Language Evolution: The Windows Approach addresses the question: "How can we unravel the evolution of language, given that there is no direct evidence about it?"
The LINGUIST List is dedicated to providing information on language and language analysis, and to providing the discipline of linguistics with the infrastructure necessary to function in the digital world. LINGUIST is a free resource, run by linguistics students and faculty, and supported primarily by your donations. Please support LINGUIST List during the 2016 Fund Drive.
Review of Morphosyntactic Expression in Functional Grammar
Date: Sun, 27 Nov 2005 18:29:34 -0700 From: Inge Genee Subject: Morphosyntactic Expression in Functional Grammar
EDITORS: de Groot, Casper; Hengeveld, Kees TITLE: Morphosyntactic Expression in Functional Grammar SERIES: Functional Grammar Series 27 PUBLISHER: Mouton de Gruyter YEAR: 2005
Inge Genee, Department of Modern Languages, University of Lethbridge
This is a collection of papers addressing various aspects of the expression rule component in Functional Grammar. The specific brand of functionalism intended here is the framework formerly called Functional Grammar (FG; Dik 1978, 1980, 1997). However, while the title of the volume still uses the older name of the theory, the contents of a number of the papers already places them firmly within the latest version of the theory, which, to reflect its much more rigorously systematic attempt at incorporating relevant aspects of discourse structure into the grammar, now goes by the name Functional Discourse Grammar (FDG; Mackenzie and Gómez-González 2004; Hengeveld and Mackenzie 2005, fc.).
As mentioned by the editors in the preface, ''morphological and syntactic issues have received relatively little attention in Functional Grammar (...) partly due to the fact that this grammatical model, given its functional orientation, assigns pride of place to pragmatics and semantics'' (p. v). In other words, Functional Grammarians tended to focus on providing a satisfactory underlying representation for a specific utterance, without necessarily always worrying too much about the expression rules that would model the derivation of the actual surface form of the utterance based on this underlying representation. The present volume is intended to begin to rectify this situation with respect to morphosyntax.
I discuss the papers in groups on the basis of similarities in their approaches and/or focus, not necessarily in the order in which they are presented in the book or in the groupings suggested by the editors in the introduction.
A first group of four papers address the mechanics of the expression rule component in a way that brings the model closer to a real production model, without necessarily intending it to be a full model of the speaker.
Dik Bakker's chapter, ''Agreement: More arguments for the dynamic expression model'' (p. 1-40), builds on a number of his earlier publications (1999, 2001) that propose a new way of handling the sandwiching of form and ordering rules in the derivation of utterances in which form and order codetermine one another. Dynamic expression generates so-called expression trees. These trees adhere to three tree-construction principles: 'top-down', 'left-to-right' and 'depth-first', and two information-passing principles: 'inheritance' and 'percolation'. Bakker shows that a number of fairly straightforward form-order interactions can be accounted for by means of these five principles. However, some complicated subject agreement facts in Arabic call for a fourth tree-construction principle which he calls 'limited look-ahead'. The constructions in question exhibit various combinations of person, gender and number agreement that crucially depend on word order: the further a postverbal subject is away from the verb, the less gender and number distinctions are encoded in agreement morphology. Bakker accounts for these facts by assuming that the semantic and pragmatic information required to correctly express these categories becomes accessible at different points in the generation of the expression tree. He distinguishes three degrees of accessibility: global, relative and local. This accounts for the fact that in the Arabic dialect under consideration person is always encoded (globally accessible), gender less frequently (relatively accessible), and number even less frequently (locally accessible).
The paper by Anna Siewierska and Dik Bakker, ''The agreement cross- reference continuum: Person marking in FG'' (p. 203-247), takes a more typological approach to the treatment of agreement and cross- reference morphology. Based on Bakker's dynamic expression model and a slightly idiosyncratic interpretation of the basics of Hengeveld's FDG, they propose to handle all crosslinguistically possible types of agreement and cross-reference as resulting from language-specific differences in the interaction between pragmatics and morphosyntax, rather than as signs of semantic differences between languages. Since reference is treated as a pragmatic, interpersonal phenomenon in FDG, referentiality is not relevant to the semantic analysis of a given utterance. This allows us to posit identical semantic representations for what look like wildly different ways of expressing similar States of Affairs in different languages, by capturing the crucial intuition that whether a specific entity is expressed not at all or multiple times through various combinations of full NPs, pronouns, or cross- reference and agreement markers, with some of them possibly even occurring in more than one copy (as in e.g. clitic doubling), it is only present once in the semantics. In a final section, Siewierska and Bakker show how this approach can account for a series of unexpected speech errors in bilingual situations.
John Connolly's paper, ''Constituent ordering in the expression component of Functional Grammar'', suggests some modifications to Bakker's (1999, 2001) dynamic expression model in terms of how it handles linearization templates. After stating that the dynamic model is to be preferred over the standard model (Dik 1978, 1980, 1997), Connolly proposes the following refinements: (i) ''full templates should be employed within the dynamic model'' (47) rather than deleting unfilled entries - this eliminates the need for a ''pruning mechanism''; (ii) trace phenomena can be handled by inserting an empty string into the node's configuration slot - this allows for appropriate expansions of the node where necessary to account for their syntactic effects, an option not available to ''real'' empty nodes; (iii) functional categories such as Subject or Modifier are assigned to an element as it is inserted into the appropriate slot in the linearization template - this prevents unnecessary copying of information while preserving useful functional and syntactic information; (iv) where more than one constituent can be placed in one slot, an ''elastic template'' (49) can be used to specify the number of constituents in that slot.
Kees Hengeveld's paper, ''Dynamic expression in Functional Discourse Grammar'', shows how Bakker's (1999, 2001) dynamic expression model can be incorporated into the grammar component of FDG. Within the grammatical module of FDG, an utterance is analyzed at four hierarchically ordered levels: the Interpersonal level takes care of the pragmatic aspects of the utterance, the Representational level of the semantic aspects, and the Structural and Phonological levels of respectively the morphosyntactic and phonological aspects. The Interpersonal and Representational levels are generated by a process called Formulation; the Structural and Phonological levels by a process called Encoding. The various Formulators and Encoders select appropriate ''building blocks'', called primitives, from the relevant section of the Fund, which is split up into three parts and contains frames, templates, prosodic patterns, lexical and grammatical morphemes, and various kinds of operators. Hengeveld shows how Bakker's principles 'Depth First' and 'Maximal Depth' can be incorporated into the FDG grammar module by specifying different pathways through the grammar. Most importantly, it follows from this analysis that not all levels of representation are always relevant to the generation of an utterance. For instance, when a specific illocution is expressed through prosodic means (e.g. the rising intonation in a yes/no question), the Representational and Structural levels are not relevant and are bypassed in that part of the generation process. Hengeveld exemplifies the dynamic nature of the generation process by providing detailed derivations for three one-word utterances from typologically different languages.
A second group of four papers addresses theoretical and practical issues around specific aspects of FDG, generally focusing on the detailed analysis of problematic or otherwise interesting data.
Niels Smit's paper, ''Noun incorporation in Functional Discourse Grammar'' (p. 87-134), shows how FDG is capable of handling noun incorporation (NI). Based on a reinterpretation of Mithun's (1984) ''functional-lexicalist'' (97) approach, Smit offers a functional- syntactic account of NI. His analysis capitalizes on the fact that, within FDG, nouns can be introduced in three different ways. Minimally, a noun represents a zero-order entity and is associated with a predicate frame f. Such nouns are non-modifiable and non-referential: they do not correspond to an entity and no referential subact is associated with them at the Interpersonal level. Alternatively, a noun represents a first-order entity and is associated with a term x. Such nouns are modifiable, but not necessarily referential: a referential subact may or may not be associated with them at the Interpersonal level. Based on this distinction between three types of noun-headed constituents, (f), (x) and (R), Smit now shows how they map onto three types of NI, associated with different combinations of referentiality and modifiability of the incorporated noun (IncN): fNI (IncN is non-referential and non- modifiable), xNI (IncN is non-referential but modifiable) and RNI (IncN is referential and modifiable). Finally, Smit proposes a new implicational hierarchy for NI: RNI >xNI > fNI. This needs to be read slightly differently from the normal way in which such hierarchies are usually interpreted. Rather than arguing that languages would develop NI from right to left on this scale, Smit sees xNI as a kind of pivot point, which all NI languages have. Languages with rigid part-of- speech categories will additionally include constructions to the left to include RN, while languages with flexible categories will extend to the right to include fNI.
Casper De Groot's paper, ''Morphosyntactic templates'' (p. 135-161), shows how redundancy in the generation process can be avoided by allowing multiple different semantic and pragmatic configurations to map onto one morphosyntactic template. Templates are selected by the morphosyntactic encoder from the second level of the Fund, on the basis of relevant distinctions specified at the Interpersonal (pragmatic) and the Representational (semantic) levels. The output of this encoding process is a representation at the Structural (morphosyntactic) level. De Groot shows that, in Dutch, eleven different inflectional and derivational word-formation processes resulting in various kinds of modifiers and participles map onto two word-level templates. For Hungarian he shows that four semantically different types of modifiers map onto one term-level template. Finally, he shows that in the Western Brazilian language Oro Nao a number of different question word questions, which share the same pragmatic representation at the Interpersonal level but different semantic representations at the Representational level, are expressed using one clause-level word order template. These three case studies underscore the usefulness of templates in the Fund to account for word, phrase and clause structure.
Francis Cornish's paper, ''A Crosslinguistic study of 'locative inversion': Evidence for the Functional Discourse Grammar model'' (p. 163-202), provides an FDG analysis of sentences with initial locative/temporal constituents, such as French 'Dans l'armoire se trouvaient les chaussures' or English 'On the wall hung an antique chimney hook'. The analysis is based on data from French, English, Italian, Portuguese, Turkish and Arabic, chosen because they have different basic word order patterns. All these languages have similar 'locative inversion' constructions, which share the following characteristics: (i) a locative/temporal constituent is in initial position and fulfils a thematic function; (ii) the position of the subject varies, in accordance with the word order rules of the language in question, but it is always in a special rhematic/focus position and signals new information; (iii) the verb does not predicate. Drawing on Hannay's (1991) idea of ''message management modes'', Cornish analyses the constructions under discussion within the FDG framework as instances of a special kind of transitional discourse Move, characterized by a specific configuration of: (i) a referential subact with the pragmatic function SubTopic/Stager, (ii) a second referential subact with the pragmatic function Presentational Focus, and (iii) the absence of any ascriptive subact. Finally, Cornish offers an explanation for the lack of agreement morphology on the postposed subjects in some of the languages under discussion.
Evelien Keizer's paper ''Close appositions'' (p. 381-417) revisits the analysis of constructions like (1a) 'the actor Orson Welles', (1b) 'the word recession' and (2) 'Orson Welles the actor' from an FDG point of view and based on natural data from the British component of the International Corpus of English. Her paper may serve as a stern reminder of the dangers of using made-up examples. Some previous analyses of close appositions have suffered from being based on such non-natural ''data'' - Keizer's data shows that various constructions said by earlier scholars not to be grammatical are perfectly fine given the right context, significantly undermining the value of their analysis (especially Acuña-Fariña 1996, but also Burton- Roberts 1975). Reevaluating issues like definiteness, referentiality, and headedness, Keizer proposes underlying representations for (1a) and (1b) that reflect the fact that the determiner has scope over both nouns, the two elements are non-referential, and the second element functions as a modifier. Her representation for (2) reflects the fact that the proper name is intrinsically definite but non-referential, while the second element is a modifier with its own specification for definiteness. Both types represent, at the Interpersonal level, a referential subact consisting of two ascriptive (non-referential) subacts. She then classifies the uses of these constructions, based on the communicative function of the two contributing elements: (1) functionally identifying, as in 'the number four' or 'the name Algernon'; (2) descriptionally identifying, as in 'the Jaguar boss Tom Walkenshaw' or 'Humphrey the Cabinet cat'; (3) introductory, as in 'Roald Dahl the author' or 'this bloke Mark'; and (4) contrastive, as in 'the CRITIC Paul Jones or the AUTHOR Paul Jones?'. Various combinations of pragmatic function assignment and identifiability allow for the formulation of matching underlying representations at the Interpersonal and Representational levels that account for the discourse functionality of the four types.
The remaining papers revisit problems left unsolved by the canonical FG model or begin the analysis of phenomena not previously addressed within this framework. These analyses are generally still formulated in terms of the older framework, but can be easily ''translated'' into the newer versions of the model.
Two papers deal with non-verbal predication, both taking Hengeveld's (1992) influential typological framework for the treatment of non-verbal predication as their starting point.
Eva van Lier's paper, ''The explanatory power of typological hierarchies: Developmental perspectives on non-verbal predication'' (p. 249-280), tests the explanatory power of synchronic hierarchies for several types of developmental data. She investigates two predicability hierarchies and two expression hierarchies: Predicability hierarchies: (i) [equative > ascriptive] (ii) (within ascriptive:) [locative > adjectival > nominal > possessive] Expression hierarchies: (iii) zero-1 hierarchy: no copula, non-verbal predicate is marked as intransitive verb: [bare > referential > relational] (iv) zero-2 hierarchy: no copula, no morphology: simple juxtaposition: [equative > ascriptive]
In addition, two diachronic pathways for copula development are taken into account: (v) diachronic pathway 1: [verb > locative > adjective > noun > possessive] (vi) diachronic pathway 2: [pronoun > identifying > classifying > adjective/noun].
Van Lier then tests the applicability of these hierarchies and pathways to five types of developmental data: (synchronic variation as) change in progress, internally motivated and contact-induced language change, and first and second language acquisition, hypothesizing that these data should be explainable in terms of the hierarchies and pathways under investigation. Data come from African American Vernacular English copula deletion patterns, copula development in Ibero-Romance, Chinese, Sranan and Swahili, non-verbal predication in embedded clauses in Hungarian spoken outside of Hungary, the development of copula support in young children learning English, and the acquisition of the Spanish double-copula system by L2 learners with English L1. All of these follow the hierarchies under discussion, suggesting that the synchronic hierarchies are applicable also to developmental and diachronic data. In particular, the zero-1 hierarchy in (iii) and the first predicability hierarchy in (i) are relevant to contact- induced change, and the second predicability hierarchy in (ii) is relevant to both language-internal synchronic variation and to first and second language acquisition processes.
''Non-verbal predicability and copula support rule in Spanish Sign Language'' (p. 281-315), by Ángel Herrero-Blanco and Ventura Salazar-García, applies Hengeveld's (1992) framework to a non-oral language. To my knowledge this is the first systematic attempt at applying F(D)G to (aspects of) a sign language, which makes it all the more interesting: a theory which claims typological adequacy should be applicable to all languages, including non-oral ones. The paper presents a wealth of data concerning non-verbal predication in Spanish Sign Language (LSE), with the aim of determining which types of non-verbal predicates are predicable (in Hengeveld's 1992 sense) in this language. With respect to LSE, Herrero-Blanco and Salazar-García posit a number of hypotheses (286), which boil down to the claim that LSE will conform to the predicability and expression hierarchies mentioned in the discussion of Van Lier's paper above. It turns out that it does, confirming that Hengeveld's typological generalizations also hold for signed languages, which strengthens the theory's claims for typological adequacy.
A couple of facts about ESL are especially interesting. First, ascriptive and equative constructions such as 'my friend is a/the famous physician' are expressed by simple juxtaposition: MY FRIEND PHYSICIAN FAMOUS; but specification constructions, such as 'the capital of France is Paris', require a question-answer construction format that seems to stem from the fact that the specifying element, which is the argument in the construction, carries the pragmatic function Completive Focus, as in the answer to a question: FRANCE C A P I T A L y/n, PARIS (spacing marks the prosodic marker for a yes/no question, which takes the form of raised eyebrows). When present, aspect markers appear on non-verbal predicates just like on verbal predicates, confirming a basic 'zero-1' strategy for nominal and adjectival predicates. Second, presentative predications such as 'there is a dog in my house' use a form of the verbal predicate HAVE: MY HOUSE DOG HAVE. This is interesting given the existence of parallel constructions in Spanish. The authors suggest that this HAVE was actually borrowed from Spanish, given the fact that its sign takes the form of the Spanish verb 'hay' in fingerspelling. Third, the indirect deictic predicate THERE(i), which is normally used to reference 'a space outside the communication setting'(299), functions as a copula in non-presentative locative predications such as 'my friend is in Alicante': MY FRIEND ALICANTE THERE(i); that it functions as a copula is confirmed by its taking on aspect markers, such as the perfective/egressive marker END in 'my mother is not in Alicante anymore': MY MOTHER ALICANTE THERE(i) END.
This is the only paper in the collection that, in my opinion, could have benefited from more careful editing. It is noticeably clear throughout that English is not the authors' native language, and in a number of cases the points they want to make would have benefited from more explicit explanations and examples.
Annerieke Boland's paper, ''A new view on the semantics and pragmatics of operators of aspect, tense and quantification'' (p. 317- 350), reassesses the treatment of operators that have traditionally in FG been treated as belonging to level 1 (predicate operators) or 2 (predication operators). Level 1 operators ''contribute to the description of the property or relation'' and level 2 operators ''contribute to referring to the real or imaginary world'' (346). Basing herself on Bohnemeyer's (1998) distinction between pre- and post states on the one hand and pre- and post times on the other, Boland shows phasal, (im)perfective, and perspectival aspect to belong to the descriptive level: ''they select a part of the temporal structure of the property or relation and only this part is ascribed to the argument(s)'' (329). Standard FG (Dik 1997:239) treats perspectival aspect (prospective and perfect) as a level 2 operator, mainly on the basis of scope facts, but Boland shows that scope can also operate within levels and is thus not an automatic reason for relegating something to a higher or lower level.
Based on Klein's (1994) notion of topic time, she proposes that tense locates the State of Affairs in time, ''but only the part that is relevant to topic time'' (331), and that it is a level 2 (referring) operator. This allows for the analysis of apparent contradictions, such as the use of Simple Present for scheduled future events, as in 'The train leaves at five tomorrow', and, in combination with her previous analysis of aspect, for a very elegant analysis of complicated conflations of tense and aspect as in 'He will have written a letter' and the habitual reading of 'He runs the marathon'.
Following Klein (1994) and Anstey (2002), finally, she shows that there are two different types of quantification: property quantification (level 1) and event quantification (level 2). Habituality expressed by 'used to' and frequentativity expressed by 'keep -ing' (with telic States of Affairs) in English are shown to be types of event quantification. 'Keep -ing' (with non-telic States of Affairs), however, is seen as property quantification. In her conclusion, Boland states that her new treatment of the semantics of these operators ''would fit nicely into the model of FDG''. Both the level 1 and level 2 operators belong to the representational level (where the semantics are accounted for), but only the highest ones, at level 2, ''require a link with the interpersonal level'' (where pragmatics are accounted for), since ''mutual knowledge and pragmatic inferences based on context, situation and world knowledge (...) play an important role in the interpretation of operators of tense and event quantification'' (347).
Ahmed Moutaouakil's paper ''Exclamation: Sentence type, illocution or modality?'' (p. 351-379) argues, to me convincingly, that there are serious problems in treating Exclamation as a sentence type, as well as in treating it as a special kind of illocution, as is done in standard FG (Dik 1997). He proposes instead to treat it as a special type of subjective modality. Subjective modality concerns specifications of the source of S's evaluation of the proposition, including personal opinion regarding its truth and desirability. Moutaouakil now proposes to add a third type, which he calls ''Impression/emotional reaction''. He then shows, with data from English, French and various Arabic dialects, that Exclamative modality can be positive (appreciative) or negative (depreciative) and that there are various degrees of Exclamation, which have different expression forms.
In the final part of the paper he explores the consequences of his proposal for the layered structure of the clause. After first arguing that Exclamative modality can have a proposition, a term, or a stretch of discourse in its scope, he proposes a Discourse structure, a Clause structure and a Term structure that would take the scope facts into account, and ends with a proposal for a ''transmodular'' approach to subjective modality in general. However, his proposal for a modular grammar model has, I think, been superseded by more recent work in FDG, which, surprisingly given the other contributions in this collection, is completely ignored.
Two fascinating papers address aspects of direction/inversion morphology in two Native American languages, a topic that, as far as I am aware, has not been dealt with in this theoretical framework before, but turns out to lend itself very well to a classical FG analysis. The first deals with Plains Cree (Algonquian, Canada) and employs the term inversion; the second deals with Mapudungun (Araucanian, Chile/Argentina), and uses the term direction for the same phenomenon.
Arok Wolvengrey's paper, ''Inversion and the absence of grammatical relations in Plains Cree'' (p. 419-445), argues convincingly that the Algonquian languages lack the syntactic functions Subject and Object as defined in standard FG. One obvious consequence of this is that they do not have case agreement. In fact, not only do they not mark case, they do not mark semantic function either. Person marking affixes merely signal the involvement of a first, second or third person entity. In the case of transitive verbs, semantic role disambiguation is achieved indirectly, by so-called direct/inverse marking: a separate morpheme on the verb that indicates whether a specific person hierarchy is either in alignment (direct) or out of alignment (inverse) with a specific semantic function hierarchy (as in Aissen 1997). A second consequence of the analysis of Cree as lacking syntactic functions is the lack of an active-passive distinction. FG analyses active/passive oppositions as resulting from alternative syntactic function assignment - given the absence of syntactic functions, active/passive distinctions are therefore not possible. Constructions analyzed by others as involving active/passive on the basis of quantifier scope facts, as well as unspecified-actor constructions, are demonstrated to crucially depend on pragmatic information status rather than subject- or objecthood. Wolvengrey concludes that ''The interaction of pragmatic and semantic functions is enough to disambiguate all necessary interactions without recourse to a third level of grammatical functions'' (441).
Ole Nedergaard Thomsen's paper, ''Direction diathesis and obviation in Functional Grammar: The case of the inverse in Mapudungun, an indigenous language of south central Chile'' (p. 447-482), focuses on the role of direction/inversion morphology in the marking of obviation distinctions. His analysis is probably broadly applicable to all languages that combine direction/inversion with obviation, including Algonquian, and deserves to be read by anyone concerned with the question of what exactly it is that is coded by obviation. Obviation is a morphological distinction on verbs (and sometimes, but not in Mapundungun, on nouns) that subdivides the third person category in two: the more topical/central entity is marked proximate, and the less topical/central entity is marked obviative. In terms of the hierarchy alignment involved in direction/inversion, proximate ranks higher than obviative, so that in a transitive verb, when agent is proximate and patient is obviative, the direct morpheme applies, while the inverse morpheme applies when agent is obviative and patient is proximate. The paper addresses the question as to what kind of topicality it is that is encoded by obviation: a kind of vantage point or speaker's focus of (visual) attention, or rather a kind of textual aboutness or discourse topicality.
In order to test this, an experiment similar to Tomlin's (1995) Fish Film experiment is conducted, ''to see whether voice and subject selection in utterances correlate with visual attention focusing'', the hypothesis being ''that Proximate would be chosen to code a visually primed participant in Tomlin's experiment, and that a primed Agent (Proximate) would select direct voice whereas a primed Patient would select inverse voice'' (450). This turns out not to be the case: visual priming has no effect on Proximate choice or inverse voice selection. Regardless of visual priming, the agent was always encoded as Proximate, the Patient as Obviative, resulting in Direct voice selection. Nedergaard Thomsen concludes that ''focal attention does not determine Proximate status (empathy)'' but is rather determined by ''textual topicality'' (472). The Proximate/Obviative distinction therefore appears to encode pragmatic rather than syntactic (perspectivizing) functions. Nedergaard Thomsen finally argues that the new top-down FDG model, which situates pragmatic functions at the highest, i.e. Interpersonal level, and consequently early on in the derivation of the utterance, would be better suited to handle these facts than the older, bottom-up FG model.
Johan Lotterman and Lachlan J. Mackenzie, in their paper ''Unexpected insertion or omission of an absolutive marker as an icon of a surprising turn of events in discourse'' (p. 483-501), describe the grammar of the Absolutive marker 'a' in Tanggu, a Sepik- Ramu language of Papua New Guinea. This marker appears regularly on the second argument of transitive predicates and on the sole argument of intransitive predicates, including non-verbal predicates and postpositions. At this level this marker can be interpreted as the expression of the basic Ergative/Absolutive semantics of Tanggu, and can thus be accounted for at the Representational level in FDG. However, there are exceptions that require the involvement of the Morphosyntactic and Interpersonal levels as well as the Fund. Morphosyntax is involved in accounting for the normal absence of the Absolutive marker in one-place predications, which Lotterman and Mackenzie interpret as resulting from the non-activation of the operator that would be responsible for its presence. The Fund is involved in accounting for the lexically determined absence of the Absolutive marker with selected verbal and non-verbal predicates, which does not appear to reflect any kind of semantic distinction in those predicates and therefore needs to be separately represented in their lexical entries. The most interesting exceptional facts involve the unexpected presence or absence of the marker in contexts where the opposite would normally be expected. The authors show that this surprising presence/absence iconically reflects a surprising turn of events in a narrative. In FDG terms they analyze these utterances as corresponding, at the Interpersonal level, to Discourse Acts ''that convey the speaker's desire to engender in the hearer an effect of surprise'', this effect being ''marked by an inversion of the normal marking of the Absolutive'' (498).
The final paper of the collection is ''Pronominal expression rule ordering in Danish and the question of a discourse grammar'' (p. 503- 524), by Lisbeth Falster Jakobsen. Falster Jakobsen discusses the complicated interaction between ordering and case morphology in the Danish pronoun system, which is complicated by the kind of high degree of synchronic variation related to register and age indicative of an ongoing change in progress. While the details of the Danish facts are a bit different from those of English, the situation reminds us of the general confusion on the part of native speakers of English with regard to their pronoun system, which reflects the ongoing generalization of the original accusative forms. In fact, in a number of cases a more idiomatic translation into English would have better reflected what is going on in the Danish examples, e.g. in example (20) (p. 513), where Danish 'mor og mig(acc)' is translated as 'Mum and I', while the perfectly grammatical colloquial 'Mum and me' would have underlined the parallel development. Similarly closer translations could have been given for examples (21), (22), (23), (25). The central problem is how to model such variation. It is not enough to simply distinguish between written and spoken language and formulate different expression rules for each modality. Nevertheless, a top-down FDG-like dynamic expression model will be better able to handle this type of variation: distinctions of mode and style will need to be specified at the Interpersonal level, which interface with the expression level to produce appropriate output for all possible combinations of mode (written vs. spoken), formality, and Speaker attributes such as age and social class.
This book will be the last in the Functional Grammar Series, published by Foris since 1985 and by Mouton the Gruyter since 1992. The series editors have decided that, given the now more generally available publication options for work in this framework, a dedicated series is no longer warranted. Taken as a sign of the mainstreaming of F(D)G, this may be seen as a positive development.
If the editors of the Functional Grammar Series intended it to go out with a bang, they have certainly succeeded. This is a very valuable collection of papers that contributes both to the development of the theory into one that truly takes discourse into account, as well as to the understanding of problematic phenomena in a variety of languages. Linguists already working within F(D)G or in other functionally oriented frameworks will find much to inspire them in this book.
That said, a few critical comments are also in order. First, I was disappointed that some of the papers display a lack of detailed derivations. In good old FG style, they are still long on underlying representations and short on derivations. Some do not even offer fully specified underlying representations. Perhaps this reflects the fact that the new FDG format is still in quite a bit of flux and many details still need to be worked out. Nevertheless, in a volume on morphosyntax one would expect the details of the proposed derivations and all the steps that lead from the underlying representation to the resulting expression, or at least to its phonological form, to be worked out in their entirety, especially since it is the explicitly stated goal of the collection to contribute to the development of the expression rule component of the grammar. Notable exceptions are the papers by Bakker, Bakker and Siewierska, Hengeveld, Smit, de Groot, Cornish.
Secondly, aspects of some of the FG-based papers are already partially superseded by newer developments in the FDG framework. For instance, in Herrero-Blanco and Salazar-García's contribution on non-verbal predication, pragmatic functions are still represented at the level of the predication, as add-on functions assigned to terms. Boland's otherwise excellent analysis of operators of tense and aspect, to the extent that it refers to things like discourse relevance (''topic time'') and the ''referring'' functions of tense, just seems to be asking for an updated FDG analysis. She mentions herself, at the very end of her paper, that her account ''would fit nicely into the model of FDG'' (347), but this leaves me unsatisfied: I can see the outlines of how it would fit, but I would like to see the details worked out. Moutaouakil presents his analysis of exclamation initially in the ''upward layering'' framework, and then later on suggests the outlines of a modular approach which appears not to take into account any work after 1998. If he has reasons for preferring his version of the modular approach to the one developed in the FDG framework, I think they need to be made explicit. Keizer, in her otherwise exemplary treatment of close appositions, gives convincing FG representations of the constructions under consideration, and then discusses in quite a bit of detail how aspects of these representations, especially definiteness and referentiality, would be represented at the Interpersonal level. Why not give the FDG representation of the constructions at all three relevant levels, so we can see how it would work?
Similar comments would apply to the papers by Wolvengrey and, to a lesser extent, by Nedergaard Thomsen, Lotterman and Mackenzie, and Falster Jakobsen. Nedergaard Thomsen argues, in some detail, that the FDG model would be better able to handle obviation than the FG model, because of its radical top-down structure - unfortunately, again, no actual representations or derivations are provided. Lotterman and Mackenzie explicitly take the FDG model as their framework, giving a few new representations of term structures to account for the parallel Absolutive marking in predications and terms. However, their main point, illustrated with copious textual materials, hinges on speaker intention, which is represented at the Interpersonal level. A very general Move structure is proposed, which leaves many questions open, most importantly the nature and representation of the intended 'surprise': is this a function to be assigned to an Act, or an Act type, or something else altogether? Finally, FALSTER JAKOBSEN suggests that aspects of mode and style and Speaker attributes need to be accounted for at the Interpersonal level, but does not attempt to show how or where this would exactly take place.
In conclusion, this collection itself provides ample supporting evidence for the idea that synchronic variation is often indicative of ongoing change-in-progress. Some of the variation reflects real differences of opinion, but most of it clearly points toward the FDG model or a version of it, if I may be permitted a teleological interpretation of the direction of the change. Given that most of the more FG-oriented contributions deal with unsolved problems that were carried over from FG into FDG, and that their conclusions and solutions are easily translated into the FDG framework for the most part, this is a very valuable contribution to its development.
Let me finish by stressing that most of the contributions are also of importance to a more general audience. Stripped of their framework- specific terminology, these papers offer important insights into the phenomena they study and are required reading for linguists involved in their investigation.
Acuña-Fariña, Juan Carlos. 1996. The puzzle of apposition: on so- called appositive structures in English. Universidade de Santiago de Compostela.
Aissen, Judith. 1997. On the syntax of obviation. Language 73:705- 750.
Anstey, Matthew. 2002. Layers and operators revisited. Working Papers in Functional Grammar 77. University of Amsterdam.
Bakker, Dik. 1999. FG expression rules: from templates to constituent structure. Working Papers in Functional Grammar 67. University of Amsterdam.
Bakker, Dik. 2001. The FG expression rules: a dynamic model. Revista Canaria de Estudios Ingleses 42:15-54.
Bohnemeyer, Jürgen. 1998. Time relations in discourse. Evidence from a comparative approach to Yucatec Maya. Ph.D diss., Katholieke Universiteit Brabant.
Burton-Roberts, Noel. 1975. Nominal apposition. Foundations of Language 13:391-419.
Dik, Simon C. 1978. Functional Grammar. Amsterdam: North-Holland.
Dik, Simon C. 1980. Studies in Functional Grammar. London: Academic Press.
Dik, Simon C. 1997. The Theory of Functional Grammar. I: The structure of the clause. II: Complex and derived constructions. Edited by Kees Hengeveld. Berlin: Mouton de Gruyter.
Hannay, Mike. 1991. Pragmatic function assignment and word order variation in a functional grammar of English. Journal of Pragmatics 16:131-155.
Hengeveld, Kees & J. Lachlan Mackenzie. 2005. Functional Discourse Grammar. In: Keith Brown (ed.) Encyclopedia of Language and Linguistics. 2nd edition. Oxford: Elsevier.
Hengeveld, Kees & J. Lachlan Mackenzie. fc. Functional Discourse Grammar. Oxford: Oxford University Press.
Klein, Wolfgang. 1994. Time in Language. London: Routledge.
Mackenzie, J. Lachlan & María A. Gómez-González (eds.) 2004. A new Architecture for Functional Grammar. Berlin: Mouton de Gruyter.
Mithun, Marianne. 1984. The evolution of noun incorporation. Language 60(4):847-894.
Tomlin, Russell S. 1995. Focal attention, voice, and word order. An experimental, cross-linguistic study. In: Downing, Pamela and Michael Noonan (eds.) Word order in discourse. Amsterdam: John Benjamins, 517-554.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Inge Genee is an Assistant professor of Linguistics in the Department of Modern Languages at the University of Lethbridge in Alberta, Canada. She works on Celtic (especially medieval Irish) and Algonquian (especially Blackfoot) languages and is interested in historical syntax, language contact and shift/loss, language description and functional theories of language.