they would have realized
that (58) and its translation are ungrammatical and unacceptable. The
point is that the adverb ''sovershenno'' (completely) modifies ''ne
ostaviv'' (without leaving), but not the predicate ''ischezly''. The correct
translation of (58) should read: ''And disappeared without leaving these
traces at all, even in the literature''. This English sentence is
ungrammatical as well as its Russian counterpart. The ungrammatical
nature of both sentences can be explained in terms of case grammar. The
verb ''leave'' assigns a factitive thematic role to the argument, if this
argument is an abstract noun. Since factitive denotes ''the object
resulting from an action or state, having no prior existence but coming
by virtue of the action or state'' (Brinton, 2000; p.268), in the
negative clause this existence is denied. Hence we have semantic
anomaly: negative form of participle denies existence of ''traces'' while
the determiner ''these'' implies their existence. Negative clauses, in
which a verb assigns to its argument not a factitive, but some other
role, can take determiners. Cf. ''*He left without making these mistakes''
(''mistakes'' is a factitive); ''He left without doing these exercises''
(exercises is a patient). The same holds for Russian. Example (56), p.
271 ''Nu, ja eti tonkosti ne znaju'' (Well, I am not aware of these fine
points) is grammatical because ''tonkosti'' is a neutral, but not a
This short analysis provides evidence for one important constraint on
systems interaction approach suggested by the authors: this methodology
operates on condition that there is an isomorphic correlation between
form and meaning. In the case of homomorphic correlation between them,
this approach must be supplemented by a more profound semantic analysis.
Once more I have to point out authors' carelessness in presentation of
data. Commas are missing in examples (67), (68), (69). An unpleasant
impression is produced by the strange spacing in (63), (67) and many
other examples. In (67) two words that must be separated by a comma (byl,
iz) are merged into one word. By the way, examples (59)-(66) all admit
both cases: genitive as well as accusative, contrary to authors'
statement that some of them are unacceptable. Actually these are
examples of case alteration typical of the genitive and accusative cases
in Russian colloquial speech. These examples are a good illustration of
different forms coding one meaning, and authors' attempts to impose
their transparency principle to account for the use of different cases
seem a ridiculous simplification.
V. A. Beloshapkova, (1989; p. 426) writes: ''One shouldn't simplify
matters thinking that differences in a number of ... case forms are
always reflected in their meaning. Here (just like in phonology and word
formation) free alteration of forms is also possible....''
''Conclusions, implications, and open questions'' outlines some
implications of the authors' work for the theory of functions in
phonology;the theory of the organization of the lexicon; the theory of
the linear order as a coding means; the phrase structure theory; the
methodology of discovering the function of a linguistic form; the theory
In conclusion I'd like to summarize some advantages and drawbacks of the
book under review.
- The authors suggest an innovative systems interaction approach, whose
methodology can be successfully applied to investigation of various
linguistic phenomena. Nevertheless, functional transparency principle
restricts the application of this methodology to isomorphic correlation
between linguistic form and linguistic meaning. If this correlation is
homomorphic, systems interaction approach must be supplemented by more
profound semantic analyses.
- The monograph contributes to the development of creative thinking; its
language is simple enough to be understood by foreign readers having a
more or less decent command of English. It can be recommended for a wide
variety of linguists, from undergraduate to post doctoral students,
researchers and educators.
- Many works that analyze data from a number of languages belonging to
different families meet with the problem of correct linguistic analysis
if sentences' structures. Since annotated corpora for many languages
haven't been yet created the authors as I noticed in one of my earlier
reviews (Iatsko, Van Valin) should be consulted by native speakers -
professional linguists otherwise they make mistakes that diminish
quality if their research. Unfortunately, the authors of the monograph
haven't coped with this problem. It abounds in misprints, careless
presentation of data, and their wrong interpretations.
Generally, the monograph's style is somewhat doubtful: on the one hand
the authors display much enthusiasm about their approach using some
categorical statements; on the other hand they are careless about
important details. Such a style is more characteristic of junior
researchers, but not of esteemed scholars, who are, no doubt, Drs.
Frajzyngier and Shay.
Beloshapkova, V. A. (1989) The modern Russian language. Moscow:
Blokh, M. Y. (1983) A course in theoretical English grammar. Moscow:
Brinton, L. J. (2000) The structure of modern English. Amsterdam;
Philadelphia: John Benjamins.
Iatsko, V. (2001) Review of Van Valin, An introduction to syntax. http://linguistlist.org/issues/12/12-2380.html#1
| ABOUT THE REVIEWER:
ABOUT THE REVIEWER
V. Iatsko is a professor in the Department of English and Head of
Computational Linguistics Laboratory at Katanov State University of
Khakasia located in Abakan, Russia. His research interests include
text summarization, text grammar, TEFL, contrastive analysis of English
and Russian syntax.