Lingua
75 (1988) 251-287.
North-Holland
251
REVIEWS Eva KoktovB, Sentence adverbials in a functional description. Pragmatics & Beyond VII: 2. Amsterdam/Philadelphia, John Benjamins, 1986. VIII + 96 pp. Reviewed by: Milan Bily, Department of Slavic Languages, Uppsala University, Box 513, 751 20 Uppsala,
Sweden.
1 It is a nice feeling to review a book one thoroughly likes. I have had the pleasure to read Sentence adverbials in a functional description by Eva Koktova. The work in question presents a new syntactico-semantic treatment of certain adverbial expressions (adverbs as well as prepositional phrases) called Complementations of Attitude (CA), such as frankly, to my surprise, certainly, alternatively, as a matter of fact, only, also, for example, etc. etc. Such expressions have often been treated under various headings of different cross-classifications, though some kinds of similarities (exhibition of similar functional properties) have usually been recognized by the authors of many competing descriptions. In Koktova’s alternative proposal, CA:s are described on the basis of their common behaviour in the underlying representation (UR) of a sentence, i.e. on the basis of their scoping properties and certain surface-syntax properties (surface movement). Her approach differs from the earlier descriptions by avoiding the trap of cognitively-based criteria, such as paraphrasabilities by various higher predicates. The author claims that CA: s are lexically heterogeneous, but functionally homogeneous, regardless of whether they belong to the traditionally recognized class of sentence adverbials (those adverbials that are paraphrasable by higher predicates and logically representable by means of predicates or operators, and whose semantics have been viewed as related to the whole sentence), focusing adverbials (those that are not paraphrasable in the above-mentioned ways, and whose semantics have been viewed as related to a sentence part), or various disjointed minor classes postulated by other authors. Koktova also abandons the usual practice of accounting for scoping ambiguities in terms of lexical homonymy. I.1 Koktova is not the first author who has recognized the relationship between adverbs belonging to the class of CA:s and the topic-focus structure underlying surface sentences. Some Slavicists (e.g. Grochowski in Poland and the Czech linguists
0024-3841/88/%3.50
0
1988, Elsevier Science Publishers B.V. (North-Holland)
Dane; and Hajieovi) have studied this relationship for certain restricted groups of CA:s. However, hers is the first attempt at a unified description of all kinds of CA: s. The theoretical framework of the monograph is provided by the Functional Gencrative Description (FGD) cf., e.g. Sgall et al. (1986), the contemporary development 01 the teaching of the Prague school by the linguistic team at the Faculty of Mathematics and Physics of Charles University in Prague. led by Professor Petr S&all. FGD is a multilevel system of explicit description consisting of a generative component in the form of a dependency grammar and of several levels ordered from meaning to sound. The generative component generates linguistically disambiguated underlying rcprcsentations in the form of linearized dependency trees. FGD’s concept of linguistic meaning is narrow ~ it takes into account only those distinctions which have to do with the structure of the natural language system in question. while distinctions concerning the domain of the extra-linguistic (cognitive) content are excluded. The distinction is made explicit by means of certain operational criteria which keep apart genuine ambiguities of surface structures from mere vagueness, genuine synonymy from quasisynonymy. etc. For example, the genuine synonymy of two linguistic expressions exists only when the expressions arc interchangeable in any linguistic context (except for metalinguistic ones). Consequently. many instances of supposed synonymy appear to involve a mere accidental coincidence in cognitive context the two expressions may happen to dcnotc the same extra-linguistic reality. While the expression /~ornisc~ plus an object clause is genuinely synonymous with promise plus the corresponding infinitive construction. (i.c. these expressions share the same underlying representation). such a converse pair as X sold Y to Z and Z ho~~ghr Y fion7 X are not true synonyms. which becomes obvious in an appropriate differentiating context. E.g. Cl7crrks intmtionull~~ .sold (I hul cur to Torn means something completely different from Tom intcntionull~~ bought (I hurl cur ,/Uom Churlcs. A part of the underlying reprcscntation is the topic--focus articulation (TPA). involving not only the bipartition of the information conveyed into the old (given. presupposed) vs. the new, but also a more subtle hierarchy of the deep word order in UR based on a gamut of rising degrees of Communicative Dynamism (CD) of the sentence elements. The degrees of CD as well as the basic bipartition arc correlated to the varying salience of the dynamic, changing stock of knowledge shared by the speaker and the hearer of a discourse. The elements of the stock that arc activated over a certain threshold of importance/interest at the given moment are considered contextually bound. The concept of contextual boundness comprises co-textual boundness (the linguistic context). situational boundness. and boundness due to the permanent activation of indcxical elements and their derivatives in the memory of users of natural languages. Contextual boundness has a recursive character (as the dependency structure of a sentence generally has at any depth of the dependency tree) ~ any linguistically explicit predication. any subtrcc of the dependency tree (i.e. every sequence of sister nodes) has its own deep word order. For example. an embedded non-bound node may constitute an embedded focus of a sentence. as in Tcrr~~ is lookip
,fiw an
cx-co)7
t’ict bcith LI RED shirt.
1.2 There is a so-called systemic ordering (basic underlying ordering of complements for maximally context-free sentences) which is based on the empirically discovcrcd communicative weight of the single types of complementations. While these systemic orderings are similar for particular languages, they are not necessarily identical. The following systemic ordering is assumed for English: Time ~ Actor - Addressee ~ Patient - Origin ~ Effect ~ Manner -- Directional (which way) - Directional (from (where to) - Location where) ~ Accompaniment - Instrument ~ Directional Condition - Concession ~ Cause ~ Purpose. The positions of the elements in the deep word order become quite obscured in languages with ‘grammatically fixed word order’, as in English, while in languages with ‘free word order’, such as Slavic languages (with the exception of Bulgarian and Macedonian, which have lost the case inflection) the surface ordering shows more or less straight correspondence to the ordering of elements in UR. which expresses the salience of the elements of the stock of shared knowledge related to the actual universe of discourse. Nevertheless, a surface representation of a sentence is often ambiguous among several possible underlying sources differing as to the placing of the topic-focus boundary. It should also be noted that, for example, the English sentence pair Terry is READING vs. TERRY is rruding are two different. non-synonymous sentences. not one and the same sentence ‘pronounced in different ways’. Similarly, the ‘free word order’ of Slavic languages is not a matter of some luxurious and redundant ‘scrambling rules’. It reflects the different UR:s; it is (mostly) the result of the existing lack of synonymy. The TFA of one or several UR:s corresponding to one surface structure is identified objectively by the negation test or the question test (cf. e.g. Sgall and Hajicova (1977,1978). Koktova describes the scoping properties of CA expressions in this way: any natural language expressions with the semantic properties of operators (ie. with scoping properties), CA expressions as well as natural language quantifiers such as every, many, a etc, have wide scope over all expressions standing to the right in the UR of a sentence (i.e. over those carrying a higher degree of CD than the operator expression). 1.3 In the remainder are utilized:
of this text some of Koktova’s
conventions
for linearized
UR: s
- The topic-focus boundary is indicated by a slash. - The scopes of CA: s are indicated by arrows. ~ Auxiliaries of tense and modality are connected with their main verbs by means of a hyphen. ~ CA expressions stand in scope-unambiguous positions.
~ All complementations are ordered in agreement with the rising degrees of CD. ~ The verb stands either in the topic-final position (with some exceptions), in the focus-initial position, or in the second position of the focus. The last case concerns the majority of examples analyzed, since the ‘unmarked’ position of CA: s (the position according to the systemic ordering of complements) is the leftmost one (the position with lowest CD). - In case of ambiguous surface sentences, only the relevant UR: s (relevant readings) are taken into account.
2 The following primary possibilities are obtained: The scope of a CA extends over the whole (topicless) (1) Probably,
Terry will run from Manhattan
The UR of the primary (la)
/ Probably
reading
sentence,
as in (1)
to Brooklyn.
of (I) is (la):
will-run Terry from Manhattan
to Brooklyn..
Such a topicless sentence may serve as an answer to a general question such as What is the news? The scope of a CA extends over the predicate consisting of the verb and possibly one or several complements, as in (2) (2) Terry will probably
run from Manhattan
to Brooklyn.
The UR of (2) is (2a): (2a) Terry / probably
will-run
from Manhattan
to Brooklyn.
(2) can serve in this reading as an answer to What about Terry? The scope of a CA extends over the complements but not over the verb, as in (3). (3) Terry will run probably The UR for the intended
from Manhattan
reading
(3a) Terry will-run / probably In this reading
(3) answers
to Brooklyn.
is (3a): from Manhattan
the question
to Brooklyn.
Where will Terry run? The UR of (3a) also
Reviews gives rise to the unambiguous meaning of (3) viz. (3’): (3’) It is probably
surface sentence
from Manhattan
The scope of a CA extends intended reading of (4)
255 with the same meaning
to Brooklyn
where Terry will run.
over some complement(s)
(4) Terry will run from Manhattan
probably
as the primary
of the predicate,
to Brooklyn
as in the
(but not to the Bronx).
The UR is (4a): (4a) Terry from Manhattan
will-run
An unambiguous sentence corresponding comes about via the same UR (4a): (4’) It is probably
to Brooklyn
/ probably
to Brooklyn.
to the intended
reading
of (4) is (47, which
where Terry will run from Manhattan.
Both (4’) and (4) in the same reading can be answers to Where will Terry run from Manhattan? Other primary cases of possible UR: s are usually expressed on the surface by means of marked intonation, or by the construction with there, as in (5) corresponding to the UR (Sa) and (6) corresponding to the UR (6a): (5) Probably TERRY (5a) From Manhattan
(but not Mary) will run from Manhattan to Brooklyn will-run / probably Terry.
to Brooklyn.
(6) There will run to Brooklyn probably Terry from Manhattan. (6a) To Brooklyn will-run / probably Terry from Manhattan.
(Unfortunately, my English informants strongly opposed the grammatical correctness of (6). The same reaction has been obtained to (11) (13) and (14) of 2.1.) To sum up, in the primary case a CA stands in the focus-initial position, its scope extending over the rest of the focus. A CA can never occur in underlying focus-medial, nor in focus-final positions. CA expressions cannot be pronominalized. (Cf. Terry will run probably to Brooklyn. *Henry will run so to Manhattan.) In the systemic ordering of complements, the leftmost position, i.e. the position before Time in the ordering presented in 1.2, is established for CA: s. As CA can be considered a means of the speaker’s commentary on how the focus of a sentence holds, Koktova recognizes the class of CA: s as open-ended, i.e. new CA: s may be
156
Rovkwr
created. This summary is assumed to hold not only for English, but for every language possessing the category of CA:s, and every natural language is assumed to have means for expressing how the focus of a sentence holds.
In case of the multiple occurrence of CA: s, only the leftmost CA in the UR can occur in the focus-initial position. All other CA: s are focus-medial, extending their scope forward (rightward), i.e. they have in their respective scope portions of the focus to the right of the CA in question: (7) Terry will run probably only to Brooklyn. (7a) Terry will-run / probably only to Brooklyn.
It is worth mentioning that the occurrence of this type of CA cluster is rather restricted in the surface structure it may appear mainly in the surface medial position, while it is mostly excluded in the surface initial and surface final position. in coordinations, and in elliptic answers: *Probably only, *Terry will run *Terry will run, (8) (7”‘) (Will Terry run (7’) (7”)
Terry will run to Brooklyn. to Brooklyn, probably only. surprisingly and amusingly only, to Brooklyn. to Brooklyn?) *Yes, probably only.
Furthermore, these clusters are limited to two members and only certain CA: s can occur twice in the same clause. with the leftmost CA in the UR moved to the surfaceinitial position, such as in (9) with the UR of (9a): (9) Only,, Terry forgot to wash only, his own cup. (9a) Terry forgot to wash / only, only, his own cup.
However, other clusters allow the above-mentioned final, and coordinated clusters, as well as elliptic interpretation: (10’) (10”) (11)
excluded answers,
Not surprisingly, Terry will run to Brooklyn. Terry will run to Brooklyn, not surprisingly. Terry will run, not surprisingly and amusingly, comment on (6) in 2.) (IO”‘) (Will Terry run to Brooklyn?) Yes, not surprisingly.
surface-initial, surfaceand they get another
to Brooklyn.
(Cf.
the
The UR for (10) as well as (10’) and (IO”) is (lOa), where the leftmost CA has in its scope only the nearest CA to the right and the whole CA cluster has in its scope the rest of the focus: (10) Terry will run, not surprisingly, (I Oa) Terry will-run / not surprisingly
I
The scopes of the (rather hypothetical) are assumed to be as in (12a))( 14a):
to Brooklyn. to Brooklyn.
multiple
(12) Terry will run, hopefully not surprisingly, (12a) Terry will-run / hopefully not surprisingly
occurrence
of more than two CA: s
to Brooklyn. to Brooklyn.
I (13) Terry will run, not surprisingly amusingly to Brooklyn. (13a) Terry will-run / not surprisingly amusingly to Brooklyn.
1
I
(14) Terry will run, not surprisingly probably correctly, to Brooklyn. (14a) Terry will-run / not surprisingly probably correctly+ to Brooklyn.1
I
(Cf. the comment
on (6) in 2.)
However, it is not explained why the CA: s are hierarchically organized different scopes) in the way outlined in (12a) vs. (13a), or why the hierarchy as it is assumed and not, e.g. as in (14b): (14b) Terry will-run
/ not surprisingly
probably
correctly
(and have in (14a) is
to Brooklyn.
It thus becomes obvious that there are lexical differences between various kinds of CA: s with respect to their scoping properties, both regarding the hierarchical organization of CA:s in the examples mentioned and the difference between the former (sentences (7)-(9)) and the latter (IO)-(14) interpretations and the allowed sentential positions in syntactic surface structures. In other words, the CA:s are not such a homogeneous group of complementations as the author believes. To be honest, Koktova recognizes eleven lexical subgroups of CA expressions (cf. p. 28) and in chapter 6 (esp. pp. 6675) takes up the question of CA: s’ lexically determined surface distribution and the primary scoping interpretations (the hierarchical scale of scoping relations between the various CA subgroups).
258
Reviews
2.2 The partial heterogeneity of CA: s becomes also quite evident in regard to the possibility of functioning as the only element of the focus. In sentences which do not serve as answers to yes-no questions, only a limited number of CA: s can be the only element of the focus, extending its scope backward (in the UR) over the whole topic: (15) (Henry will run to Brooklyn.) Terry will run to Brooklyn (15a) Terry to Brooklyn will-run / also.
also
Some of these CA: s (as well, either, too, etc.) can occur only with such backward scope; others can occur also with forward scope (i.e. in the focus-initial position in the UR), e.g. also, in comparison with, from the viewpoint of (16) (Terry will run to Brooklyn.) He will run also to Manhattan. (16a) He will-run / also to Manhattan.
Otherwise ‘the majority (full or elliptic) answers
of CA expressions’ can occur as the only element to yes-no questions:
of focus in
(17) (Will Terry run to Brooklyn?) Yes, probably. (17a) Terry to Brooklyn will-run / probably.
Similarly, there are restrictions on the acceptability of CA: s with scope over some part(s) of the topic, i.e. in sentences such as (18) with the UR of (18a): (18) Terry will run only from Manhattan to Brooklyn because of Mary. (18a) Terry only from Manhattan to Brooklyn will-run / because of Mary.
2.3 CA: s may occur at any depth of non-clausal subtrees of the sentential dependency tree. On the surface such CA: s are placed before the highest embedding subtree: (19) Terry is looking (19a) Terry is-looking
probably for an ex-convict with a RED shirt. for an ex-convict with a shirt / Probably red.
3 On the surface,
CA: s most
often
occur in scope-ambiguous
positions.
The only
259
Reviews positions that may be unambiguous as in (20) resp. (21):
are immediately
or remotely
postverbal
positions,
(20) Terry will run probably to Brooklyn (but not to Manhattan). (20a) Terry will-run / probably to Brooklyn.
(21) Terry will run from Manhattan probably to Brooklyn (but not to the Bronx). (21a) Terry from Manhattan will-run / probably to Brooklyn.
However, as soon as an element of the focus constitutes a non-clausal subtree (as mentioned in 2.3, a head-noun with an attribute is enough), even such positions become ambiguous. In my opinion, this is true even for placement of unmarked stress in the constituent in question, because the interpretation of the kind mentioned in 2.3 is always possible. Thus unambiguous scopes of postverbal CA: s must be rather an exception. All other surface positions are guaranteed scope-ambiguous ones. The UR:s (20a), (22a), and (23a) correspond to the surface structures of (20) and (22)-(24) in the following way: (22a) / Probably
will-run
(23a) Terry / probably
Terry to Brooklyn.
will-run
23a
UR:
Surface level:
to Brooklyn.
22
23
24
20
(22) Probably, Terry will run to Brooklyn. (23) Terry will probably run to Brooklyn. (24) Terry will run to Brooklyn, probably. Allmost all CA: s may occur in the surface sentence-initial position (which seems to be primary in English), surface preverbal position (which seems to be primary in Czech) or the surface final position. Koktovi states that the CA in the sentence-initial position may but need not occur parenthetically, while the CA in the sentence-final position in English (but not in German or French) is usually parenthetical. (As an example of a rare non-parenthetical sentence-final CA she quotes McCawley’s A WAR will break out probably, where only a war is in the scope of probably, while wihbreakout belongs to the topic.)
However, one is obliged to question the importance of the distinction parentheticals vs. non-parentheticals, especially in the surface sentence-initial position. It is true that certain CA:s demand a pause accompanied with a drop in pitch and a comma in writing - e.g. in rn_r opinion, according to, etc. Other CA: s cannot be separated from the rest of the surface sentence by any pause, e.g. only (in sentences with one CA only), no (in sentences that are not answers to questions). But many CA: s can be either ‘parenthetical’ or ‘non-parenthetical’ and, as far as I can see, there is no difference in meaning. Both alternatives stem from the same types of UR: s as those exemplified by Koktova’s UR: s (20a), (22a). (23a): (25) Happily, Terry will run to Brooklyn (26) Happily Terry will run to Brooklyn. 3.1 In the second half of the book, the author describes the existing ways of forming CA: s and analyzes the lexical properties of different subclasses of CA: s. such as their paraphrasability. omissibility. pronominalization possibilities. distribution in the mentioned surface-sentential positions, and the primary scoping relations between different kinds of CA: s (which are important for instances of the multiple occurrences of CA: s ~ cf. 2. I). Finally, there is a list of English CA: s classified into eleven subgroups and their equivalents in Czech. (Since the whole book is concerned with the English CA:s, their Czech translations seem rather unmotivated. I dare say that Czech linguists, not only the members of Professor Sgall’s team but also others e.g. J. Firbas or F. DaneS have done more for the understanding of TFA than anybody else and that Czech is a language with a more transparent relationship between the surface word order and the TFA of the UR: s than many other languages with the socalled free word order. I guess that the book reviewed is a fragment of a larger material from which a more profound comparison of CA: s in Czech vs. English had to be deleted for the sake of economy of space.) I abstain from an account of this part of the book, which contains many interesting highlights, but which is not on a par with the thoroughly original first part. 3.2 It is a pity that Koktovi has not been able to iron out various wrinkles caused by difficult personal circumstances under which the book has been written. What I have in mind is the occurrence of surprisingly in examples (2). (2a), (p. 2) (51))(60) (pp. 24f.), (I l6)-( I 19) (pp. 43 f.). where prohah/y should stand, or examples (3) and (3a), where surprisingly should be deleted, all according to the personal communication of the author. (Because of these emendations I have not kept the original numberings of quoted example sentences and UR: s.)
At the same time it is quite obvious
from such emendations
that in writing
a book
about a language other than her mother-tongue the author was handicapped by the necessity of relying heavily on informants and authors quoted in regard to the interpretations of more exclusive (and rather marginal) instances of several CA:s in one sentence and especially for the less usual distributional surface word order possibilities, which a non-native speaker can hardly master in toto. An improvement could have been obtained by excluding the discussion of examples as (6) (1 I), (13) and (14) referred to above. Their sole purpose is to give substance to the less important theoretical claims on the very marginal occurrences of multiple CA: s. Generally speaking, the book should be read in minor portions, not in one sitting after reading the fiftieth or one hundreth variation on Terry will run probably to Brookl_vn the reader has hardly any own judgement left otherwise. There is an obvious dilemma: Koktova tries to present as exhaustive a description as possible and thus has to trust the authors quoted, but the rather usual practice of English-speaking generativists to describe the (real or only imagined) pecularities of one’s personal idiolect does not offer too firm a ground. She avoids the morass of claimed nuances between question-marked, double question-marked, asterisk-marked, etc. sentences, but her inspiring work might have gained in credibility had she observed the old rule of thumb proposed by Chomsky in Syntactic Structures (and forgotten by Chomsky himself decades ago), viz. to construct the rules of grammatical descriptions on the basis of sure, generally accepted (or rejected) cases and to determine the correctness of all cases in the shadow zone between with the help of the rules themselves. (Cf. the preceding paragraph.) 3.3 Finally, I cannot resist regretting the use of the terminology employed by the author. I understand well why the followers of Professor Sgall have abandoned the original terms of theme and rheme for topic and focus (and Functional Sentence Perspective for Topic-Focus Articulation). They try to propagate their ideas in the ‘disguise’ of the widespread ‘Western’ terminology. However, the terms originally used had the advantage of being different and used in another sense by very few other linguists (e.g. by Halliday). I want to assert that the use of the same terminology as in many qualitatively inferior and cheerfully intuitivistic and superficial descriptions (remember, e.g. Chomsky’s topic defined as the leftmost NP that is immediately dominated by S in the surface structure and that is a major category), does not make the issue clearer. However, all the minor discords I felt when reading Koktova’s book cannot obscure the fact that here is an important work surpassing the fragmentary descriptions of CA: s by other authors and corroborating the frugality of the Functional Generative Description, which I, without being an adherent of myself, wish would achieve a better place in the linguistic sun than many fashionable theoretical schools.
References Sgall, P. and E. HajiEovB, 1977. Focus on focus 1. Prague S-34. Sgall, P., 1978. Focus on focus II. Prague Sgall, P., E. HajiEovi pragmatic
aspects.
and J. PanevovB, Prague:
Bulletin of Matematical
Bulletin of Mathematical 1986. The meanings
Linguistics
of the sentence
Linguistics 29
28.
, 23-41.
in its semantic
and
Academia.
Jenny Cook-Gumperz (ed.), The social construction of literacy, New York, Cambridge University Press, 1986. x + 248 pp. Reviewed by: William Grabe, English Department, Northern Arizona University, P.O. Box 6032, Flagstaff, AZ 8601 I, USA. This volume, the third in the Cambridge series Studies in Interactional Sociolinguistics, brings together a set of articles which views literacy development as an outcome of classroom interaction patterns. Arguing against definitions of literacy in terms of standard assessment instruments, the volume calls for more careful analyses of discourse strategies in classrooms as the means for understanding individual and group variation in literacy development. The use of interactional language in school contexts is seen as a major factor in determining outcomes of learning processes and the development of literacy skills. In particular, the different ways that children speak and are spoken to in school will influence how they are perceived and judged, and will, in turn, affect their literacy skills development. The volume comprises ten chapters that fall into two groups; the first three, by Cook-Gumperz, Cook-Gumperz, and John Gumperz, provide background resources for the chapters to follow. The latter seven, by Gordon Wells ‘Five-year-old children at home and school’; Sarah Michaels ‘Narrative presentations by first graders’; James Collins ‘Differential instruction in reading groups’; Donna Eder ‘Constraints on reading group mobility’; Douglas Campbell ‘Mathematical literacy in a bilingual classroom’; Herbert Simons and Sandra Murphy Spoken language strategies and reading acquisition’, and Collin and Michaels ‘Discourse strategies and the acquisition of literacy’, are all specific ethnographic data-based studies. Before reviewing individual chapters in detail, a few general comments are in order. While the title of the book, The Social Construction qf Literacy, is intriguing, the links between the study of classroom interaction and the actual development of literacy skills are not entirely straightforward; the volume as a whole does not go very far towards clarifying these links. Few researchers would suggest that the social context of classroom interaction is irrelevant to the development of literacy skills. At the same time, however, ignoring the psychological processes central to literacy development. as if they were inconsequential matters, leads to a limited view of literacy development, albeit from a different perspective. In fact, there is very little information in this volume specifically on reading and writing development or individual ability differen-